A subway train slip prevention method based on gradient descent

Abstract

The application discloses a subway train slip prevention method based on gradient descent, comprising the following steps: S1, measuring the train axle braking torque and the overall braking rate of the train body; S2, determining the algorithm learning rate according to the creep rate; S3, iteratively controlling the maximum braking rate according to the algorithm learning rate, the axle braking torque value and the acceleration value; S4, implementing the corresponding braking torque according to the target braking rate and the iteration value, so as to reach the target braking rate of the train or converge to the extreme point, thereby preventing the occurrence of slip. The application measures the axle braking torque and the braking rate of the train body, and uses the gradient descent method to make the braking rate of the train approach the maximum braking rate provided by the track in real time, thereby avoiding the occurrence of train slip.

Description

Technical Field

[0001] This invention relates to the field of urban rail transit signal control technology, specifically to a method for preventing subway train slippage based on gradient descent. Background Technology

[0002] During subway train operation, the track environment is constantly changing, and the adhesion between the wheels and rails is affected by external environmental factors such as rain, snow, dust, and fallen leaves. The maximum friction between the wheels and rails also varies over time. If the braking force applied by the subway train exceeds the maximum friction force that the track can provide, slippage will occur, leading to wear on the wheels and rails. In severe cases, it can damage the rails, thus affecting the operational safety of the entire line. Therefore, methods to prevent slippage are crucial.

[0003] Currently, methods for preventing slippage include the following:

[0004] First, environmental sensors acquire external environmental parameters, which are then input into a model to predict track conditions, thereby preventing train slippage. For example, CN113147843B discloses a train automatic control method based on environmental perception and signaling systems. This method uses meteorological station data from the track section and a Kalman filter to estimate track conditions. The system then allocates an appropriate braking rate based on the track conditions to prevent slippage.

[0005] These methods rely on the completeness of empirical model algorithms. Once a scenario not considered by the model occurs, or a factor that is difficult for sensors to measure, the predictions given by the model algorithm will deviate significantly from the actual situation.

[0006] Secondly, by inputting the wheel speed, train acceleration, and other wheel motion sensor signals of each train wheel into a preset algorithm for comparison, it is determined whether the wheels are slipping. If wheel slippage is detected, the braking force is immediately and significantly reduced, and after the slippage disappears, the braking force is slowly increased. If slippage is detected again, the above process is repeated. For example, CN102991489B discloses a safe train speed and distance measurement system and method for detecting and compensating for wheel slippage and wheel lock-up, which uses train speed and acceleration information and fixed thresholds to detect wheel slippage or lock-up and can compensate for train speed.

[0007] These methods require sensors to detect abnormal wheel and vehicle movement after train slippage has already occurred to determine the slippage status. Such methods cannot prevent train slippage; they can only take measures to mitigate its impact after it has happened. In special slippage scenarios where the wheel and vehicle movement is outside the algorithm's consideration, false positives and false negatives are prone to occur.

[0008] Third, the train reports the slippage status to the central processing unit, which then notifies other trains to enter rain and snow mode in specific areas to reduce braking force to prevent slippage. For example, CN105549587B discloses a train automatic driving control method and system for rain and snow weather, which sends a command to all trains to reduce braking force after determining that it is rain or snow weather.

[0009] These methods require the first or a few trains to slip before determining that the entire line or area has entered a rain / snow mode and instructing subsequent trains to reduce braking force. This method cannot prevent the first or a few trains from slipping. Furthermore, because real-time track conditions are unavailable, uniformly lower braking force is applied across the entire line or area to avoid subsequent trains slipping, severely impacting subway operational efficiency.

[0010] Fourth, some patents utilize torque. For example, CN101985300B discloses a torque control method to prevent drive wheel slippage. It uses a fixed value of the wheel speed difference threshold to determine whether the vehicle has slipped, and after slippage, it determines the amount of torque reduction based on the wheel speed difference and a lookup table. CN112389435B discloses a slippage torque determination method, device, and vehicle. It uses wheel speed to determine whether the vehicle has slipped and calculates the amount of slippage, determining the value of reducing vehicle torque based on the level of slippage. CN114734981A discloses a vehicle anti-slip control method, control device, storage medium, and processor. It determines whether the wheels are in a slipping state based on the driving status, and then determines the response torque threshold of the power system corresponding to the current driving mode based on the current driving mode. CN107962981A discloses an active torque reduction strategy for electric vehicles in slippage conditions. It uses wheel speed difference to determine slippage and determines the value of reducing vehicle torque based on the level of slippage. CN117507842A discloses a torque control method, device, and vehicle for electric vehicles. After slippage, the longitudinal acceleration value and the rate of change of acceleration are used to determine the torque reduction value during slippage by looking up a table. CN113085863A discloses a method, device, equipment, and storage medium for preventing slippage. Based on wheel speed and axle speed information, it determines whether the vehicle is in the starting phase and limits the output torque in stages to prevent idling. CN103052552B discloses a method for controlling wheel slippage in an electric traction vehicle. When slippage is detected, the torque is reduced exponentially within a preset time window. After the slippage is determined to have disappeared, the torque is increased linearly until a new slippage occurs. CN118061806A discloses a control method, device, axle controller, and vehicle for a wheel axle controller. It determines whether slippage occurs by the ratio of wheel speed change to torque change, and after slippage, sets the torque to the torque estimated by the detection device.

