Patent Application
20210027556
The present invention relates to an abnormality sign diagnosis apparatus and an abnormality sign diagnosis method for diagnosing abnormality signs of equipment.
In operating various kinds of equipment, it is preferable to grasp degrees of their deterioration and to diagnose their abnormality signs before they actually fail in view of the need for maintaining safety or performance levels. The diagnosis of abnormality signs of a motor vehicle as typical equipment is important for ensuring its safety. As a recent trend in the vehicles, some devices have started to be introduced to improve the safety of vehicle users. However, additional costs involved have thwarted the popularization of such devices.
Meanwhile, the surge in physical distribution and expansion of the car-sharing business tend to increase the operating time of both commercial and private vehicles. Thus it is more necessary than ever to monitor the status of the vehicles and detect their abnormality as soon as possible so as to notify drivers, car dealers, transport companies, and fleet management companies of the findings.
There exist techniques for diagnosing abnormality signs using CAN data and OBD data. These prediction techniques involve using measurement data such as CAN data and OBD data on the operating time of vehicles to detect equipment abnormality from the difference between previously acquired data on the vehicle in normal driving on one hand and the measurement data on the vehicle at a time targeted for diagnosis on the other hand.
Patent Document 1 cited below describes failure diagnosis that involves comparing driving data on multiple parameters stored in an electronic control unit (ECU) of a vehicle upon failure with reference values constituted by driving data at normal time.
With the above-mentioned existing techniques, it has been difficult to detect abnormality accurately by comparison with normal data if the surrounding environment and status of the vehicle in driving are different. As an example, in the case of detecting abnormality in the brake system of the vehicle, it is difficult to make accurate detection if the status in which the brake system is activated does not match. For instance, depending on the location, on the weather, or on the traffic conditions at which the brake is activated, the behavior resulting from similar activation of the brake may vary, which makes appropriate detection difficult.
In view of the above, the present invention aims to provide an abnormality sign diagnosis apparatus and an abnormality sign diagnosis method for enabling more accurate predictive abnormality detection by comparison between the behaviors having taken place in similar environments.
Thus the present invention provides “an abnormality sign diagnosis apparatus that uses measurement data on equipment in operation to find a difference between previously acquired data on the equipment in normal operation and the measurement data on the equipment at a time targeted for diagnosis so as to detect abnormality of the equipment, the abnormality sign diagnosis apparatus including: input means that inputs the measurement data on the equipment in operation and a surrounding environment and status of the equipment in operation; storage means that stores the data on the equipment in normal operation; and processing means that selects from the storage means the data on the equipment in normal operation in a surrounding environment and status of the equipment, the surrounding environment and the status being similar to the surrounding environment and the status of the equipment at a time which is targeted for diagnosis and at which the data on the equipment in normal operation has been measured, the processing means further detecting abnormality of the equipment using the data on the equipment in normal operation at the time targeted for diagnosis and the data on the equipment in normal operation selected from the storage means.”
The present invention also provides “an abnormality sign diagnosis method that uses measurement data on equipment in operation to find a difference between previously acquired data on the equipment in normal operation and the measurement data on the equipment at a time targeted for diagnosis so as to detect abnormality of the equipment, the abnormality sign diagnosis method including: by a processing means, obtaining the measurement data on the equipment in operation and a surrounding environment and status of the equipment in operation; storing the data on the equipment in normal operation; and using the data on the equipment in normal operation at a time targeted for diagnosis and previously stored data on the equipment in normal operation in a surrounding environment and status of the equipment, the surrounding environment and the status being similar to the surrounding environment and the status of the equipment at the time at which the data on the equipment in normal operation has been measured, so as to detect abnormality of the equipment.”
According to the present invention, it is possible to detect abnormality more accurately by comparison with previously acquired data in the matching status in which the brake has been activated, for example.
Some embodiments of the present invention are described below in detail with reference to the accompanying drawings.
Of the constituent elements in
The mobile information terminal in the vehicle includes mobile information terminal sensor data acquisition means for acquiring sensor data such as velocity, acceleration, and GPS data from sensors installed inside the vehicle. The mobile information terminal in the vehicle outputs mobile information terminal sensor data D11 thus acquired. The mobile information terminal in the vehicle further includes abnormality detection and reporting means for displaying and reporting, inside the vehicle, diagnostic result data transmitted from the vehicle abnormality sign diagnosis apparatus 100. Some of the functions implemented by the mobile information terminal may be executed by a mobile terminal carried by a passenger on the vehicle.
As part of the input section 110, the stationary information terminal may be an information terminal installed at a meteorological observatory, for example. The stationary information terminal acquires data such as weather and map data on the environment external to the vehicle, and transmits the acquired data as network data D12 via the network N. The information terminal includes network data acquisition means for data acquisition and transmission purposes.
