Patent Application
20130274992
This application claims the benefit of the French patent application No. 1253383 filed on Apr. 12, 2012, the entire disclosures of which are incorporated herein by way of reference.
The present invention relates to the diagnosis of elements of complex systems, in particular of aircraft, and more particularly to a method, devices and a program for computer-aided preventive diagnostics of an aircraft system, using critical event charts.
The latest failure diagnosis systems in aircraft generally use failure models designed by the manufacturers and their equipment suppliers during the aircraft development cycle. They can be used for the purpose of setting up preventive diagnostics on board the aircraft in question or on the ground via web services for example.
These diagnostic systems can use messages originating from equipment monitoring systems comprising self-diagnosis software applications, also called built-in test equipment (BITE), reporting maintenance messages involving items of equipment suspected of failure as soon as the monitoring systems detect them.
Thus, for example, the diagnostic systems known as on-board maintenance systems (OMS), in particular used in the Airbus A380 (Airbus and A380 are trade marks) make it possible to group messages received from equipment monitoring systems and to access reports generated in flight in order to carry out statistical analysis allowing potential future failures to be identified.
The grouping of messages is here carried out by a software application of a centralized maintenance system (CMS) which collects and consolidates these maintenance messages in order to identify the more relevant maintenance messages allowing the maintenance teams on the ground to complete the necessary repairs. Such messages indicate failed equipment as well as providing information on possible failures based on statistical analyses such as mean time between failures (MTBF).
Access to reports generated in flight typically involves access to reports known as aircraft condition monitoring system (ACMS) reports which are generated systematically at certain phases of each flight or when particular events are detected, for example when a predetermined threshold is exceeded by a given parameter of the aircraft. Such reports thus represent a view of the status of a certain number of parameters and equipment of the aircraft. When collated, these ACMS reports allow the airline operating the aircraft to monitor its status and to intervene when deemed necessary.
The ability to prevent possible effects of future failures in the cockpit (called flight deck effect (FDE)) is offered by some aircraft manufacturers in a ground system called airplane health management (AHM) interfaced with the reports issued by an aircraft. To this end, the AHM calculates and adapts a time remaining for carrying out maintenance (called time to failure (TTF)) for the maintenance messages reported by a centralised maintenance computing function (CMCF) of the aircraft and based on the history of these messages.
In order to plan preventive maintenance tasks, an airline needs to have advance knowledge of a future dysfunction. But this is not sufficient on recent-generation aircraft where the systems are very interdependent, incorporate components with complex malfunction modes, and have architectures that have single failure tolerance.
A failure tolerance capability allows an aircraft to remain available even if an item of equipment has failed. A list of the minimum operational equipment (called minimum equipment list (MEL)) sets the conditions according to which an aircraft in which at least one item of equipment has failed can remain operational (dispatch reliability). By way of illustration, an airline can be permitted to operate an aircraft for 10 days with certain equipment failed. Thus, these operating conditions are encompassed by the MEL and are often accompanied by mandatory maintenance operations to inspect the equipment in working order associated with the failed equipment and/or to provide safe manual deactivation of the failed equipment.
A failure tolerance capability also allows an airline to operate an aircraft while, at the same time, preparing for the purchase and procurement of spares as well as the associated maintenance.
In this context, it is necessary not only to obtain a list of equipment failures in an aircraft in order to decide on its operation but moreover, the airline operating this aircraft wishes to know exactly the tolerance margin remaining before a dysfunction with greater impact occurs, for example a situation named NO GO in the MEL, which does not permit the airline to operate the aircraft in this condition or a situation according to which the passenger experience would not accord with the image the airline wishes to project (for example if the cabin video system no longer functions).
There is a need to provide predictive maintenance and failure tolerance information.
The invention makes it possible to resolve at least one of the problems previously set out.
A subject of the invention is thus a method for the computer-aided preparation of a diagnostic report for a complex system of an aircraft comprising a plurality of subsystems, at least one subsystem of said plurality of subsystems comprising means of monitoring and notification of at least one detected event, this method being characterized in that:
it implements a critical event chart at least partially modelling said complex system, said critical event chart comprising a plurality of nodes, each node of said plurality of nodes being linked by a logical implication relation to at least one other node of said plurality of nodes, said plurality of nodes comprising,
a plurality of nodes each representing a notification message capable of being received;
at least one node representing a critical event; and,
a plurality of nodes each representing an element of said complex system, each element represented by a node being liable to failure;
it comprises the following steps,
receiving at least one notification message of the occurrence of said at least one detected event;
creating a minimal diagnostic set relative to said at least one detected event, comprising a plurality of elements each represented by a node of said critical event chart, each element of said minimal diagnostic set being determined according to at least one logical implication relation of said critical event chart with a node representing said at least one received notification message; and
The method according to the invention makes it possible in particular to facilitate decision-making on the ground, for example by the maintenance control centre of the aircraft operating company because the diagnostic result is classified by relevance. Moreover, this method being based on physical and topological knowledge of the system, for example physical and topological knowledge of the aircraft consistent with failure mode effects analysis (FMEA) and the minimum equipment list (MEL), it allows in particular information to be obtained on the remaining failure tolerance margin based on the knowledge of the architecture of the systems. It also allows awareness of a list of the equipment still remaining operational, heading towards a future significant dysfunction. This information can be obtained in real time and transmitted to a remote system, for example of an aircraft in flight to a system on the ground.
Said critical event chart can be at least partially generated by instantiation of at least one generic subchart in order to simplify the creation and management thereof.
According to a particular embodiment, the method comprises moreover a step of obtaining data representative of a diagnostic history of said complex system, said ranking step being at least partially based on said data representative of said diagnostic history.
