METHOD FOR CONTROLLING THE PRESSURE OF THE HEAT TRANSFER FLUID IN A HEATING SYSTEM

Abstract

A method for controlling the pressure of heat transfer fluid in a heating system including a thermo-sanitary appliance, in particular a gas boiler, the method including the following steps. Extrapolating minimum values of the fluid pressure from the original minimum values, within a time observation window corresponding to a given number of previous days. Identifying one or more ramps of the minimum values of the pressure of the fluid, within the time observation window. Calculating a plurality of characteristics relating to the trend of the pressure of the fluid on the basis of said one or more ramps of the minimum values. Providing the calculated characteristics to a machine learning classification model, adapted to produce a three-class classification, each of which is representative of the more or less high probability that in a future time window the thermo-sanitary appliance will block due to an excessively low value of fluid pressure.

Description

FIELD OF THE INVENTION

The object of the present invention is a method for controlling the efficiency state of a heating system, adapted to prevent an excessive pressure drop of the heat transfer fluid circulating in said system and to promptly signal the need for maintenance interventions.

The invention falls, without any limiting intent, in the field of thermo-sanitary appliances (such as for example gas boilers, heat pumps, hybrid boilers) for closed circuit heating systems in which a heat transfer fluid circulates that must remain stably within a suitable range of pressure values (typically about 1.2/1.5 bars) so that the system correctly performs the functions of space heating and production of sanitary water.

BACKGROUND OF THE INVENTION

For the sake of brevity and clarity of description, in the remainder of the present description reference shall be made exclusively to a gas boiler; however, it is understood that the teachings of the invention are similarly applicable to any type of thermo-sanitary appliance operating with a heat transfer fluid circulating in the system.

In the prior art systems, in case of reduction in the pressure of the heat transfer fluid beyond a certain threshold, generally due to leaks in the hydraulic circuit, the boiler produces a low-pressure warning and enters a blocking condition, in order to avoid major problems to the various components of the boiler and the system.

In order for the boiler to resume operation, the user or the service technician must supply additional water to the hydraulic circuit so that the pressure rises and the heat transfer fluid resumes normal circulation.

An example of a prior art document that aims to solve the problems set out above is the prior art patent DE102021111467, which describes a method for controlling the pressure of the heat transfer fluid in a heating system, by monitoring the minimum pressure values and the refill periods of such system.

It may also happen that the boiler operates normally with pressure values of the heat transfer fluid lower than the standard range, without having leaks from the circuit but only incurring a progressive but slow reduction in pressure over time: even in these conditions, however, the boiler can generate frequent blocking errors, with consequent inconveniences for the user.

SUMMARY OF THE INVENTION

The object of the present invention is to obviate such type of drawback, by providing a control method for monitoring the pressure values of the fluid circulating in the system and to avoid, or at least reduce, the blocking conditions and failure of the heating and thermo-sanitary appliance.

Another object of the present invention is that of providing a method operating effectively with the components and means already present in a typical gas boiler or in other thermo-sanitary appliances of the closed-circuit heating systems, without the need to resort to additional devices with increasing costs.

A further object of the present invention is that of providing methods for signaling the need to control the pressure of the fluid, through warning systems in the form of notifications that provide improved ease of use for the user.

These and other objects which shall appear clear hereinafter, are achieved with a method for controlling the efficiency state of a heating system, adapted to prevent an excessive drop in the pressure of the heat transfer fluid circulating in said system and to promptly signal the need for maintenance interventions, in accordance with the provisions of the independent claims.

Other objects may also be achieved by means of the additional features of the dependent claims.

BRIEF DESCRIPTION OF DRAWINGS

Further features of the present invention shall be better highlighted by the following description of a preferred embodiment, according to the patent claims and illustrated, purely by way of a non-limiting example, in the accompanying drawing tables, in which:

FIG. 1 shows the hydraulic diagram of a typical gas boiler, in which the control method according to the invention may be implemented;

FIGS. 2 and 3 schematically depict two examples of fluid pressure signals used by the control method according to the invention, in order to identify a plurality of pressure values of the heat transfer fluid circulating in the heating system (note that such FIGS. 2 and 3 are not correlated and consequential to each other, referring to different periods of execution of the method);

FIG. 4 is a graph showing the last step of the method according to the invention, producing a first type of notification; and

FIG. 5 is a graph showing the last step of the method according to the invention, producing a second type of notification.