Returning to
Whereas the vehicle abnormality sign diagnosis apparatus 100 in
In
The braking period extraction means 151 in the data processing section 150 extracts, from the mobile terminal and network data D0 output by the data detection section 130, the mobile information terminal sensor data D11 corresponding solely to the period in which the brake is activated, on the basis of an empirical model stored in the operation database DB. The empirical model here is a model for determining whether the vehicle is decelerating, i.e., whether the brake is activated, in accordance with a motion equation of the vehicle. The empirical model will be discussed later in detail.
The parking/stopping data exclusion means 152 excludes the data on the vehicle during parking and stopping by searching for zero-velocity data through vehicle velocity data. Here, the data other than the data on the vehicle in driving, such as the data on the vehicle during parking and stopping, is excluded as unnecessary data because current and past operations in normal driving are targeted for comparison.
The vehicle straight-ahead travel data extraction means 153 extracts only the data on the vehicle traveling straight based on vehicle velocity data and vehicle acceleration data included in the mobile information terminal sensor data D11 by excluding the data on the vehicle turning right or left. In the first embodiment, the straight-ahead travel data is extracted because an empirical model is created from a straight-ahead driving equation. If there is an empirical model taking vehicle turns into consideration, the extraction of the straight-ahead travel data may be omitted.
The road gradient calculation means 154 performs road gradient calculations using vehicle GPS information included in the mobile information terminal sensor data D11.
The above-described series of processing by the data detection section 130 or by the data processing section 150 is carried out repeatedly in the computer system. The result of the processing, every time the processing takes place with the vehicle in driving, is obtained as previously processed driving data D4. The previously processed driving data D4 is stored successively into the operation database DB. The previously processed driving data D4 will be discussed later with reference to
The previously processed driving data D4 immediately after shipment of the vehicle from the factory or immediately after subsequent periodical maintenance of the vehicle may be said to be what is known as training data on the vehicle in the normal state. On the other hand, the previously processed driving data D4 obtained after driving time reflects the vehicle in the current state. There is thought to be an increasing difference between the training data and the current-state data.
The similar external environment detection section 160 focuses on the external environment data from among the network data D12 stored consecutively into the operation database DB. The similar external environment detection section 160 extracts two kinds of data: existing driving data D5 as past data in an external environment similar to the current external environment (previously processed driving data D4 at a past point in time), and the most recent previously processed driving data D4. Details of the processing by the similar external environment detection section 160 will be discussed later with reference to
The abnormal state detection section 170 detects an abnormal state by comparing the previously processed driving data D4 in the current state with the previously processed driving data D4 in the past, both data being from similar environments. The result of the detection is output by the output section 120 as abnormal state display data D6. As depicted in
What follows is a detailed explanation of the processing according to the present invention on the basis of specific data examples. In a typical data flow in
Of these items of data, the “velocity, acceleration, and GPS” data D1 is configured with a velocity X (D13), a velocity Y (D14), acceleration X (D15), acceleration Y (D16), an X coordinate (D17), and a Y coordinate (D18) of the vehicle at a given time interval D12. The velocity X (D13) is the velocity of the vehicle traveling straight ahead. The velocity Y (D14) is the velocity of the vehicle traveling laterally. The acceleration X (D15) is the acceleration of the vehicle traveling straight ahead. The acceleration Y (D16) is the acceleration of the vehicle traveling laterally. The data with a given timestamp is provided with an ID (D11) starting from 1.
The times are given in units of ms. The times that have elapsed from the beginning are written. The velocities are given in units of km/h and written as one-dimensional data strings. The acceleration is given in units of m/s2 and written as a positive one-dimensional data string when the vehicle accelerates and as a negative one-dimensional data string when the vehicle decelerates. The X and Y coordinates are data obtained from GPS and written as two-dimensional data strings indicative of longitude and latitude. For this apparatus, a reference location is determined in advance, and the distance of the vehicle from the reference location is calculated on the basis of longitude and latitude information. The X coordinate is in the longitude direction and the Y coordinate is in the latitude direction. The X and Y coordinates are given in units of m. It is to be noted that these units and coordinate arrangements are only examples. The formats of the data may be changed.
The “map information” data D2 is based on maps obtained from external data acquisition means of the information terminal. The maps constitute three-dimensional data and include longitude, latitude, and height information. As with GPS information, the data in the longitude direction and the data in the latitude direction are converted to an X coordinate D21 and a Y coordinate D22. The height information is written as a Z coordinate D23.
The “weather information” data D3 is based on weather information obtained from the external data acquisition means of the information terminal. The “weather information” data D3 includes time D31, weather D32, temperature D33, and humidity D34.