Thus by using the diagnostic history in order to consolidate a degree of certainty of analysis by ranking diagnostic results on which a preventive analysis can be based, it is possible to promote maintenance of the more often suspected candidate items, which avoids leaving an item in unresolved failure for too long. This is particularly useful in the case of operation of an aircraft that does not return to its main base after a series of flights and on which different maintenance teams work. In fact, in this case, the maintenance operatives are not the same staff from one airport to another, only making a spot check on the aircraft in a given airport. The results obtained using the steps described previously make it possible to benefit from previous diagnostic histories. Moreover, the method makes it possible to facilitate decision-making on the ground, for example by the maintenance control centre of the aircraft operating company because the diagnostic result will already be classified as a function of the history, avoiding the need for the staff of this centre to carry out the work manually from flight to flight.
Still according to a particular embodiment, said step of ranking at least some of the elements of said minimal diagnostic set comprises a step of ranking at least some of said minimal diagnostics.
The method comprises moreover, preferably, a step of calculating the persistence weighting of each element of a plurality of elements of said minimal diagnostic set, said persistence weighting calculation being based on the presence of each element of said plurality of elements of said minimal diagnostic set in a minimal diagnostic set of said diagnostic history, said ranking of at least some of said minimal diagnostics being at least partially based on results of said persistence weighting calculation.
Still according to a particular embodiment, the method comprises moreover a step of prioritizing elements of said minimal diagnostic set.
Said step of ranking at least some of the elements of said minimal diagnostic set comprises, preferably, a step of ranking problems resulting from said minimal diagnostics. According to a particular embodiment, the method comprises moreover a step of calculating the persistence weighting of each problem of a plurality of problems resulting from said minimal diagnostics, said ranking of problems being at least partially based on results of said persistence weighting calculation.
Still according to a particular embodiment, the method comprises moreover a step of selecting at least one receivable notification message represented by a node of said critical event chart, and a step of identifying the elements of said minimal diagnostic set capable of leading to the generation of said at least one selected notification message, said identified elements forming part of said diagnostic report. Attributes can be obtained and assigned to said identified elements.
The method according to the invention thus uses exhaustive physical knowledge to indicate candidate items that are not yet declared to have failed but a malfunction of which could result in a critical event. This information is very important for decision-making. In fact, such information makes it possible, for example, to prevent an aircraft departing if the tolerance margin is due only, for example, to the life of a line-replaceable unit (LRU) that is very expensive to send to the place of destination of the aircraft. In this case, the risk of a long immobilization of the aircraft while waiting for a replacement LRU is high. On the other hand, if the tolerance margin is encroached but the logistics and maintenance of the replacement parts do not pose a problem in terms of costs and operationally, it is less risky to allow the aircraft to depart.
Advantageously, the method comprises moreover a step of calculating the remaining distance before the imminent effect for at least one of said identified elements, said remaining distance being calculated as a function of the number of elements that do not form part of said minimal diagnostic set and a malfunction of which is necessary to the generation of said at least one selected notification message.
Still according to a particular embodiment, the method comprises moreover a step of selecting at least one failure resolution procedure involving at least one element of said minimal diagnostic set.
A subject of the invention is also a computer program comprising instructions stored in a non-transient memory adapted to the implementation of each of the steps of the previously-described method when said program is executed on a computer as well as an aircraft maintenance system comprising a calculator comprising means for implementing each of the steps of the previously-described method. The advantages achieved by this computer program and this system are similar to those mentioned previously.
Other advantages, aims and features of the present invention will become apparent from the detailed description which follows, by way of non-limitative example, with reference to the attached drawings in which:
In general, the invention relates to a system of preventive diagnosis and failure tolerance analysis for an aircraft system, using critical event charts (or failure condition graphs) constructed here on the basis of fault trees developed during safety studies.
As shown in
After identification of failures, several steps and/or sequences of steps are carried out independently in order to select an optimum failure resolution procedure for each item of equipment that could potentially fail, the malfunction of which could result in the current configuration of the system (phase 120), analyze the failure tolerance of the system in its current configuration (phase 125) and issue a preventive diagnosis thereof (phase 130). The results obtained during these steps allow a diagnostic report to be prepared (phase 135).
As shown by the arrow in broken lines, the latter phases are, preferably, repeated in order to allow analysis of all the detected events, for example as they are detected.
According to a particular embodiment, the modelling of the critical event chart is carried out on the basis of modelling charts of the critical events of several aircraft systems, preferably all. The critical event chart can be regarded as the extension of the fault trees developed during the safety studies. Here it has the following features:
the chart is directed, it can comprise cycles;
the chart comprises at least three types of node:
candidate items denoting equipment, preferably replaceable items, in particular calculators of the line replaceable unit (LRU) type, software applications, cables and operating conditions such as resets of an item of equipment showing a malfunction or abnormal operating conditions of a system (such as for example an engine overboost, skidding on braking or operation in the presence of ice on the air intakes). A particular attribute is advantageously denoted in order to classify each “candidate item” node according to two groups, persistent candidate items and non-persistent candidate items. The persistent candidate items are such that having once failed, their failure is irreversible without maintenance action. The non-persistent candidate items are all others;
critical event, also called failure condition, denoting system malfunction conditions modelled by the chart; and,
logic gates denoting logic operations, for example the logic operations OR, AND, negation (NOT) or a gate of the “n AMONG” type (where n is a natural non-zero integer representing an activation threshold);
each arc of the chart is a directed arc representing a logical implication relation between the two nodes that it links, it being possible to consider the origin of the arc as a cause and the destination as an effect;
the set of nodes of the chart covers the set of the failure mode, effects and criticality analysis (FMEA) fault trees produced for safety analysis (system safety analysis or FMEA system). In other words, any fault tree shown in the FMEA system is a subchart of the critical event chart;
the set of candidate item type nodes comprises the set of line replaceable units or modules (LRU and LRM) considered in the maintenance manuals known as TSM and AMM; and,
the set of functional failures defined in the MSG-3 type analysis of the system in question is included in the set of critical event type nodes of the chart.
