Method and corresponding system for predicting expected deceleration of at least one vehicle

Abstract

Method and corresponding system for predicting expected deceleration of at least one vehicle: A computer-implemented method for predicting expected deceleration of at least one vehicle, in particular at least one rail vehicle, the method comprising the steps of: a) providing at least one braking data related to the performance of the braking system of the at least one vehicle, at least one environmental data related to environmental conditions of a route along which the vehicle will travel, and at least one vehicle data related to the structure of the at least one vehicle as inputs to a neural network; and b) predicting, via the neural network, an expected deceleration value of the at least one vehicle based on the at least one braking data, the at least one environmental data, and the at least one vehicle data.

Description

[Technical Field]

[0001] The present invention relates generally to the field of vehicles, and more particularly, the present invention relates to a computer-implemented method and corresponding system for predicting the expected deceleration of at least one vehicle. [Background technology]

[0002] The prior art will be described below with particular reference to the rail vehicle industry, but what is described below can be applied to vehicles in other industries as well where applicable.

[0003] The railway industry is pursuing new concepts for the management of rail vehicles on its network. These new management concepts aim to increase track capacity, rail vehicle reliability, and resistance to changing environmental conditions.

[0004] Railroad vehicle braking performance can be impaired by adverse environmental conditions, the effects of which can unfortunately only be partially mitigated by existing braking systems. Summary of the Invention [Means for solving the problem]

[0005] Therefore, one object of the present invention is to provide a solution that allows for the prediction of vehicle performance, in particular vehicle deceleration performance, which can be used to improve the performance of the vehicle's braking system even under the worst environmental conditions. For example, the prediction of vehicle deceleration performance can be used for the following applications: - Reducing the distance of moving block functions ("moving block trains") and thereby increasing track capacity; -Easying the integration of autonomous vehicles into operations by providing information that in prior art was manually provided by the driver; - To be able to provide infrastructure managers with information relating to the deceleration conditions achievable on the route / track.

[0006] The above and other objects and advantages are achieved according to one aspect of the present invention by a computer-implemented method for predicting the expected deceleration of at least one vehicle having the features defined in claim 1, and according to a further aspect by a system for predicting the expected deceleration of at least one vehicle having the features defined in claim 10. Preferred embodiments of the present invention are defined in the dependent claims, the content of which is to be understood as an integral part of this description. [Brief explanation of the drawings]

[0007] The functional and structural features of some preferred embodiments of the system for predicting the expected deceleration of at least one vehicle and the computer-implemented method for predicting the expected deceleration of at least one vehicle according to the present invention will now be described with reference to the accompanying drawings, in which:

[0008] [Figure 1] FIG. 2 illustrates an exemplary neural network that may be used in a computer-implemented method for predicting the expected deceleration of at least one vehicle according to the present invention. [Figure 2] FIG. 1 illustrates two exemplary activation functions for a neural network. [Figure 3] Figure 1 shows the state of neurons in a "feed-forward" architecture. [Figure 4] 1 illustrates an exemplary vehicle including a computer configured to perform a method for predicting expected deceleration in accordance with an embodiment of the present invention; DETAILED DESCRIPTION OF THE INVENTION

[0009] Before describing several embodiments of the present invention in detail, it should be made clear that the present invention is not limited in its application to the details of design and arrangement of components set forth in the following description or illustrated in the drawings. The present invention is capable of other embodiments and may actually be implemented or constructed in different ways. It is also to be understood that the phraseology and terminology used is for convenience of description and is not to be construed as limiting. The use of the terms "including" and "comprising" and variations thereof is intended to encompass the elements set forth below and equivalents thereof, as well as additional elements and equivalents thereof.

[0010] By way of example, and with reference to FIG. 4, a first embodiment of a computer-implemented method for predicting the expected deceleration of at least one vehicle V, in particular at least one rail vehicle, will now be described.

[0011] The method comprises: Step a): providing at least one braking data 102 relating to the performance of the braking system of at least one vehicle, at least one environmental data 104 relating to the environmental conditions of a route travelled by the vehicle, and at least one vehicle data 106 relating to the structure of at least one vehicle V as inputs to a neural network 100; Step b): predicting / determining an expected deceleration value 108 of at least one vehicle V based on the at least one brake data 102, the at least one environmental data 104 and the at least one vehicle data 106 via the neural network 100.