The operation database DB in
In
In
The mathematical expression (1) may be transformed into the mathematical expression (2) below. According to the expression (2), the acceleration ak is given as a linear function of the braking force Fk, the velocity squared vk2, and the road gradient θk at the same point in time. The motion equation and travel physical model in this case are stored in the operation database DB.
The empirical model obtained from the travel physical model is explained with reference to
In the first processing step S710 in the flowchart of
For that reason, the braking period is extracted using the empirical model prepared beforehand in the operation database DB. Using the empirical model permits determination of the braking state at a given point in time. An aggregate of data on the braking states associated with the time of the determination constitutes the braking period. This makes it possible to extract the braking period.
In a processing step S720, vehicle straight-ahead travel data is extracted by determination based on the lateral velocity and acceleration. As depicted in
In a processing step S730, parking/stopping data is excluded by searching for locations at zero velocity. In order to diagnose brake failure, the data on the use of the brake is effectively utilized. The other data is excluded. The mobile information terminal sensor data D11 depicted in
In a processing step S740, the road gradient is calculated by computing the height using the XY coordinates D17 and D18 of GPS information and the map information D2 in combination. The mobile information terminal sensor data D11 includes the GPS information as well as the XY coordinate data D17 and D18 indicative of longitude and latitude. Network data 220 includes the map information D2. The road gradient is calculated using both the GPS information and the map information.
The previously processed driving data D4 thus obtained is sent successively to the operation database DB for storage therein.
Next, the similar external environment detection section 160 uses as its input data the previously processed driving data D4 stored in the operation database DB. Given the input data, the similar external environment detection section 160 searches for and extracts existing operation data in a similar external environment as the existing driving data D5 in the similar external environment.
For the present embodiment, the vehicle behavior and the external environment are indicated in
In a processing step S920, the degrees of influence of the external environment elements are evaluated. Here, the operation database DB stores existing vehicle operation data, which is used to evaluate the degrees of influence of the external environment elements. The method of evaluation involves dividing the existing operation data into two groups A and B, extracting an appropriate braking period from the group A, and searching the group B for the most similar and the most dissimilar vehicle behaviors.
The method discussed in Non-Patent Document 1 may be adopted as the method for determining similarity. The method may be used to calculate the similarity between the acceleration time series data on two vehicle behaviors. In the case where the vehicle behaviors are similar and so are the external elements, the degree of influence is incremented by 1. In the opposite case, the degree of influence is decremented by 1. For example, a search is made through the group B for the most similar acceleration time series data T2 to the acceleration time series data T1 in a single braking period in the group A. The degrees of similarity are then evaluated between the external environment elements of the time series data T2 and those of the time series data T1. For example, in the case of initial velocity, the difference in initial velocity between the time series data T1 and the time series data T2 is written as an absolute value T12. A search is further made through the group B for the most dissimilar acceleration time series data T3 from the acceleration time series data T1. The difference in initial velocity between the data T1 and the data T3 is calculated as an absolute value T13. In the case where the absolute value T12 of the difference in initial velocity is larger than the absolute value T13, the degree of influence of the external environment element “initial velocity” is incremented by 1. In the case where the absolute value T12 is smaller than the absolute value T13, the degree of influence is decremented by 1. In the group A, the degrees of influence of all external environment elements are calculated using all braking periods.
In a processing step S930, similar external environment elements are selected. The external environment elements are arrayed in descending order of their degrees of influence. The elements whose degrees of influence are decremented are omitted. The external environment elements are then weighted in accordance with the degrees of their influence calculated in the processing step S920. For example, the weight is 1 for the external environment element with the highest degree of influence. The degrees of influence of the remaining external environment elements are calculated using the following mathematical expression (3):
[Math. 3]
Weight=degree of influence of a given external environment element/the highest degree of influence (3)
In a processing step S940, the vehicle behavior in the similar external environment is output. A search is made for the most similar external environment using the weights calculated in the processing step S930. Because there are multiple external environment elements, the degrees of similarity of these external environment elements are evaluated using their individual weights. For example, if there are external environment elements A, B, and C, the degrees of similarity between two of these external environment elements are written as Sa, Sb, and Sc, and their weights are given as Wa, Wb, and Wc. The degree of similarity in this case is obtained by the following mathematical expression (4):
[Math. 4]
Degree of similarity=Wa*Sa+Wb*Sb+Wc*Sc (4)
The method for calculating the degrees of similarity Sa, Sb, and Sc of the external environment elements is explained here. In the previously processed driving data D4 in
Of the weather information as another external environment element, the temperature D33 and humidity D34 are displayed numerically. The difference between the values is obtained as an absolute value for evaluation. The degree of similarity is low in the case where the difference in absolute value is large, and is high where the difference in absolute value is small. All differences in absolute value are normalized to numerals ranging from 0 to 1 and written with Dif. The degree of similarity is calculated as 1-Dif.