The critical event chart can comprise several thousands of nodes and arcs.
It should be noted that a chart can have a variable level of completeness. For example, the candidate items associated with the wiring can not be shown in an intentionally simplified version of the chart of a system. However, this simplified chart enables a first level of diagnostics that is useful for online maintenance and allows a mode of implementation in which the manufacturer offers a detailed diagnostic service based on a complete chart.
Each subsystem of a system can be represented by a critical event subchart. Thus, when a critical event chart is associated with a system comprising several subsystems, each subsystem being associated with a critical event subchart, there are in the critical event chart, critical event type nodes which act as an interface between the critical event subchart, representing cause and effect relationships between the corresponding subsystems. Such nodes are, preferably, identified with a particular attribute.
Again, the circles represent nodes of the critical event chart and the arrows represent the arcs of the chart. The circles in solid lines represent critical event type nodes, the circles in broken lines represent of the nodes of the logic gates type and the circles in dash-dotted lines represent candidate item type nodes. The circle in double solid lines represents a critical event type node acting as an interface between two systems.
By way of illustration, the detection of a fault in the circuit-breaker 310 or in the main supply bus 315 is a cause of the critical event “loss of power supply on the bus” (320), according to the OR logic gate (325), in the critical event subchart 305-2. The critical event “loss of power supply on the bus” (320) being a node acting as an interface between subcharts 305-1 and 305-2, it is the cause of the critical event “loss of power supply to the actuator” (330) in the critical event subchart 305-1 according to the arc 335.
The advantages of such a representation in the form of a critical event chart are in particular associated with its consistency with models used for performing safety analyses which make it possible, using the same formalism, to represent knowledge of a system, from a high-level critical event to a critical event at the level of a component of the system and thus to combine the knowledge of equipment suppliers and of a manufacturer in a single database. It also allows the establishment of formal evidence, by using the theory of the charts' coverage, that the critical events are, from a safety point of view, satisfactorily covered by the critical event chart used in the diagnostic aid system.
After modelling a critical event chart, a subsequent phase (phase 105 of
A maintenance message, fault report, monitoring parameter of the aircraft condition monitoring function (ACMF), a message of the electronic centralised aircraft monitor (ECAM) type, a warning from the flight warning system (FWS) or pilot entries in the electronic logbook are in particular automatic notifications of the occurrence of critical events in an aircraft. These messages as well as, if applicable, similar messages are therefore associated with critical events in the critical event chart. To this end, notification type nodes are added to the critical event chart and directed links are established between these new nodes and critical event type nodes.
Such a relationship can be established using a simple first-order logic. Thus, for example, as shown in
It is noted here that a detected event, notified by a message, corresponds to a particular instantiation, over time, of a critical event or of a conjunction of critical events. Thus, for the sake of clarity, although the critical event chart here comprises notification type nodes, critical events of the critical event chart can be obtained directly from a notification message without the need to implement notification type nodes in the critical event chart.
By way of illustration, a BITE monitoring unit detecting that a hydraulic fluid pressure value is less than 345 bars and transmitting a corresponding message is a means of notifying the occurrence of the critical event of the “hydraulic pressure too low” type. A link can thus be established between this message and this critical event. Similarly, a monitoring unit detecting that the pressure of a hydraulic accumulator for a brake is less than 8 bars is another means of notifying the critical event of the “hydraulic pressure too low in the accumulator for the braking function” type.
In other words, this phase makes it possible to introduce an item of known information associated with the monitoring systems messages in the previously-modelled critical event chart.
This phase makes it possible in particular, using the same formalism, to group together maintenance messages, messages from the FWS, in particular ECAM type messages and warnings, ACMF monitoring parameters, as well as results of tests carried out on the aircraft on the ground, in connection with the corresponding critical events.
It also makes it possible to obtain a simple representation, on the basis of first-order logic, of detected events in monitoring systems in a critical event chart that is easily understood by non-expert users of the system in question. Moreover, it makes it possible to produce formal evidence of the cover and diagnostic accuracy of the software of the monitoring systems (built-in test) of these systems issuing the maintenance messages, by calculating the critical event subchart generated by the notification nodes and all their predecessors (i.e. all the candidate item type nodes having a logical implication link to the notification-type node in question). Thus, for example, the subchart referenced 410 in
Independence between the software of the monitoring systems (built-in test) provided by different equipment suppliers is ensured by using interface-type critical events nodes in the model. These nodes facilitate and formalize the specification of the interfaces between systems. Moreover, this representation allows automatic analysis of the consequences, in a single system or in others, of a modification of an item of equipment of the aircraft, in its functionalities or its failure modes. Such an analysis can be carried out using an algorithm that automatically runs through the chart step by step, and listing the critical events that may be generated by this equipment modification.
This phase also allows a manufacturer to define the coverage objectives of the malfunction or failure management procedure (also called troubleshooting) to be achieved with each maintenance message. Finally, it can be used as a reasoning model for troubleshooting on the ground because it represents all the possible branches of dysfunction that could result in a critical event notified in flight.
These phases of modelling critical event charts and assigning failure message codes (phases 100 and 105 of
A chart pattern is here a generic chart in which the nodes representing critical events and candidate items denote events and generic items which can take on as many values as there are similarities in the system in question.
By way of illustration, an aircraft generally has two ventral undercarriages that are symmetrical and similar. It would be unnecessary to analyze and model these two undercarriages because the critical event charts obtained would have the same form, only the names of the nodes would be different, the first chart referring to elements of the left-hand undercarriage and the second to elements of the right-hand undercarriage. Similarly, for assignment of the message codes, if the monitoring techniques of the left-hand undercarriage are similar to those of the right-hand undercarriage, it is unnecessary to perform the analysis twice.