[0012] The data 102, 104, 106 provided to the neural network in step a) were selected because they may affect the outcome of the vehicle slowdown.

[0013] In other words, the at least one braking data 102 may indicate the response of the vehicle's braking system, as well as a potential deterioration state of the braking system and the use of special brakes, such as electrodynamic brakes, magnetic track brakes (MTBs), sand distribution means, eddy current brakes, etc.

[0014] The at least one environmental data 104 may be indicative of environmental conditions in which the vehicle travels.

[0015] The at least one vehicle data 106 may be indicative of a characteristic of the vehicle that may affect the vehicle's deceleration and therefore braking distance.

[0016] Preferably, the neural network 100 has a feed-forward structure.

[0017] A feedforward neural network is a structure in which the layers do not form a ring, but rather information flows from the input layer to the hidden layer to the output layer. This structure is relatively simple and requires limited computational power, making it particularly suitable for implementation in real-time systems such as the control units of braking systems.

[0018] The following is an exemplary definition of a neuron-based artificial intelligence structure with reference to the exemplary neural network shown in FIG.

[0019] The input of neuron a is a linear function, b is the bias, w is the weight, p is the data input, and h is the output of the neuron. The output is the same as the activation function.

number

[0020] For example, the activation function can be of two types: sigmoid and rectifier.

[0021] Each has its own unique advantages: -Sigmoid activation functions allow for more complex data structures and ensure that the output is regular and avoids sudden changes in values. -Rectifier activation function (Relu) requires less computational effort.

[0022] Figure 2 shows two graphs. The left graph shows a rectifier activation function. The right graph shows a sigmoid activation function.

[0023] Referring to Figure 3, the neurons in a feed-forward structure can be represented by an input layer, a hidden layer, and an output layer.

[0024] Preferably, the computer-implemented method for predicting the expected deceleration of at least one vehicle, in particular at least one railway vehicle, may comprise, prior to steps a) and b), performing training of a neural network.

[0025] In such cases, training the neural network may include: providing as inputs to the neural network at least one brake data, at least one environmental data, at least one vehicle data, and an expected deceleration value that is a function of the at least one brake data, the at least one environmental data, and the at least one vehicle data; - determining a value of at least one parameter of said neural network according to the received at least one braking data, at least one environmental data, at least one vehicle data and an expected deceleration value.

[0026] In this manner, the predicted deceleration value that is a function of the at least one brake data, the at least one environmental data, and the at least one vehicle data represents a known predicted deceleration value (e.g., a measured predicted deceleration value that may be derived from past experimental measurements) obtained using the at least one brake data, the at least one environmental data, and the at least one vehicle data.

[0027] Preferably, the training is based on the "backpropagation" algorithm.

[0028] In other words, a feed-forward neural network can be trained with a backpropagation algorithm, and the output (determination of vehicle deceleration) checked for consistency. The weights and biases (parameters) are frozen when training is complete, once the output is accurate.

[0029] Once vehicle deceleration is predicted, this value may be transmitted to a control means located on the vehicle or to a remote infrastructure manager.

[0030] Preferably, the at least one braking data 102 relating to the performance of at least one vehicle's braking system may include at least one of the following types of data: at least one deceleration data indicating a deceleration value of at least one wheel W or at least one axle of at least one vehicle, - at least one piece of data indicating the number of wheels W in a skidding phase of at least one vehicle, - at least one piece of data indicative of deceleration of at least one vehicle during an initial braking phase; at least one piece of data indicative of the actuation speed of at least one braking means of the braking system; at least one piece of data indicative of a skid velocity of at least one wheel or at least one axle of at least one vehicle, - at least one piece of data indicative of a steady-state value of at least one braking means of the braking system, - at least one data indicating activation of a sanding means of at least one vehicle, - at least one data indicating activation of at least one vehicle magnetic brake pad MTB; - at least one data indicating the activation of a system / function arranged to compensate for deviations from the expected deceleration value; - at least one piece of data indicating that there is a malfunction of the braking means of the braking system; -At least one data indicative of a number of activations over time of an exhaust valve associated with at least one braking means of a braking system of at least one vehicle.