DETAILED DESCRIPTION OF THE INVENTION

Please The features of a preferred variant of the control method according to the invention are now described using the references in the figures.

With reference to FIG. 1, with 1 is indicated a gas boiler, in particular a condensing gas boiler, wherein the components necessary for the operation thereof, well known to the man skilled in the art are housed; only those components strictly relevant to the object of the invention will be mentioned herein.

As is known, a boiler 1 comprises:

• • a heating water circuit (primary circuit 10), intended to supply the heating bodies adapted to the indoor heating,
  • and a circuit for the sanitary water (sanitary circuit 20), intended to heat the water intended for the users.

Anyone of the two circuits is provided with its own heat exchanger, respectively a “primary exchanger 100” and a “secondary exchanger 200”.

In the primary exchanger 100, the heat of the flame of a burner is transmitted to the heat transfer fluid (hereinafter abbreviated to “fluid”); in the secondary exchanger 200, the heat is instead transferred from the hot fluid circulating in the primary circuit 10 to the cold sanitary water circulating in the sanitary circuit 20.

As mentioned, in order for the heating system comprising the boiler 1 to operate correctly, it is necessary that the pressure of the fluid circulating in the primary circuit 10 (as well as in the secondary circuit 20 on the primary side) remains stable, or in any case above a certain reference threshold.

Reference 4 indicates a pressure sensor (or similar means, such as a pressure switch or a pressure gauge) relating to the hydraulic circuit, adapted to send a signal representative of the pressure value of the fluid circulating in the hydraulic heating circuit to a control unit (not shown in the figure).

In the systems of the prior art, on the basis of such value the control unit may produce a warning in the event of low pressure and block the boiler 1, until the circuit is topped up to restore the correct pressure value of the fluid.

The method of the present invention instead proposes to use such values coming from the pressure sensor 4 in combination with a plurality of further parameters, in order to extrapolate certain characteristics to be provided to a previously trained machine learning classification model, capable of returning a prediction, within a next time window, of the more or less high probability of the risk that the boiler 1 will go into a block due to an excessive drop in the pressure value of the fluid.

More precisely, the method that shall be described below uses a machine learning classifier model SVM, trained to recognize the behavior of the pressure signals and provide as an output a risk class that is representative of the probability that the boiler 1 will, preferably in 10 days, find itself in a blocking condition due to a low pressure value of the circulating fluid.

For the sake of brevity but without any limiting intent, hereinafter it will be assumed that such future time window T.prox is equal to the following 10 days.

According to a variant of the invention, the pressure values, the additional parameters, the data reprocessing and the production, as an output, of the risk class, are all processed by the same control unit, normally adapted to manage the usual functions of the boiler 1 and components thereof, with which it can communicate bidirectionally according to the most appropriate methods, for example via wired and/or wireless connections (radio type, such as WLAN, ZigBee, Bluetooth protocols or the like).

However, for the purposes of the invention, nothing prevents all or some of the data processing means underlying the operation of the method of the invention from being managed by additional control units or performed on devices external to the heating system: by way of a non-limiting example, such additional means may comprise smartphones, tablets, PDAs, remote servers (e.g. in the Cloud), provided that they are appropriately equipped with software and/or applications suitable and compatible for the purpose.

For the sake of brevity, hereinafter the description of the invention shall assume that all the processing capabilities reside in the control unit of the boiler 1 but, more generally, “control unit” means the entire set of data processing elements, united or distributed in multiple subsets, which are necessary to implement the method according to the invention.

Such control unit is advantageously provided and/or cooperating with:

• • means of acquisition and/or reception of input data for the method of the invention,
  • means for calculating and processing said input data, adapted to provide output data and/or information aimed at determining the risk class of the heating system,
  • memory means for storing, at least temporarily, said input and/or output data, and
  • means for transmitting said output data and/or information to a display and notification interface, said interface being able to consist of a typical HMI interface integrated in the heating system and/or in the aforementioned devices external to the heating system.

The control method shall now be described in detail with reference to the various sequential steps of which it is composed.

Step 1: extrapolation of the minimum values of the pressure of the fluid circulating in the system.

In Step 1, the method of the invention takes into account the pressure values of the fluid within an observation time window T.analysis, corresponding to a determined number of previous days, for example 20 days.

For the sake of brevity but without any limiting intent, hereinafter it shall be assumed that such observation time window T.analysis is equal to the previous 20 days.

