Method for estimating the condition of a motor vehicle, and associated system

Abstract

A method for estimating the state of a first motor vehicle (100), characterized in that it comprises the following steps, when the first vehicle (100) is in motion: Measurement, by said first vehicle (100), of a first plurality of physical quantities (the physical quantities varying when the first vehicle is in motion), the first plurality of physical quantities comprising a first set of physical quantities and a second set of physical quantities; Transmission, by the first vehicle (100), of the first set of physical quantities, via a mobile communication network (500), to a server (400) comprising a machine learning model (410); Determination, by the machine learning model (410), of a first physical datum, from the first set of physical quantities. Figure for the abstract: Figure 2

Description

Title of the invention: Method for estimating the condition of a motor vehicle, and associated system

[0001] The invention relates to preventive and predictive maintenance of motor vehicles.

[0002] Today, a great deal of information is collected in a motor vehicle. Much of this information, because it is representative of the physical condition of the motor vehicle or the physical stresses experienced by the motor vehicle, can serve as a basis for predictive maintenance of the motor vehicle.

[0003] This information, because it is representative of the constraints experienced by the parts of motor vehicles, can also be used to dimension motor vehicles during their design.

[0004] However, sending this information, which constitutes a large volume, to a server for analysis can be costly. Sending this information may also not comply with personal data privacy regulations.

[0005] To remedy this drawback, the invention relates to a method for estimating the condition of a first motor vehicle, characterized in that it comprises the following steps, when the first vehicle is in motion (for example, on a road): - Measurement, by (in other words: in) said first vehicle, of a first plurality of physical quantities (the physical quantities varying when the first vehicle is moving), the first plurality of physical quantities comprising a first part of physical quantities and a second part of physical quantities (NB: preferably, the first part and the second part form disjoint sets), - Transmission, of the first part of physical quantities, via a mobile communication network, by the first vehicle, to a server comprising a machine learning model (in other words: an algorithm trained by (automatic) learning), - Determination, by the machine learning model (of the server), of a first physical data, for example, representative of the physical state of the first motor vehicle or of the physical constraints suffered by the first motor vehicle (for example, fatigue or wear of the motor vehicle) resulting from the circulation of the first vehicle, from the first part of physical quantities (in input to the learning model).

[0006] Thus, the machine learning model, by determining the first data from the first part of physical quantities, makes it possible to avoid the emission of data by the first vehicle, for example from the second part of physical quantities.

[0007] According to one embodiment, the first data point is determinable (in other words: can be determined) by a deterministic algorithm (a deterministic algorithm is one which, given a particular input, will always produce the same output, the underlying state machine always passing through the same sequence of states) from (as input to the deterministic algorithm) (one or more values) of the second set of physical quantities or, from (one or more values) of the second set of physical quantities and (a value) of the first set of physical quantities. In other words, according to one embodiment, the first data point is determinable (in other words: can be determined) deterministically from the second set of physical quantities and possibly from the first set of physical quantities.

[0008] The deterministic algorithm can be, for example, a (simple) arithmetic operation or an algorithm.

[0009] Alternatively, the first data point is one of the physical quantities of the second part.

[0010] Alternatively, the first data cannot be obtained deterministically either from the second part of physical quantities, or from the second part of physical quantities and the first part of physical quantities.

[0011] For example, (according to example 1) the first data is NC3 (NC to the power of 3) and NC7 (NC to the power of 7) where NC is the product of the (or an) (instantaneous) speed of the (propulsion) engine of the first vehicle by the (or an) torque of the (propulsion) engine of the first vehicle.

[0012] NC3 is used, for example, to dimension the torque transfer durability of the transmission. NC7 is used to determine the endurance limit of the transmission.

[0013] For example, (according to example 1 bis) the first part of physical quantities includes the (or a) engine (propulsion) speed of the first vehicle and the (or a) speed of movement of the first vehicle when it is travelling (on a road).

[0014] The first part of physical quantities may include (also or alternatively) the following physical quantities: - An (instantaneous) temperature of the engine oil (of the propulsion engine of the first vehicle), - An ambient (instantaneous) temperature outside the first vehicle, - Information indicating that the anti-lock braking system (ABS) of the first vehicle intervenes (to limit wheel lock-up or wheel slippage) or that the electronic stability control system (ESP) of the first vehicle intervenes (to correct the trajectory by acting on the braking system as well as on the engine torque of the propulsion engine of the first vehicle), - A geographical altitude of the first vehicle.

[0015] For example, (according to example 2) the first data is a state of charge of a battery of a first vehicle with a thermal propulsion motor.