[0031] The steady-state value of at least one braking means of a braking system may be understood as the value at which the required braking force is reached after a transient (for example, in a pneumatic braking system, the transient corresponds to the time it takes to bring the brake cylinder to nominal pressure).

[0032] Preferably, said at least one environmental data 104 relating to the environmental conditions of the route travelled by the vehicle may comprise at least one of the following types of data: - at least one image data or at least one video data of the route (e.g., one captured image or video of the route), - at least one temperature data showing the temperature on the route, - at least one rainfall data indicating that there is rainfall on the route; - at least one moisture data showing the moisture level on the route; - at least one adhesion data indicating the adhesion level on the route; - At least one route data showing the outline of the route.

[0033] Preferably, the at least one vehicle data 106 relating to the configuration of the at least one vehicle may include at least one of the following types of data: - at least one nominal deceleration data indicating a nominal deceleration value of at least one vehicle; at least one wheel or axle data indicating the number of wheels or axles of at least one vehicle; at least one sand distribution device data indicating the number of sand distribution devices of at least one vehicle; - at least one data of magnetic pads, indicating the number of MTB magnetic brake pads for at least one vehicle; - at least one anti-skid system data showing that the anti-skid system WSP acts on each bogie or each axle of the vehicle; -At least one data item regarding the presence of deceleration compensation indicating that the vehicle has at least one vehicle deceleration compensation system / function.

[0034] Obviously, the various types of brake data 102 relating to the performance of the brake system, the various types of environmental data relating to the environmental conditions of the route traveled by the vehicle, and the various types of vehicle data 106 relating to the structure of at least one vehicle may be combined in any manner, some possible combinations being given below as examples:

[0035] Preferably, in an example embodiment, the at least one brake data 102 relating to the performance of the braking system of the at least one vehicle may include at least one data indicative of the skid speed of at least one wheel or axle of the at least one vehicle, the at least one environmental data 104 relating to the environmental conditions of the path traveled by the vehicles may include at least one adhesion data indicative of the adhesion level along the path, and the at least one vehicle data 106 relating to the structure of the at least one vehicle may include at least one wheel or axle data indicative of the number of wheels or axles of the at least one vehicle. This example is primarily aimed at monitoring adhesion levels and their potential adverse effect on deceleration of the various wheels of the vehicle.

[0036] In a further embodiment, the at least one brake data 102 relating to the performance of the at least one vehicle's brake system may preferably include at least one data indicating the number of wheels in a skid phase of the at least one vehicle, the at least one environmental data 104 relating to the environmental conditions of the path along which the vehicle is traveling may include at least one adhesion data indicating the adhesion level along the path, and the at least one vehicle data 106 relating to the structure of the at least one vehicle may include at least one anti-skid system data indicating the operation of an anti-skid system WSP for each bogie or each axle of the vehicle. For example, the at least one vehicle data 106 relating to the structure of the at least one vehicle may further include at least one data regarding the presence of deceleration compensation, indicating that the vehicle has at least one vehicle evasion deceleration compensation system / function. This embodiment is primarily intended to monitor the adhesion level between the various wheels of the vehicle and the positive impact on deceleration that may be produced by the at least one vehicle evasion deceleration compensation system / function.

[0037] In yet another embodiment, preferably, the at least one brake data 102 relating to the performance of the brake system of the at least one vehicle may comprise at least one data indicative of a steady state value of at least one brake means of the brake system and at least one data indicative of the presence of a fault in a brake means of the brake system, the at least one environmental data 104 relating to the environmental conditions of the route travelled by the vehicles may comprise at least one of the data items from the above list, and the at least one vehicle data 106 relating to the structure of the at least one vehicle may comprise at least one of the data items from the above list. This embodiment is primarily aimed at monitoring the state of the brake system and the adverse effect on deceleration that a faulty brake means may have.