Starting from the original minimum pressure values, measured by the pressure sensor 4 (or by similar means), the algorithm underlying the method extrapolates minimum pressure values in consecutive time intervals (for example 3 hours each) within said time window T.analysis, in accordance with rules and modalities shortly described in detail.

The graph of FIG. 2 supports the understanding of such Step 1 of the method: in this figure, like the following FIGS. 3, 4 and 5, the fluid pressure value is indicated on the ordinate axis, while the time is represented on the abscissa axis.

As mentioned, the purpose of the Step 1 is to obtain minimum pressure values P.MIN starting from the original minimum values P.RAW: hereinafter, the term “pressure values” should be understood as the pressure quantities or, indifferently, the signals deriving from the measurement of such quantities, as monitored by the pressure sensor 4 and/or stored and processed by the control unit.

In detail, in such Step 1 the pressure values P.MIN (represented in FIG. 4) are obtained by taking into account a series of conditions and events in which the heating system found itself during the time window T.analysis of the previous 20 days.

More specifically, the achievement of the pressure signal P.MIN is filtered from limit and/or anomalous conditions, deriving from certain events that affect the coherence and homogeneity of the original pressure values P.RAW.

By way of a non-limiting example, the algorithm does not take into account the oscillations of the pressure signal P.RAW, in particular the deep and sudden drops thereof, deriving from at least one of:

• • the measurement of a pressure value by the pressure sensor 4, which detects an incorrect value equal to 0 bars,
  • the maintenance of the system by a technician, who again empties and refills the circuit with fluid.

Likewise, in order to obtain the pressure signal P.MIN, the original pressure signal P.RAW is purified from pressure increases caused by:

• • the increase in the pressure value due to the heat demand, a condition in which there is typically an increase in the pressure of approximately 0.5 bars,
  • the topping up operation from the fluid inside the hydraulic circuit, by a technician or directly by the user.

FIG. 2 shows the pressure signal P.MIN obtained following such Step 1 of the method, for the identification whereof the sudden pressure drop P.RAWO of the signal P.RAW at day dated 21 August is not taken into consideration, precisely because it represents an emptying and subsequent filling of the circuit by a technician.

Step 2: identification of the ramps of the minimum values of the fluid pressure.

Once Step 1 of the method is concluded and the pressure values P.MIN are obtained, the algorithm proceeds with the implementation of Step 2, in which the identification of the ramps of such minimum pressure values of P.MIN takes place.

For the purposes of the method according to the invention, “ramp” should be understood as a portion of the pressure signal P.MIN comprised between two determined events or conditions; or, in the absence of determined events or conditions, the ramp corresponds to the entire portion of the pressure signal P.MIN identified in Step 1.

In both cases, such ramp is always identified within the same observation time window T.analysis, i.e. in the same period of the previous 20 days that was analysed by the algorithm for the execution of Step 1.

The aforementioned events or conditions comprise at least one of: increase in the minimum pressure value P.MIN caused by a top-up of the fluid in the hydraulic circuit, such as to raise this value by a pre-established threshold (for example 0.4 bars), blocking error of the boiler 1, in the event that the pressure sensor 4 detects a pressure value lower than a pre-established threshold (for example 0.3 bars);

absence of the pressure signal for a pre-established duration (for example at least 6 hours), caused by the switching off of the boiler 1.

The graph in FIG. 3 shows the identification of three distinct ramps, delimited by the occurrence of two events A and B, respectively consisting of an increase in pressure P.MIN caused by a top-up (event A) and a blocking error of the boiler 1 (event B).

In such FIG. 3, three distinct ramps of the minimum pressure values P.MIN of the fluid can therefore be identified: a ramp preceding the dashed line representing the occurrence of the event A; a second ramp between such event A and the dashed line representing the event B; a third ramp following such event B.

Step 3: calculation of a plurality of characteristics relating to the trend of the fluid pressure starting from the ramps of the minimum pressure values.

The Step 3 of the method provides for the preparation of characteristic data (hereinafter “characteristics”) to be then provided to the classification model SVM in the subsequent Step 4.