Thus, steps 100 and 105 of
A first step relates to identifying all the similarities of the system to be modelled (step 500), i.e. all the groups of subassemblies of this system having similar structures. It can be carried out automatically by analysis of the system according to predetermined criteria or by an operative.
In a subsequent step (step 140), a generic critical event chart is modelled and failure message codes are assigned thereto, as described with reference to steps 100 and 105 of
The generic charts thus modelled are then analyzed (step 505) in order to identify, for each generic node, the parameter(s) which, in the name of the node in question, change from one similarity to another.
Thus, for example, assuming that a node of a generic chart has the name “Loss of the automatic door locking signal sent by the calculator of the door [x]” and that there are ten similar doors in the aircraft, named P1 to P10, the parameter in the name of this node being [x], it takes on the values of instantiations P1 to P10.
Then, for each generic chart, a table of instances of parameters is created according to the values of the parameters of the chart in the corresponding subassemblies (step 510). Such a table comprises for example the names of the generic parameters in the chart in question and, for each subset of the group in question, the value of the parameters.
A table of instances of parameters is given as an appendix by way of illustration (Table 1). Each line represents here a parameter of a given generic model. The first column contains the name of the parameter and the subsequent columns contain the possible values of this parameter for each instantiation, i.e. for each subset represented by the generic chart. By way of illustration, the table of instances of parameters comprises here the parameters #Param1#, #Param2#, #Param3# and #Generic_candidate_item#, each being capable of instantiation according to three values. This table derives from the chart pattern shown in
Turning again to
By way of illustration, the generic chart shown in
As in
The diamond 620 corresponds to a maintenance message of ECAM type the purpose of which is to avoid the occurrence of a critical event.
A fault in the generic equipment represented by the reference 600 is capable of triggering the generic critical event represented by the reference 605 which itself is capable of triggering the generic critical event represented by the reference 615, the latter being able to be triggered by another cause. The occurrence of the generic critical event represented by the reference 615 triggers the generic maintenance message represented by the reference 620.
The generic parameters involved in the generic chart shown in
The instantiation of the generic chart shown in
Because of the values defined in the table of instances of parameters, the critical event chart comprises three specific branches pertaining to each instance (referenced 600′-i and 605′-i where i represents the instantiation number varying from 1 to 3) and a common branch (referenced 610′, 615′ and 620′).
The advantages achieved by the steps described with reference to
Turning again to
This phase of obtaining messages and items of information has the further purpose of determining a minimum list of parameters used in the logical expressions used in the critical event chart, in particular ACMS parameters, to make it possible to carry out given diagnostic operations and to access the values of these parameters to allow these logical expressions to be evaluated.
A subsequent phase (phase 115) consists in particular of using the critical event chart (static and a priori knowledge), identified parameter values and notifications of the monitoring systems (dynamic knowledge collected in real time) in order to produce a diagnostic aid for the system corresponding to the critical event chart at a given moment.
To this end, the critical event chart makes it possible to establish causality links between critical events the corresponding notifications of which have been received and to isolate the critical events at the source of the propagation of the others. This chart moreover makes it possible to deduce therefrom a diagnostic aid via suspecting a minimum number of candidate items by calculating the set of the minimal vertices (or hitting sets), i.e. the sufficient set of the configurations of candidate items that could have resulted in each critical event in question.
After receiving at least one notification from a monitoring system (step 700), the corresponding notification type node(s) Ni are identified (step 705) in the critical event chart, according to the previously-established associations (phase 105 of
In a subsequent step (step 710), the identified notification type nodes Ni of are used to run through the critical event chart and select the set O of the source critical events, i.e. critical events capable of triggering the critical events directly associated with the identified notification type nodes Ni. Each of the source critical events of the set O is such that:
In order to ensure a causality relationship between the events, a condition of inclusion between the time of occurrence of the messages associated with the identified notifications is, preferably, used when creating group O. According to this condition, O is a subset {Ej}j∈J Ni such that for any element E′ included in Ni and any element Ej included in O, either E′ does not imply Ej ((E′
E)), or the time slot of occurrence of Ej is not included in the time slot for the occurrence of
In a subsequent step (step 715), the algorithm runs through the subchart of the antecedent nodes of each source critical event of the set O. The algorithm runs through the subchart as far as the candidate items and, on its way, applies the logic gates of the critical event chart in order to construct the simplified logical expression formed on the basis of candidate items and of Boolean operators AND, OR or NOT. This expression constitutes the logical explanation of the source critical event in question. To this end, the logical predicate Ab(•) is introduced (Ab meaning abnormal). It represents the logical function allowing a candidate item to be suspected. Thus, for example, Ab(Actuator) means that failure of the actuator is suspected. By way of illustration and as shown in
In a subsequent step (step 720), the source critical events are grouped together as follows: two critical events Ei and Ek are grouped together in the same set Pj if their (previously determined) associated logical explanations contain at least one common candidate item operand.
Turning again to the previous example based on
Thus, two groups Pj and Pk constitute two groups of different sources and allowing different sets of suspect candidate items to be isolated: by considering the set of candidate items suspected by Pj and that of the candidate items suspected by Pk, these sets are separated. Each group Pk reflects the presence of a problem Fk the diagnostics of which will be formulated on the basis of the candidate items which can be deduced from the group.
For a group Pk, the problem Fk is the subset of critical events such that:
Thus, for example, if group Pk is equal to {E1, E2, E3}, by using the chart shown in
In a subsequent step (step 725), the minimal vertices (minimal hitting sets) of candidate items covering each source critical event Ei of each set Pk are calculated.