[0038] Preferably, if there are multiple types of braking data 102 relating to the performance of at least one vehicle's braking system, each type of braking data may undergo a respective data fusion process before being provided to the neural network.

[0039] For example, if there are 100 pieces of braking data including 50 pieces of data indicating the actuation speed of at least one braking means of the braking system and 50 pieces of data indicating the skid speed of at least one wheel or at least one axle of at least one vehicle, the 50 pieces of data indicating the actuation speed of at least one braking means of the braking system may undergo their own data fusion processing, and the 50 pieces of data indicating the skid speed of at least one wheel or at least one axle of the at least one vehicle may undergo their own data fusion processing before being provided to the neural network.

[0040] Preferably, when there are multiple types of environmental data 104 relating to the environmental conditions of the route traveled by the vehicle, each type of environmental data may undergo a respective data fusion process before being provided to the neural network.

[0041] For example, if there are 100 pieces of environmental data including 50 pieces of temperature data indicating the temperature on a route and 50 pieces of rainfall data indicating that there is rainfall on the route, the 50 pieces of temperature data may undergo their own data fusion processing and the 50 pieces of rainfall data may undergo their own data fusion processing before being provided to the neural network.

[0042] Preferably, when there are multiple types of vehicle data 106 relating to the configuration of at least one vehicle, each type of vehicle data may undergo a respective data fusion process before being provided to the neural network.

[0043] For example, if there are 100 vehicle data sets including 50 wheel or axle data sets and 50 sand spreader data sets, the 50 wheel or axle data sets may undergo their own data fusion processing and the 50 sand spreader data sets may undergo their own data fusion processing before being provided to the neural network.

[0044] Preferably, the data integration process may include at least one of the following: -Determining the sum of the data, determining the average of the data, determining the absolute minimum of the data, determining the absolute maximum of the data.

[0045] For example, the data may be integrated to obtain vectors from the original data matrices of the time series. This integration step represents a preliminary analysis of the data. The integration of environmental data may be more advanced in the case of live video streaming, for example, when foreign object recognition as defined in "WO2021100003" is used. The choice of computational method depends on the characteristics of the input data.

[0046] Preferably, the at least one brake data 102 relating to the performance of the at least one vehicle's brake system, the at least one environmental data 104 relating to the environmental conditions of the route traveled by the vehicle, and the at least one vehicle data 106 relating to the structure of the at least one vehicle may be correlated by a time variable. Such a time variable may also be provided to the neural network or may be known, such that the neural network may correlate the at least one brake data 102 relating to the performance of the at least one vehicle's brake system, the at least one environmental data 104 relating to the environmental conditions of the route traveled by the vehicle, and the at least one vehicle data 106 relating to the structure of the at least one vehicle received at the input.

[0047] In a further aspect, the present invention relates to a system for predicting the expected deceleration of at least one vehicle, in particular at least one railway vehicle, comprising at least one computer arranged to carry out a method according to any of the preceding claims.

[0048] For example, the computer 101 may include at least one control means such as a processor, microprocessor, controller, microcontroller, FPGA, PLC, control unit, control box, or the like.

[0049] Preferably, the computer is arranged to receive from communication means of the at least one vehicle at least one brake data relating to the performance of a braking system of the at least one vehicle, at least one environmental data relating to environmental conditions of a route travelled by the vehicle, and at least one vehicle data relating to the structure of the at least one vehicle for providing to the neural network.

[0050] The communication means may be, for example, the vehicle's CAN network.

[0051] Preferably, the computer is arranged to receive from control means of additional vehicles travelling along the route at least one brake data relating to the performance of the braking system of the at least one vehicle, at least one environmental data relating to the environmental conditions of the route travelled by the vehicles, and at least one vehicle data relating to the structure of the at least one vehicle, for providing to the neural network.

[0052] Preferably, the computer may be arranged to receive from a control system of the at least one vehicle at least one brake data relating to performance of a braking system of the at least one vehicle, at least one environmental data relating to environmental conditions of a route travelled by the vehicle, and at least one vehicle data relating to the structure of the at least one vehicle, for providing to the neural network. The at least one brake data relating to performance of a braking system of the at least one vehicle, at least one environmental data relating to environmental conditions of a route travelled by the vehicle, and at least one vehicle data relating to the structure of the at least one vehicle may be generated by the control system as a function of a predetermined adhesion map.