The algorithm processes a series of characteristics on the basis of the ramps identified in the previous Step 2; in particular, the fluid pressure signal outlined by the ramps is analyzed according to the following routines:

• • description of the behavior of the pressure of the last ramp identified in the Step 2 (or of the single ramp, in case a single ramp has been identified within the observation time window T.analysis due to the absence of the events or conditions listed in the Step 2): with such routine, the algorithm extracts various characteristics such as, by way of a non-limiting example:
    • the descent speed of the pressure value P.MIN of the last identified ramp,
    • the minimum pressure value P.MIN of said last ramp,
    • a binary value (0 or 1) if, in the last week, for at least 50% of the time the pressure value P.MIN remained stable on values lower than the standard suggested by the installer: in such case the binary value is equal to 1;
  • analysis of the validity of the data processed in Step 2: with such routine, the algorithm extracts various characteristics such as, by way of a non-limiting example:
    • percentage of the minimum pressure values P.MIN within the 20 days observation time window T.analysis,
    • percentage of such values below a certain threshold,
    • number of the ramps identified in the window T.analysis;
    • percentage of valid pressure values P.MIN of the last ramp identified in the window T.analysis (i.e. the percentage of the pressure values P.MIN remaining after the removal of the original pressure values P.RAWderiving from the limit and/or anomalous conditions mentioned in Step 1);
  • count of the ramps identified in Step 2;
  • description of the pressure signal outlined by the ramps, within a time interval equal to a sub-multiple of the observation time window T.analysis, by dividing the time window T.analysis into N intervals of equal duration: with such routine, the algorithm extracts various characteristics from each time interval (for example, lasting 2 days) such as, by way of a non-limiting example:
    • the descent speed of the pressure value P.MIN,
    • the pressure value P.MIN at the beginning of the interval,
    • a binary value (0 or 1) if the last ramp identified in the specific interval
    • coincides with the last ramp of the entire time window T.analysis: in such case the binary value is equal to 1,
    • the value of the variance of the pressure signal P.MIN,
    • the average of the pressure signal P.MIN,
    • the difference between the maximum value and the minimum value of the pressure signal P.MIN,
    • the minimum value of the pressure signal P.MIN,
    • the maximum value of the pressure signal P.MAX;
  • analysis of the overall trend of the signal within the entire observation time window T.analysis: with such routine, the algorithm extracts various characteristics such as, by way of a non-limiting example:
    • the linear regression of the descent speeds of the pressure value P.MIN of the N intervals of equal duration into which the observation window T.analysis has been divided,
    • the Pearson correlation coefficient of such linear regression.

The features extracted with the above listed routines are those considered most suitable for the correct execution of the subsequent Step 4 of the method; however, the method of the invention can also work with the extraction of a smaller number of features, while maintaining the execution of the routines according to the sequence above.

Step 4: classification by the machine learning model.

The features calculated in Step 3 are inserted into a machine learning model, based on a classifier SVM (the so-called support-vector machine), previously trained and able to return a prediction, within a future time window T.prox equal to the next 10 days, of the more or less high probability that the boiler 1 will block due to an excessive drop in the fluid pressure value.

More precisely, such classifier produces as output one of the three classes on which it has been trained, each of such three classes being representative of the probability that the boiler 1 will be, in a future time window T.prox (preferably 10 days), in a blocking condition due to a low pressure value of the circulating fluid.

The model is trained with the historical data of a multitude of heating systems, to which the above mentioned three output classes have been assigned.

Such preliminary training step, similarly to the classifying and non-regressor machine learning models, is preferably carried out once, before the first execution of the method of the invention; however, nothing prevents further training steps from being repeated subsequently, even through self-learning methods.

That is, the training of the machine learning model can continue even during the normal operation of the method, in accordance with well-known artificial intelligence algorithms that provide, precisely, a progressive learning with incremental training stages through a sort of data back-propagation, which allows the model to adapt over time through the training and storing of additional data in the control unit, through input provided by a technician or directly by the user during the normal use of the boiler 1.

The three output classes of the classification model are:

• • green class: exemplifying a high probability that the boiler 1, in the next 10 days, will not run into any blocking problem caused by an excessive drop in the pressure value of the fluid circulating in the system;
  • yellow class: representative of the fact that the boiler 1, in the next 10 days, will have a rather low-pressure value, lower than the recommended value; however, the boiler 1 may also continue to work in this condition still for a long time, although a minimal variation in the circuit that may cause a sudden drop in pressure below the threshold value and cause the boiler 1 to block cannot be excluded;
  • red class: exemplifying a high probability that the boiler 1 will incur a blocking error within the next 10 days due to an excessively low-pressure value.

In the event that Step 4 produces the yellow or red class as an output, the method according to the invention provides that such classes are also associated with a notification to the technical assistance centre and/or the user.