A vertex of the set Pj of candidate items covering a given critical event is here defined as a conjunction of predicates on these candidate items which is consistent with the logical expression associated with the critical event Ei.
Thus, by way of illustration with reference to
A minimal vertex is here defined as follows: in a set of vertices {Vn}, a vertex Vm∈{Vn} is called minimal if there is no other vertex of {Vn} which can be logically deduced from Vm.
Thus, for example, the vertex Ab(Actuator) is deduced from the vertex Ab(Actuator) AND Ab(Supply cable). Consequently, the vertex Ab(Actuator) AND Ab(Supply cable) is not a minimal vertex of a set which contains these two vertices.
These minimal vertices represent here the minimal diagnostics for each problem Fk associated with a group Pk. In other words, the minimal vertices of a group Pk are the minimal logical expressions of candidate items capable of explaining all the critical events of group Pk. According to the example given previously with reference to
By way of illustration, the vertex V4 (Ab(CandItem1) AND Ab(CandItem7) AND Ab(CandItem4)) is not a minimal vertex of group P1 because the minimal vertex V1 (Ab(CandItem1) AND Ab(CandItem7)) is deduced therefrom.
The minimal vertices of candidate items of each group Pk can then be grouped in order to represent all the candidate items allowing the explanation of all the critical events identified through the messages notifying detected events.
The use of a critical event chart in a diagnostic aid system makes it possible to increase the level of accuracy of the diagnostics by the ability to carry out crosschecking by minimal vertices (minimal hitting sets), making it possible in terms of time to optimize the troubleshooting procedures on the ground and, as a result, to reduce maintenance costs.
Furthermore, the level of completeness of the final diagnosis is increased. In fact, the diagnosis is expressed on the basis of the candidate items of the critical event chart. Due to its construction, all the known origins capable of explaining the subsequent malfunctions are covered: line-replaceable units (LRU), software, cables or operational conditions such as a reset of an item of equipment or abnormal operating conditions.
Moreover, the relationships established between a diagnosis and notified messages or warnings which can be consulted on the critical event chart can be useful during online aircraft maintenance operations at a stopover for resolving causes associated with a particular symptom (ECAM type messages, warnings, etc.) reported by the pilot in a flight logbook. By using the critical event chart, the diagnostic aid system does not find a correlation relationship between failures and symptoms but establishes causality relationships consistent with the safety analyses, which can in particular be used in enquiries, in particular in the context of accidents.
Moreover, combined with a diagnostic result, the critical event chart can be used in a troubleshooting procedure. In fact, such a procedure typically consists of testing the lower branches of the chart, linked to the candidate items, concerning which there is uncertainty with respect to failures, because the set of information notified was not sufficient to remove this uncertainty. In order to remove ambiguities, the troubleshooting procedure can rely on the chart in order to clearly define the areas of uncertainty, then refer to new types of notifications provided by ACMF parameters or avionic test results.
Turning again to
Thus, the step referenced 120 in
This step can be broken down into two parts.
During a first part of this step, a search is carried out in the troubleshooting manual for the references of the procedures relating to testing all the candidate items of each previously calculated minimal vertex the perimeter procedure of which is minimal. This set of procedures forms, for each minimal vertex, an optimal list of troubleshooting procedures.
A second part involves identifying the procedures which are common to several vertices.
The information thus obtained, linked to the troubleshooting procedures, is advantageously combined with the diagnostic report in order to allow an optimal and effective failure test.
It is noted here that the search for procedures in the troubleshooting manual as described above can be improved by assigning priorities to the procedures, for example according to their execution time, from the quickest to be executed to the longest, or according to their implementation, favouring those that do not require any tool over those requiring ground specific equipment (GSE).
By way of illustration, it is assumed here that the problem F1 the presence of which is reflected by group P1 is diagnosed by the minimal vertices V1={L1, L2} or V2={L3} and that the problem F2 the presence of which is reflected by group P2 is diagnosed by the minimal vertex V3={L1, L4}. It is also assumed that the troubleshooting manual contains the following procedures:
TSM1: procedure relating to testing LRUs L1, L2 and L4
TSM2: procedure relating to testing LRUs L1 and L3
TSM3: procedure relating to testing LRU L3
TSM4: procedure relating to testing LRU L3
Consequently, the result obtained after completing step 120 of selection of procedures is the following
for the problem F1 the presence of which is reflected by group P1,
the minimal vertex V1 is optimally dealt with by the procedure TSM1; and
the minimal vertex V2 is optimally dealt with by the procedure TSM3 or the procedure TSM4;
for the problem F2 the presence of which is reflected by group P2,
the minimal vertex V3 is optimally dealt with by the procedure TSM1.
The procedure TSM1 is therefore common to the resolution of problems F1 and F2 the presence of which is reflected by groups P1 and P2. This procedure is therefore favoured over the others.
The advantages achieved by such a step of selecting the failure resolution procedures are in particular the following:
Turning again to
A first step (step 900) relates to producing a list of detections of failure notifications from the critical event chart for which a failure tolerance analysis must be carried out. In other words, step 900 involves selecting notifications of failures capable of being detected and used in the critical event chart, for which a failure tolerance analysis must be carried out. Such a list of selected detections can be predetermined, produced by an operative or produced automatically according to given criteria.
Attributes are advantageously associated with each selected detection. Such attributes comprise, for example, the following attributes:
a reference to a family associated with the detection, according to a predetermined classification, capable in particular of comprising elements such as aircraft_effect, maintenance_effect and operation_effect; and
a degree of the associated operational impact, on a predetermined scale, that can in particular comprise three levels (low, average and high).
These attributes are not necessarily used during failure tolerance calculations but are helpful in deciding whether or not to undertake a preventive maintenance activity.