[0053] An adhesion map can be understood as a correspondence between geographical locations on a route (e.g., a railway line) and relative wheel adhesion values ​​and available rolling surfaces (e.g., the rails of a line).

[0054] Preferably, the at least one vehicle may comprise at least one rail car, or the vehicles may be multiple and associated with one another to form a train of vehicles, such as a rail car.

[0055] However, the present invention is preferably applicable to any type of vehicle, including, for example, rail cars / trains, road vehicles, automobiles, trucks (e.g., highway semi-trailer trucks, mining trucks, logging trucks, etc.), and the path may be, for example, a track, road or path.

[0056] Below we describe an example of a real-time embodiment applicable to rail vehicles or trains.

[0057] For example, in a railroad train consisting of multiple rail cars, a deep (e.g., "feed-forward") neural network may be fed with pre-analyzed input data in the cars to generate predictions of car deceleration. Training may be performed offline by providing input data acquired during a test run of the cars and comparing the car deceleration generated by the neural network to actual average car deceleration based on measurements.

[0058] The trained neural network can be incorporated into a brake control system to predict train deceleration.

[0059] Neural networks can be used in four vehicle architectures, depending on the origin of the data used for prediction.

[0060] The local architecture allows the control unit of each vehicle braking system to calculate the expected vehicle deceleration based on locally available data.

[0061] The established architecture allows a control unit of a vehicle's braking system to determine the expected deceleration of the vehicle based on data shared over a communication means (e.g., a braking system bus) from all other control units of the vehicle's braking system.

[0062] The data from the preceding vehicle allows the control unit of the braking system to calculate the expected deceleration of the vehicle based on the data shared from the preceding vehicle by the control means TCMS, which acts as a control system agent for the vehicle.

[0063] Vehicle data from the adhesion map allows the control unit to calculate the expected deceleration of the train / rail car based on shared data from the TCMS that is derived from the adhesion map.

[0064] Thus, an advantage realized is that a solution has been provided that makes vehicle performance predictable, particularly in terms of vehicle deceleration performance, which prediction can be used to improve braking system performance even under the worst environmental conditions.

[0065] Various aspects and embodiments of a computer-implemented method for predicting the expected deceleration of at least one vehicle and a system for predicting the expected deceleration of at least one vehicle according to the present invention have been described. It should be understood that each embodiment may be combined with any other embodiment. Furthermore, the present invention is not limited to the described embodiments, but various modifications are possible within the scope defined by the appended claims.

Claims

1. 1. A computer-implemented method for predicting expected deceleration of at least one vehicle, in particular at least one rail vehicle, comprising: a) providing at least one brake data (102) relating to the performance of the braking system of said at least one vehicle, at least one environmental data (104) relating to the environmental conditions of a route traveled by said vehicle, and at least one vehicle data (106) relating to the structure of said at least one vehicle as inputs to a neural network (100); b) predicting, via the neural network (100), an expected deceleration value (108) of the at least one vehicle based on the at least one brake data (102), the at least one environmental data (104), and the at least one vehicle data (106); 10. A computer-implemented method comprising:

2. training the neural network (100) before steps a) and b); The training of the neural network comprises: - providing as inputs to said neural network at least one braking data, at least one environmental data, at least one vehicle data, and an expected deceleration value that is a function of said at least one braking data, said at least one environmental data and said at least one vehicle data; - determining a value of at least one parameter of the neural network according to the received at least one brake data, the at least one environmental data, the at least one vehicle data and the predicted deceleration value.

3. 3. The method of claim 1, wherein the training is based on a "backpropagation" algorithm.