The notification can be visual and/or acoustic, for example shown in the HMI interface of the heating system and/or of the possible external devices connected thereto and/or through written communications via email, SMS, or the like, to the same external devices, that may be checked directly by the user, or to remote servers that may be checked by the assistance centre.

FIGS. 4 and 5 graphically show the results of the method according to the invention at the end of Step 4, in the event that the machine learning classification model produces the yellow class and the red class as output.

FIG. 4 (yellow class) shows how the expected trend of the pressure signal P.MIN is decreasing, even without reaching the limit threshold P.BLOCK (for example equal to 0.5 bars) below which the boiler 1 goes into the blocking condition.

Such result produces a yellow class classification, with emission of the notification W1 which suggests topping up the fluid in the system circuit in order to restore a higher pressure value.

In FIG. 5 (red class), the trend of the pressure signal P.MIN decreases beyond the limit threshold P.BLOCK, causing the machine learning model to insert the result into the red class, with the issuing of the notification W2 which, being representative of a high probability that in the next 10 days the boiler 1 will block, indicates the need to top up the fluid in the system circuit.

The above-described control method is particularly effective in preventing blockage situations of the boiler 1 and of the prior art heating systems.

However, it should be emphasized that the algorithm underlying the described method uses a machine learning classification model that is different from a regressor model, i.e. it does not return the exact moment in time in which the heating system will reach a critical pressure value of the fluid circulating in the system.

In other words, the control method described herein, through the classification into the three-classes mentioned above, is able to discriminate the more or less high probability that the critical value of the fluid pressure may be reached in a future time window T.prox (for example equal to the next 10 days): however, the uncertainty remains on the precise time moment of such achievement, which may occur at any time within such time window T.prox.

From the above description it is clear that the method of the invention can achieve the underlying objects, in particular that of indicating in advance the need for an intervention to prevent the heating system from incurring a block in operation, through the use of components and means already natively present in the system.

Claims

1. A Method for controlling the pressure of the heat transfer fluid circulating in a heating system, said system comprising at least: a thermo-sanitary appliance, in particular a gas boiler;a primary circuit having a primary exchanger for the heating of the water intended to supply the heating bodies for the space heating;a secondary circuit having a secondary exchanger for the heating of the domestic water intended for the users;a pressure sensor adapted to measure the value of the pressure of said heat transfer fluid;a control unit adapted to receive said pressure value of the fluid;such method comprising at least the following steps in sequence: Step 1: extrapolating the minimum values (P.MIN) of the pressure of the fluid from the original minimum values (P.RAW), within a time observation window (T.analysis) corresponding to a given number of previous days;Step 2: identifying one or more ramps of the minimum values (P.MIN) of the pressure of the fluid, within said time observation window (T.analysis);characterized in that such method further comprises in sequence the following steps: Step 3: calculating a plurality of characteristics relating to the trend of the pressure of the fluid on the basis of said one or more ramps of the minimum values (P.MIN) identified in Step 2;Step 4: providing said characteristics calculated in Step 3 to a machine learning classification model, previously trained with the history data of a multitude of heating systems, said model producing a three class classification in outlet, each of which is representative of the more or less high probability that in a future time window (T.prox) said thermo-sanitary appliance (1) will block due to an excessively low value of the pressure of said fluid, said three classes comprising: green class: representative of a high probability that said thermo-sanitary appliance will not incur, in said future time window (T.prox), into any blockage problem caused by an excessive drop in the pressure value of the fluid;yellow class: representative of the fact that said thermo-sanitary appliance (1) will present, in said future time window (T.prox), a pressure value of the fluid lower than the recommended value, but not an obstacle to the continued operation of said appliance;red class: representative of a high probability that said thermo-sanitary appliance (1) will incur, in said future time window (T.prox), into any blockage problem caused by an excessive drop in the pressure value of the fluid.

2. Control method according to claim 1, wherein: following said classification into the yellow or red classes, the further step of the output of a notification to the technical assistance centre and/or the user is provided, said notification comprising: a warning with the suggestion of proceeding with a top up of the fluid in the circuit of the system in case of a yellow class classification;a warning with the need of proceeding with a top up of the fluid in the circuit of the system in case of a red class classification.