[0016] For example, (according to example 2 bis) the first part of physical quantities includes a number of starts and / or stops of the first vehicle and a mileage of the first vehicle.

[0017] The first part of the physical quantity may also include, or alternatively include, other quantities.

[0018] According to one embodiment, the learning model is a decision tree forest (also called a random forest) or a gradient amplification. Other models may also be used.

[0019] For example, the estimation process includes a step of training the learning model from training data labeled with outcome data.

[0020] According to one embodiment, the process further comprises the following steps (preferably repeated for several different third vehicles and with different parameters): - Circulation of a second vehicle, - Measurement by (in other words: in) said second vehicle, of a second plurality of physical quantities (the physical quantities varying when the second vehicle is moving), - The result data being obtained from the second plurality of physical quantities (for example the result data are identical to the physical quantities).

[0021] The measurement, by the first vehicle or the second vehicle, is for example carried out by a sensor on the vehicle.

[0022] According to one embodiment, the process further comprises the following steps (preferably repeated for several different third vehicles and with different parameters): - Calculation of a second plurality of physical quantities by simulating the movement of a third (fictitious) vehicle according to predetermined parameters, - The result data being obtained from the second plurality of physical quantities (for example the result data are identical to the physical quantities).

[0023] For example, the predetermined parameters are (instantaneous) traffic speeds and (instantaneous) torques of the (or one) propulsion motor of the third vehicle.

[0024] For example, the second plurality of physical quantities may include the examples given above for the first part of physical quantities.

[0025] According to one embodiment, the training data are obtained from the second plurality of physical quantities.

[0026] Alternatively, the training data are predetermined values.

[0027] For example (in the case of example 1 and Ibis): - The second plurality of physical quantities includes the (or a) engine speed (of propulsion) of the second or third vehicle, the (or a) engine torque (of propulsion) of the second or third vehicle and the (or a) speed of movement of the second or third vehicle. - The training data includes the engine (or a) speed of the second or third vehicle, and the travel speed of the second or third vehicle. - The result data includes NC3 (NC to the power of 3) and NC7 (NC to the power of 3),

[0028] Alternatively (in the case of examples 2 and 2bis): - The second plurality of physical quantities includes the state of charge of a battery in the second or third vehicle, the number of starts and / or stops of the second or third vehicle, and the mileage of the second or third vehicle. - The training data includes the number of starts and / or stops of the second or third vehicle and the mileage of the second or third vehicle. - The results data include a state of charge of a battery of the second or third vehicle with internal combustion propulsion engine.

[0029] The invention also relates to a system comprising a vehicle and a server configured to implement the steps of the process according to the invention.

[0030] Thus, according to one embodiment, the invention relates to a system for estimating the condition of a first motor vehicle, characterized in that it comprises: - A server including a machine learning model, - A first vehicle configured to implement the following steps: - Measurement of a first plurality of physical quantities (the physical quantities varying when the vehicle is moving), the first plurality of physical quantities comprising a first part of physical quantities and a second part of physical quantities (NB: for example, the first part and the second part form disjoint sets), - Transmission, via a mobile communication network, to the server, of the first part of physical quantities, for example, representative of the physical state of the motor vehicle or of the physical constraints suffered by the motor vehicle (for example, fatigue or wear of the motor vehicle), from the first part of physical quantities (in input to the learning model). - The machine learning model, being configured to determine a first piece of data, from the first part of the physical quantity.

[0031] The characteristics and advantages of the system according to the invention are identical to those of the method according to the invention above, therefore they are not repeated here.

[0032] An element (such as a vehicle, server, or other element) is understood to be "configured to" perform or implement a step or operation by the fact that the element includes means to (in other words, "is designed to" or "is adapted to") perform the step or operation. These are preferably electronic means, for example, a computer program, data in memory, and / or specialized electronic circuits.

[0033] When a step or operation is performed (in other words: implemented) by such an element, this generally implies that the element has means for (in other words, "is designed for" or "is adapted for" or "is configured for") performing the step or operation. These may also include, for example, electronic means, such as a computer program, data in memory, and / or specialized electronic circuits.

[0034] Other features and advantages of the present invention will become more apparent from the following detailed description, which includes embodiments of the invention given by way of non-limiting examples and illustrated by the accompanying drawings, in which.

[0035] [Fig.1] represents a system according to an embodiment of the invention.

[0036] [Fig.2] represents the implementation of the method according to the invention, according to an example of realization, by the system of [Fig.l].