By way of illustration, the message ECAM EM1 in
In a subsequent step (step 905), the determination of the failure tolerance is carried out. It involves in particular determining, for each of the selected detections, whether or not the corresponding failure tolerance has been encroached and identifying the pathways of the critical event chart which can lead to the corresponding selected detection, starting from the candidate items suspected by the diagnostics previously carried out (step 115 of
It is noted that the detections selected during step 900 are situated in general at the top of the critical event chart because they refer to high-level critical events. By way of illustration, such a detection can relate to an ECAM message reporting a loss of function of the aircraft to the pilot, who must apply an adapted piloting procedure as a consequence (flight crew operating manual (FCOM) procedures).
Step 905 thus involves identifying the list of previously selected detections which are such that at least one candidate item is suspected on at least one branch of the chart leading to this detection. Only these detections are advantageously investigated subsequently because these are the only ones which are impacted by the suspected failures in the aircraft. In fact, the distance which separates them from total unavailability is reduced due to the failures.
It should be noted here that in certain circumstances, for example according to the phases of the flight, warning messages are not immediately displayed in order to avoid disturbing the pilot. Consequently, failures can exist which have not been indicated to the pilot. It can therefore be useful to know preventively that a warning is imminent.
As shown in
The diagnostics carried out during step 115 of
During step 910, subcharts are extracted from the previously produced critical event chart (step 100 of
In a subsequent step (step 915), the detections of notifications of events of the set SG belonging to the list of the selected detections produced in step 900 are identified. This step makes it possible to obtain a list of selected detections (for which a failure tolerance analysis must be carried out) which can be notified in the near future due to their association with suspect candidate items (due to the fact that these detections belong to the set SG). This list of detections is called hereinafter the list of imminent effects.
The minimal vertices of candidate items are then calculated for each detection Ii of the list of imminent effects (step 920). To this end, it is possible to use the method described previously with reference to
Ii→Vi={v1={CandItemn}n, v2={CandItemm}m, . . . }
where {CandItemn}n represents a set of candidate items (CandItem) defined by the set of the (not necessarily continuous) values of the index n.
The candidate items of the set Vi (set of the minimal vertices) are then selected (step 925), for each detection Ii of the list of imminent effects, so as to retain only the vertices (sets of candidate items) which comprise at least one candidate item suspected by the diagnostics carried out previously (during step 115 of
Ii→Wi∈Vi
Wi={w1={CandItemo}o, w2={CandItemp}p, . . . }
Each of the vertices wi therefore contains at least one suspect candidate item.
It is observed here that, alternatively, the selection of the vertices which comprise at least one candidate item suspected by the diagnostics previously performed can be carried out within the short-length minimal vertices, i.e. those containing a limited number of candidate items (and not among all the minimal vertices). The maximum length of the minimal vertices to be taken into account can be predetermined.
During a subsequent step, a remaining distance before imminent effect is calculated for each of the vertices wi (step 930). A remaining distance before imminent effect is calculated as being equal to the number of candidate items present in the vertex in question and not suspected during step 115 of
wi→di=Card{CandItemj} such that CandItemj∈Wi and CandItemi is not a suspect candidate item.
The data thus obtained are used to produce a failure tolerance report (step 935). More specifically, the failure tolerance report can in particular comprise the list of the detections Ii from the list of imminent effects with their attributes (for example the family and degree of operational impact), the preventive diagnoses relating thereto and the remaining distance before imminent effect for each of these diagnoses.
Table 2 given in the appendix represents an example of a failure tolerance report generated by the algorithm described with reference to
The preparation of failure tolerance reports using the algorithm described with reference to
Turning again to
As shown, a first step (step 1100) relates to accessing a diagnostic history, typically a diagnostic history of the n previous flights, for example of the four previous flights (n=4) or of the fifteen previous flights (n=15).
Lists of candidate items belonging to the sets of previously identified suspect candidate items (set Ei) are then produced (step 1105).
According to a particular embodiment, several lists of suspect candidate items are produced as a function of the cardinality of the sets of suspect candidate items Ei. More specifically, a list LSr,s is constructed for each cardinality r of the sets Ei (r varying from 1 to p) and each preceding flight s (s varying from 1 to n). The maximum value p of the cardinalities of all the sets Ei to be taken into account is preferably predetermined, for example p=4. In other words, p sets {LSr,s}r=1 . . . p are defined for each flight s.
During a subsequent step (step 1110), diagnostic persistence weightings are calculated for each of the flights of the history, for each suspect candidate item in the current flight.
According to a particular embodiment, the diagnostic persistence weightings are calculated as follows:
only the diagnostic persistence weightings of the suspect candidate items in the current flight are calculated. The suspect candidate items in the previous flights but not in the current flight are disregarded in this calculation;
if a candidate item CandItem is suspected in a flight s its diagnostic persistence weighting is zero (PCandItem,s=0), for this flight, if it is no longer suspected in the subsequent flight (s−1); and
if a candidate item is suspected in a flight s and is still suspected in the subsequent flight (s−1), its diagnostic persistence weighting PCandItem,s, for this flight (s), is defined as being the diagnostic persistence weighting of this candidate item PCandItem,s-1 in the subsequent flight (s−1) incremented by a value linked to the length of time since the flight (s) and to the cardinality (r) of the set LSr,s to which the candidate item belongs or zero if the suspect candidate item does not belong to any LSr,s set. The diagnostic persistence weighting of this candidate item on the subsequent flight (s) can then be defined by the following relationship,
where f(s) is an increasing function of s allowing the weighting of the old flights to be reduced in order to limit the influence of old diagnostics for which maintenance operations may have been carried out. By way of illustration, the function f(s) can be defined as follows:
f(s)=s=s
Table 3 in the appendix gives an example of a list of suspect candidate items and associated diagnostic persistence weightings. Each line represents here a flight identified by the value of the index given in the first column. The second column states problem(s) identified during the corresponding flight. For example, the problem FE,4 was identified during the current flight (flight s=1). The third column gives the minimal vertices obtained in response to step 115 of
where i=1, f(s)=s=3.