3. Method according to claim 1, wherein in said Step 1 of extrapolation of the minimum values (P.MIN) of the pressure of the fluid from the original minimum values (P.RAW), the latter are purified from limit and/or abnormal conditions, such as for example at least one among: the measurement of a pressure value by the pressure sensor, which detects an incorrect value equal to 0 bars;the maintenance of the system by a technician, who again empties and refills the circuit with fluid,the increase in the pressure value due to the heat demand, a condition in which there is typically an increase in the pressure of approximately 0.5 bars;the topping up operation from the fluid inside the hydraulic circuit, by a technician or directly by the user.

4. Method according to claim 1, wherein in said identification Step 2 of the one or more ramps of the minimum values (P.MIN) of the pressure of the fluid, said ramps are identified between two events or conditions such as at least one among: increase in the minimum pressure value (P.MIN) caused by a top up of the fluid in the hydraulic circuit, such as to raise this value by a preset threshold (for example 0.4 bars);block error of the thermo-sanitary appliance, in case that the pressure sensor detects a pressure value lower than a preset threshold (for example 0.3 bars);absence of the pressure signal for a preset duration (for example at least 6 h), caused by the switching off of the thermo-sanitary appliance.

5. Method according to claim 1, wherein in said calculation Step 3 of a plurality of characteristics, said ramps of the minimum values (P.MIN) of the pressure of the fluid identified in the Step 2 are processed according to the following routines: description of the behaviour of the pressure of the last ramp;analysis of the validity of the data processed in said Step 2;counting of the ramps identified in said Step 2;description of the signal of the pressures outlined by the ramps, within a time interval equal to a submultiple of the observation time window (T.analysis), each of said submultiples having equal duration;analysis of the overall trend of the signal within the entire observation time window (T.analysis).

6. Method according to claim 5, wherein through said description routine of the behavior of the pressure of the last ramp, said method calculates a plurality of characteristics, among which one or more of the following: the descent speed of the minimum pressure value (P.MIN) of the last ramp identified;the minimum pressure value (P.MIN) of said last ramp;a binary value (0 or 1) if, in the last week, for at least the 50% of the time, the pressure value (P.MIN) remained stable at values lower than the standard suggested by the installer: in such case the binary value is equal to 1.

7. Method according to claim 5, wherein through said analysis routine of the validity of the data processed in said Step 2, said method calculates a plurality of characteristics, among which one or more of the following: percentage of the minimum pressure values (P.MIN) within the observation time window (T.analysis);percentage of such values below a certain threshold;number of ramps identified in the observation time window (T.analysis);percentage of the valid pressure values (P.MIN) of the last ramp identified in the observation time window (T.analysis).

8. Method according to claim 5, wherein through said description routine of the pressures outlined by the ramps, within a time window equal to a submultiple of the observation time window (T.analysis), said method calculates a plurality of characteristics, among which one or more of the following: the descent speed of the pressure value (P.MIN);the pressure value (P.MIN) at the beginning of the interval;a binary value (0 or 1) if the last ramp identified in the specific interval coincides with the last ramp of the entire observation time window (T.analysis): in such case the binary value is equal to 1;the value of the variance of the pressure signal (P.MIN);the average of the pressure signal (P.MIN);the difference between the maximum value and the minimum value of the pressure signal (P.MIN);the minimum value of the pressure signal (P.MIN);the maximum value of the pressure signal (P.MAX).

9. Method according to claim 5, wherein through said analysis routine of the overall trend of the signal within the observation time window (T.analysis), said method calculates a plurality of characteristics, among which one or more of the following: the linear regression of the descent speeds of the pressure value (P.MIN) of the N intervals of equal duration into which the observation window (T.analysis) has been divided;the Pearson correlation coefficient of such linear regression.

10. Method according to claim 1, wherein said observation time window (T.analysis) is equal to 20 days.

11. Method according to the claim 10, wherein said submultiple of the observation time window (T.analysis) lasts 2 days.

12. Method according to claim 1, wherein said future time window (T.prox) lasts 10 days.

13. A heating system comprising at least: a thermo-sanitary appliance, in particular a gas boiler or an indoor heating heat pump or a hybrid boiler;a primary circuit having a primary exchanger for the heating of water intended to supply heating bodies for space heating;a secondary circuit having a secondary exchanger for the heating of domestic water intended for users;a pressure sensor adapted to measure the value of the pressure of said heat transfer fluid;a control unit adapted to receive said pressure value of the fluid, wherein said control unit is adapted to exchange information with said pressure sensor and to carry out the steps of the method according to claim 1.