The velocity D13, which is another external environment element, is a time series numerical string. The degree of similarity of this data item is calculated using the DTW method described in Non-Patent Document 1. The initial and end velocities in the braking period are numerically displayed, and calculated in the same manner as temperature and humidity.
After the degree of similarity of each external environment element is calculated, a search can be made for the most similar external environment to the external environment of test driving data. The vehicle behavior in the external environment resulting from the search is output. What is output by the present embodiment is the vehicle acceleration time series data in the most similar external environment.
The existing driving data D5 in the similar external environment and the previously processed driving data D4 constitute the input data to the abnormal state detection section 170. The abnormal state display data D6 is the output data from the abnormal state detection section 170. The concept of the abnormal state detection section 170 is explained below with reference to
In this case, the vehicle abnormality sign diagnosis tool is started by use of the mobile information terminal. An example of a started input screen 90 appears as illustrated in
The display in
Of these areas, the overall display area 1230 displays the degree of abnormality calculated by comparison with the training data. The time of maintenance here denotes the point in time immediately following maintenance of the vehicle at the maintenance factory. At this time, the state of the vehicle is assumed to be normal. After this time, driving data is acquired over a predetermined driving distance or following a predetermined driving time as the training data. The driving data acquired subsequent to the predetermined driving distance or driving time is referred to as test data. The driving distance or driving time for acquisition of the test data is set independently of the training data.
In the parameter setting display area 1240, the items of training data acquisition settings, test data acquisition settings, and abnormality degree threshold value settings are each displayed with selectable and settable settings. The training data acquisition settings offer two options: “set by driving distance” in km, and “set by driving time” in days. Likewise, the test data acquisition settings offer “set by driving distance” and “set by driving time” options. The abnormality degree threshold value settings have two options: set by numerical input, and “use recommended value.”
The diagnosis execution display area 1250 is provided with a “perform abnormality sign diagnosis” button. After parameters are input to the parameter setting display area 1240 and the “perform abnormality sign diagnosis” button is pressed, the acquisition of driving data and the abnormality sign diagnosis are started.
The screen display of the “abnormality sign diagnosis results 1220” tab depicted in
The abnormality degree display area 1310 provides a graphic representation. The horizontal axis stands for test data sequence and vertical axis denotes degrees of abnormality. This area displays calculated degrees of abnormality of the test data. The threshold value set by the abnormality sign diagnosis parameter tab is displayed as a horizontal line. The test data threshold value is displayed in green when not exceeded and in red when exceeded. Although the present embodiment uses the colors of green and red for the threshold value display, other colors may be used instead.
In addition, in the case where the degree of abnormality exceeds the threshold value, the alarm area 1320 displays a message “The degree of abnormality exceeds the threshold value. The brake system is recommended to be inspected.” The specific wording of the message here is an example. Some other suitable wording of the alarm may be adopted instead. As another alternative, the alarm may be replaced with “recommendations.”
In a second embodiment, the external input section 110 and the external output section 120 in
In
In the case where the abnormality sign diagnosis is performed on the brake system of the vehicle, a brake pedal force sensor and a brake pedal depression amount sensor may be used as the additional sensors. If the additional sensor data D14 acquired from the brake pedal force sensor or from the brake pedal depression amount sensor is used as the input to the brake system and the acceleration of the vehicle behavior is given as the output, then the relationship between the input and the output permits execution of the abnormality sign diagnosis on the brake system.
In the above-described first and second embodiments, the brake operation is discussed as an example in carrying out abnormality prediction. The abnormality prediction in this manner applies to other types of operations performed on the vehicle. For example, the abnormality prediction is performed by monitoring, storing, and comparing the difference between the amount of given operation performed on the vehicle by operating the parking brake, accelerator pedal, or steering wheel other than the foot brake on one hand, and the resulting state change on the other hand.
In the case of the operation other than that of the foot brake, it does not matter what models are prepared in the operation database DB or how to determine specific processes to be performed by the data processing section 150 using specific operation amounts. The point is that the data processing section 150 need only extract the operating period of a given operation performed on the vehicle and exclude the data other than that of normal operation, or need only select and evaluate the data in the similar environment for comparison purposes.
Examples are described above in which the vehicle is the target for control. Alternatively, the control target may be extended to general equipment.
The present invention is characterized in that the data regarding the normally operating vehicle as the equipment to be controlled is targeted for comparison and that the cases similar to the surrounding environment and status of the vehicle in driving are selected for comparison therebetween.
| Number | Date | Country | Kind |
|---|---|---|---|
| 2018-082844 | Apr 2018 | JP | national |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/JP2019/003156 | 1/30/2019 | WO | 00 |