4. The method according to any one of claims 1 to 3, wherein the neural network (100) has a feed-forward structure.

5. The at least one braking data (102) relating to the performance of the at least one vehicle's braking system is: at least one deceleration data indicating a deceleration value of at least one wheel or at least one axle of said at least one vehicle; - at least one data indicating the number of wheels in a skidding phase of said at least one vehicle; at least one data representative of the deceleration of said at least one vehicle during an initial stage of braking; at least one data representative of the speed of actuation of at least one braking means of said braking system; at least one data indicative of a skid velocity of said at least one wheel or said at least one axle of at least one vehicle; at least one datum representative of a steady state value of at least one braking means of said braking system; at least one data indicating the activation of the sand distribution means of said at least one vehicle; - at least one data indicative of activation of the magnetic brake pads of said at least one vehicle; - at least one data indicating the activation of a system / function arranged to compensate for deviations from the expected deceleration value; - at least one piece of data indicating that there is a malfunction of the braking means of said braking system; - at least one data indicative of the number of activations over time of an exhaust valve associated with at least one braking means of said braking system of said at least one vehicle.

6. The at least one environmental data (104) relating to environmental conditions of the route traveled by the vehicle is: - at least one image data or at least one video data of said path, - at least one temperature data indicating the temperature on said path; - at least one rainfall data indicating the presence of rainfall on the route; - at least one moisture datum indicative of the moisture level on said path; - at least one adhesion data indicative of the adhesion level on said path; - at least one route data indicating the outline of said route.

7. The at least one vehicle data (106) relating to the configuration of the at least one vehicle includes: at least one nominal deceleration data indicating a nominal deceleration value of said at least one vehicle; - at least one wheel or axle data indicating the number of wheels or axles of said at least one vehicle; at least one sand distribution means data indicating the number of sand distribution means of said at least one vehicle; - at least one data of magnetic pads MTB indicating the number of magnetic brake pads of said at least one vehicle; - at least one anti-skid system data indicating that the anti-skid system WSP acts on each bogie or each axle of the vehicle; and at least one data relating to the presence of deceleration compensation, indicating that the vehicle has a system / function for compensating for the at least one vehicle's missed deceleration.

8. if there are multiple types of braking data relating to the performance of the at least one vehicle's braking system, each type of braking data undergoes a respective data fusion process before being provided to the neural network; When there are multiple types of environmental data related to environmental conditions of a route traveled by the vehicle, each type of environmental data is subjected to a respective data integration process before being provided to the neural network; 8. The method according to claim 1, wherein, when there are multiple types of vehicle data relating to the structure of the at least one vehicle, each type of vehicle data is subjected to a respective data integration process before being provided to the neural network.

9. The data integration process includes:

9. The method of claim 8, comprising at least one of determining the sum of said data, determining the average of said data, determining the absolute minimum in said data, and determining the absolute maximum in said data.

10. A system for predicting the expected deceleration of at least one vehicle, in particular at least one railway vehicle, comprising at least one computer (101) arranged to perform the method according to any one of claims 1 to 9.

11. 11. The system for predicting expected deceleration of claim 10, wherein the computer is arranged to receive from communication means of the at least one vehicle the at least one brake data relating to performance of a braking system of the at least one vehicle, the at least one environmental data relating to environmental conditions of a route traveled by the vehicle, and the at least one vehicle data relating to a structure of the at least one vehicle for providing to a neural network.

12. 11. The system for predicting expected deceleration as described in claim 10, wherein the computer is arranged to receive from control means of additional vehicles passing through the route the at least one brake data relating to the performance of the braking system of the at least one vehicle, the at least one environmental data relating to the environmental conditions of a route along which the vehicle is traveling, and the at least one vehicle data relating to the structure of the at least one vehicle, for providing to a neural network.

13. the computer is arranged to receive from a control system of the at least one vehicle the at least one brake data relating to a performance of a brake system of the at least one vehicle, the at least one environmental data relating to environmental conditions of a route traveled by the vehicle, and the at least one vehicle data relating to a structure of the at least one vehicle, for providing to a neural network; 11. The system for predicting expected deceleration as described in claim 10, wherein the at least one brake data relating to the performance of the brake system of the at least one vehicle, the at least one environmental data relating to the environmental conditions of a route traveled by the vehicle, and the at least one vehicle data relating to the structure of the at least one vehicle are generated by the control system as a function of a predetermined adhesion map.

Images