In a next step (step 1115), diagnostic historic persistence weightings are calculated for each suspect candidate item in the current flight. The diagnostic historic persistence weighting of a suspect candidate item, called PHCandItem is the maximum value of the persistence weighting obtained by this candidate item over all of the flights. Such a weighting can be defined by the following relationship:
PH
CandItem=maxs(PCandItem,s)
By way of illustration and reverting to the example given with reference to
Table 3 of the appendix, the candidate item S1 has a historic persistence weighting equal to 2.08 which represents the maximum value of the diagnostic persistence weightings of this item which develops from 1 to 1.5 to 1.83 to 2.08 for flights s=1, 2, 3 and 4, respectively. Similarly, the candidate item L2 has a historic persistence weighting equal to 1.58 and the candidate item L5 has a historic persistence weighting equal to 1.
In a subsequent step (step 1120), the historic persistence weightings are used to rank the minimal vertices in the diagnostics of the same problem, from most relevant to least relevant. To this end, rules can be used, in particular the following rules:
Such a step makes it possible to obtain a diagnosis classified by relevance.
By way of illustration and reverting to the example given with reference to Table 3 in the appendix, three minimal vertices ({S1}, {L5}, {L2}) were identified for the current flight. These three minimal vertices have the same cardinality (1). Nevertheless, by using the sum of the historic persistence weighting of each candidate item of each minimal vertex (2.08, 1.58 and 1, respectively), it is possible to classify them: {S1}, {L2}, {L5}.
The historic persistence weightings can also be used in order to prioritize the suspect candidate items absolutely, for example by using the following rules:
By way of illustration, it is assumed that a diagnosis of the present condition comprises the following minimal vertices,
{L1, L2} or {L3} or {L4, L5}
involving the following candidate items the historic persistence weightings of which have been calculated and are given in parentheses:
L1 (3), L2 (2), L3 (1), L4 (1), L5 (2)
The use of the rules given previously makes it possible, on the basis of the cardinalities of the minimal vertices and of the historic relevance weightings, to prioritize the candidate items in the following order:
1. L3
2. L1
3. L2
4. L5
5. L4
This prioritization results from the fact that the candidate item L3 is involved in a vertex with a cardinality of one while all the other candidate items are involved in vertices with a cardinality greater than one. Moreover, in vertices with a cardinality of two, vertex {L1, L2} is more relevant than vertex {L4, L5} due to the sum of the corresponding historic persistence weightings (candidate items L1 and L2 are therefore more important than candidate items L4 and L5). Finally, candidate item L1 has a greater historic persistence weighting than that of candidate item L2 and candidate item L5 has a greater historic persistence weighting than that of candidate item L4.
The diagnostic history of the n previous flights and the critical event chart can be used to rank diagnostic problems relating to a given flight.
To this end, a first step (step 1130) consists of identifying the possible order of the problems Fi diagnosed flight after flight by using an order relationship such as the following: a problem Fi diagnosed on a flight prior to the present flight totally covers a problem Fj diagnosed on the present flight only if all the minimal vertices diagnosing the problem Fi are included in the list of minimal vertices diagnosing the problem Fj or minimizing the latter. By way of illustration, it is recalled that group {A} minimizes group {A, B} and that group {A} is included in the set of groups {{A}, {C, D}, {E, F}}. Such a relationship is denoted Fi→Fj.
In a subsequent step (step 1135), persistence weightings of the problems diagnosed during the current flight are calculated. Such a calculation can in particular be carried out according to the following steps:
finding the maximum problem length sequence F0, F-1, . . . , F-k diagnosed on consecutive flights such that F0 is detected on the current flight, F-1 is detected on the previous flight and so on as far as the problem F-k (diagnosed during the kth flight prior to the current flight) where k>0 and
F-k→F-(k−1)→ . . . →F-1→F0
if such a sequence exists, the persistence weighting of the problem F0 is equal to k and, if it does not exist, the persistence weighting of the problem F0 is equal to zero.
The problems Fi diagnosed during the current flight are then ranked by priority according to their persistence weighting, from the greatest to the smallest (step 1140) such that a problem having a persistence weighting greater than that of another problem takes priority over the latter.
In the case of equality between two problems, the composition of their respective diagnostics is advantageously used for separating them. To this end, the ranks of the minimal vertices diagnosing each problem are calculated. The rank of a minimal vertex is here equal to the number of candidate items of which it is constituted. By way of illustration, the minimal vertex {CandItemA, CandItemB} is ranked two while the minimal vertex {CandItemC, CandItemD, CandItemE} is ranked three. The problems are then classified by using the ranks of the minimal vertices such that a problem diagnosed by vertices having a lower rank takes priority over a problem diagnosed by vertices of higher rank. Thus, for example, if F1 and F2 are problems having the same persistence weighting, the problem F1, diagnosed by the minimal vertices {CandItemA} and {CandItemB, CandItemC}, takes priority with respect to the problem F2 diagnosed by the minimal vertex {CandItemD, CandItemE}. In the case of equality, the problems can be separated as a function of the number of minimal vertices, the one having fewest having the greatest priority.
It is assumed here that the MM1 and MM2 messages were notified during the previous flight and that the MM3 message was notified during the current flight. The problem F1 is diagnosed during the previous flight by the minimal vertex {S1}. The problem F2 is diagnosed during the current flight by the minimal vertices {S1}, {L1}, {L2}.
The problem F1 totally covers the problem F2 (F1→F2). The problem F1 therefore takes priority over the problem F2.