[0037] Detailed description of an example embodiment of the invention

[0038] With reference to Figures 1 to 2, in step S10, a motor vehicle 200 is traveling on a road 300. The motor vehicle 200 includes a propulsion engine 210

[0039]

[0040]

[0041]

[0042]

[0043]

[0044]

[0045]

[0046]

[0047]

[0048]

[0049]

[0050]

[0051] of a motor vehicle 200 and a battery 220. The vehicle 100 and the server 400 constitute a system for estimating the condition of the vehicle 100. At stage S20, vehicle 200 measures the following data, while the second vehicle 200 is in motion: - the operating speed of its 210 propulsion engine, - the torque of its 210 propulsion engine, and - a travel speed of the second vehicle of 200 on road 300. The measurement is, for example, carried out by a sensor on vehicle 200. Alternatively, this data can be obtained by simulation, for example by the 400 server, of a circulation of a fictitious vehicle at (instantaneous) circulation speeds and (instantaneous) torques of the propulsion motor of the fictitious vehicle (into the simulation). At stage S30, the vehicle 200 transmits, for example using a vehicle communication module, this data to the server 400, for example, via the mobile telecommunications network 500 or by another type of means of communication. At step S40, the server 400 calculates NC, the product of the speed and torque thus received by the server 400, and NC3 (NC to the power of 3) and NC7 (NC to the power of 7). At step S50, the 410 learning model is trained in a supervised manner with input the engine speed and the speed labeled by NC3 and NC7. Steps S20 to S50 can be repeated to train the learning model 410. In one embodiment, the learning model is a forest of decision trees or a gradient amplification. The training of the learning model is carried out in a manner known to those skilled in the art. At stage S60, a motor vehicle 100 travels on a road 300. The motor vehicle 100 includes a propulsion motor 110 and a battery 120. At step S70, vehicle 100 measures the following data while vehicle 100 is in motion: - the operating speed of its propulsion engine 110, and - a vehicle travel speed of 100 on road 300. At stage S80, vehicle 100 transmits this data, via a mobile communication network 500, to the server 400 comprising a machine learning model 410. At step S90, the 410 machine learning model determines NC3 and NC7 from the engine speed and velocity. Alternatively, in steps S20 to S90: The engine is replaced by the number of vehicle starts (100 and 200 and the virtual vehicle), Speed ​​is replaced by the vehicle's mileage (100 and 200 and the virtual vehicle), NC3 and NC are replaced by a battery charge state of the vehicle (100 and 200 and the virtual vehicle).

Claims

Claims

1. Method for estimating a state of a first motor vehicle (100), characterized in that it comprises the following steps, when the first vehicle (100) is traveling: - Measurement (S70), by said first vehicle (100), of a first plurality of physical quantities, the first plurality of physical quantities comprising a first part of physical quantities and a second part of physical quantities, - Transmission (S80), of the first part of physical quantities, via a mobile communication network (500), by the first vehicle (100), to a server (400) comprising a machine learning model (410), - Determination (S90), by the machine learning model (410), of a first physical data item, from the first part of physical quantities, the first data item being NC3 and NC7,where NC is the product of a speed of the engine (110) of the first vehicle (110) by the torque of the engine (110) of the first vehicle (100).,

2. Estimation method according to the preceding claim in which the first part of physical quantities comprises a speed of the engine (110) of the first vehicle (110) and a speed of movement of the first vehicle (100).

3. An estimation method according to any preceding claim, the estimation method comprising a step of training (S50) the learning model (410) from training data labeled by result data.

4. Estimation method according to claim 3 further comprising the following steps: - Circulation (S 10) of a second vehicle (200), - Measurement (S20) by said second vehicle (200), of a second plurality of physical quantities, The result data being obtained from the second plurality of physical quantities.

5. Estimation method according to claim 3 further comprising the following steps: - Calculation of a second plurality of physical quantities by simulating the circulation of a third vehicle according to predetermined parameters, the result data being obtained from the second plurality of physical quantities.

6. Estimation method according to any one of claims 4 or 5 in which the training data is obtained from the second plurality of physical quantities.

7. System for estimating (1000) the state of a first vehicle (100) automobile (100), characterized in that it comprises: - A server (400) comprising a machine learning model (410), - A first vehicle (100) configured to implement the following steps: - Measurement (S70) of a first plurality of physical quantities, the first plurality of physical quantities comprising a first part of physical quantities and a second part of physical quantities, - Transmission (S80), via a mobile communication network (500), to the server (400) of the first part of physical quantities from the first part of physical quantities.- The machine learning model (410), being configured to determine a first physical data, from the first part of physical quantity, the first data being NC3 and NC7, where NC is the product of a speed of the engine (110) of the first vehicle (110) by the torque of the engine (110) of the first vehicle (100).

Images