The steps described with reference to
promoting maintenance of the more often suspected candidate items, which avoids leaving an item in unresolved failure for too long. This is particularly useful in the case of operation of an aircraft that does not return to its main base after a series of flights and on which different maintenance teams work. In fact, in this case, the maintenance operatives are not the same staff from one airport to another, only making a spot check on the aircraft in a given airport. The results obtained using the steps described previously make it possible to benefit from the history of previous diagnoses; and
Finally, reverting to
According to a particular embodiment, the diagnostic aid system is implemented in an on-board maintenance system of an aircraft. The notifications received by the diagnostic aid system are, preferably, failure reports of the ARINC 624 type sent by the systems of the aircraft, message notifications of the ECAM type, availability messages and/or warnings transmitted by the FWS. The algorithm described with reference to
The version of the critical event chart on board an aircraft can be a simplified version without certain branches, which nevertheless makes it possible to obtain a first diagnostic result and thus to optimize the operation and maintenance procedures. A full version of the critical event chart can be used in a second embodiment, to make it possible for example for an aircraft manufacturer to sell a detailed operational and diagnostic service to an airline.
The diagnostic aid results are advantageously stored on board the aircraft. They can then be displayed via a human-machine interface. They can also be sent to a data processing system on the ground via a communication system (for example the ACARS system).
Such a system allows a low latency between the notifications of the monitored systems and the execution of the diagnostic aid algorithm. Moreover, the availability, in real time, of the diagnostic aid results on board the aircraft gives it diagnostic autonomy.
According to another embodiment, the diagnostic aid algorithm is produced by a data processing system on the ground on the basis of data transmitted by an aircraft. The diagnostic aid algorithm can be executed by the aircraft manufacturer, who preferably centralizes and validates the diagnostic aid results of several aircraft, these results being capable of validation by experts. The results, comprising a set of minimal vertices representing minimal diagnostics, can then be transmitted to the airlines operating the aircraft via a communication network such as the Internet. Alternatively or in addition, the diagnostic aid algorithm can be implemented within an airline operating the aircraft, the aircraft manufacturer being able to supply the diagnostic aid algorithm in the form of software applications. The latter can be produced with an open, modular interface architecture, allowing their integration with other aircraft fleet management services.
The data processing system 1420 comprises a knowledge base 1430 comprising in particular at least one critical event chart 1435 associated with a system of the aircraft in question. This critical event chart is used in combination with the notification messages received in order to establish a diagnostic aid according to the invention using, for example, the algorithms described with reference to
Such an embodiment makes it possible to use a centralized diagnostic aid system on the ground that can be used for establishing a diagnostic aid for several aircraft. Moreover, the diagnostic aid system can be incorporated, for example, into another maintenance information system the purpose of which is to schedule maintenance tasks and manage spares logistics. The use of such embodiment makes it possible to reduce considerably the time necessary for establishing a diagnosis. Thus, it has been observed that, combined with a trouble-shooting procedure, the gain in time can reach a factor of 50.
It is observed here that the previously-described method can also be used in post-processing of reports produced in real time, generally called current flight reports (CFR) automatically sent by an aircraft when in flight.
This method makes it possible to provide a preventive diagnostic aid on board the aircraft which allows experts on the ground to recommend preventive maintenance operations in order to avoid imminent effects that are very prejudicial to its operation.
By way of illustration, this method makes it possible to warn of imminent inhibition of the passenger cabin pressurization system owing to the non-confirmation of the closed & latched & locked status of one or more doors. This inhibition of the pressurization of the aircraft, if not prevented, poses a severe problem for the company, because it prevents take-off and the pilots are alerted of it at the embarkation door, when all the passengers are on board. By being informed in advance, the company can schedule the maintenance operations on the doors well in advance, and finally avoid any inhibition of cabin pressurization.
The device 1500 also preferably has a hard disk 1535 capable of containing the abovementioned programs as well as items of information already processed or to be processed according to the invention and a memory card reader 1540 suitable for receiving a memory card 1545 and reading from or writing to it data already processed or to be processed according to the invention.
The communication bus allows the communication and interoperability between the different elements included in the device 1500 or associated therewith. The representation of the bus is non-limitative and, in particular, the central unit is capable of communicating instructions to any element of the devices 1500 directly or via another element of the device 1500.
The executable code of each program allowing the programmable devices to implement the procedure according to the invention can be stored, for example, on the hard disk 1535 or in the fixed storage 1515.
According to a variant, the memory card 1545 can contain items of information, in particular items of information to be processed according to the invention, as well as the executable code of the abovementioned programs which, once read by the device 1500, is stored on the hard disk 1535.
According to another variant, the executable code of the programs and the items of information to be processed according to the invention can be received, at least partially, via the interface 1550, to be stored in an identical manner to that described previously.
More generally, the program(s) as well as the items of information to be processed according to the invention can be loaded into one of the storage means of the device 1500 before being executed.
The central processor unit 1510 will control and manage the execution of the instructions or portions of software code of the program(s) according to the invention, instructions which are stored on the hard disk 1535 or in the read only memory 1515 or in the other abovementioned storage elements. When powered up, the program(s) which are stored in a non-volatile memory, for example the hard disk 1535 or the read only memory 1515, are transferred to the random access memory 1520 which then contains the executable code of the program(s) according to the invention, as well as the registers for storing the variables and parameters necessary for implementation of the invention.
Naturally, in order to satisfy specific requirements, a person of ordinary skill in the field of the invention can apply modifications to the above description.
As is apparent from the foregoing specification, the invention is susceptible of being embodied with various alterations and modifications which may differ particularly from those that have been described in the preceding specification and description. It should be understood that I wish to embody within the scope of the patent warranted hereon all such modifications as reasonably and properly come within the scope of my contribution to the art.
| Number | Date | Country | Kind |
|---|---|---|---|
| 1253383 | Apr 2012 | FR | national |