[0011] These methods require the user to first enter a slipping state before determining if slipping has occurred, thus failing to prevent or avoid it. Only after slipping has taken place do they use fixed segmentation and grading to determine the amount of braking to be reduced, employing preset methods or lookup tables. If a scenario occurs that is not within the preset conditions, it is easy to reduce the braking force too much, failing to fully utilize the maximum friction, or to create a cycle of determining slipping - reducing braking force - determining no slipping - increasing braking force - then determining slipping again.

[0012] To address the above issues, a gradient descent-based method for preventing subway train slippage is proposed. Summary of the Invention

[0013] The purpose of this invention is to overcome the shortcomings of existing methods and provide a gradient descent-based method for preventing subway train slippage. By measuring the wheel axle braking torque and the vehicle body braking rate, the gradient descent method is used to make the train's braking rate approach the maximum braking rate that the track can provide in real time, thereby avoiding the occurrence of train slippage.

[0014] The technical solution to achieve the above objectives is:

[0015] A gradient descent-based method for preventing subway train slippage includes:

[0016] Step S1: Measure the braking torque of the train wheel axles and the overall braking rate of the train body;

[0017] Step S2: Determine the algorithm learning rate based on the creep rate;

[0018] Step S3: Based on the algorithm learning rate, wheel axle braking torque value, and acceleration value, iteratively control the approximation of the maximum braking rate;

[0019] Step S4: Based on the target braking rate and the iterative value, implement the corresponding braking torque to achieve the train's target braking rate or converge to the extreme point, thereby preventing slippage.

[0020] Preferably, step S1 includes:

[0021] Step S11: Install a wheel axle torque sensor on the train's brake axle to measure the train's wheel axle braking torque;

[0022] Step S12: Install an accelerometer on the train body to measure the overall braking rate of the train body, wherein the braking rate of the train body is proportional to the adhesion coefficient between the wheel and the rail.

[0023] Preferably, in step S2, a gradient descent method is used on the adhesion coefficient-creep rate curve to control the creep rate to approach the extreme point, as shown in the following formula:

[0024]

[0025] In the formula, Ct+1 Let C be the creep rate at time t+1. t Let μ be the creep rate at time t. t Let ΔT be the adhesion coefficient at time t. t Let ΔT be the time difference between time t and time t-1. t+1 Let ΔC be the time difference between time t and time t+1. t C t With C t-1 The difference, Δμ t For μ t and μ t-1 The difference, where α is the learning rate;

[0026] Since the vehicle body braking rate is directly proportional to the adhesion coefficient, and the wheel axle braking torque is positively correlated with the creep rate, and considering the adhesion coefficient-creep rate curve, the vehicle body braking rate-wheel axle braking torque curve also has an extreme point, corresponding to the optimal wheel axle braking torque and the maximum braking rate under the current rail surface conditions. From the above formula, the control formula for wheel axle braking torque is:

[0027]

[0028] In the formula, Torque t+1 Torque is the wheel and axle braking torque applied by the command at time t+1. t Let ΔBrakeRate be the wheel and axle braking torque applied by the command at time t. t ΔTorque is the difference between the braking rate measured by the accelerometer at time t and time t-1. t This is the difference between the wheel axle braking torque measured by the wheel axle torque sensor at time t and time t-1.

[0029] Preferably, in step S2, the extreme points of different rail surface conditions are distributed within a creep rate region. If the creep rate is small, no stress-induced deformation occurs on the wheel-rail contact surface, and the wheel-rail friction has not yet reached its extreme value. Conversely, if the wheel-rail friction is to reach its extreme value, a certain creep rate is required. If the creep rate is large, significant sliding has already occurred between the wheel and rail, and the extreme point cannot appear in this region. Thus, the following is obtained:

[0030] When the creep rate is small or large, a larger learning rate can be used to enable the system to converge quickly to the region where the extreme point may exist.

[0031] In regions where extreme points may exist, a smaller learning rate is used to prevent the system from exhibiting large oscillations and non-convergence around the extreme points.

[0032] This can be further described by the following formula:

[0033]

[0034] In the formula, CreepRate is the current creep rate, and CrLimit is the creep rate. Low CrLimit is the lower limit of creep rate. High α is the upper limit of creep rate. ascend α peak and α descend Both are learning rates, and α descend >α ascend >α peak >0.

[0035] Preferably, in step S3, since the operating cycle of the braking system is fixed, there is no need to consider the difference in control time intervals. Therefore, during braking, the wheel axle braking torque is calculated to approximate the maximum braking rate, and the formula is as follows:

[0036]

[0037] when At that time, ΔTorque t Set as It is typically set to 2-5 times the smallest unit of system calculation precision to prevent ΔTorque. t When the value approaches 0, the precision of the division decreases, causing the above expression to not converge.

[0038] Preferably, in step S4, the braking system determines the braking torque to be output according to the following conditions in sequence:

[0039] Step S41, the braking system determines Torque. t+1 Is it smaller than Torque? t ;

[0040] Step S42, if Torque t+1 <Torque t Then the wheel and axle braking torque applied by the train command is set to Torque. t+1 ;

[0041] Step S43, conversely, determine the magnitude of the train's current braking rate and the target braking rate;

[0042] Step S44: If the current braking rate of the train is equal to the target braking rate, then the applied wheel and axle braking torque is set to Torque. t ;

[0043] Step S45: If the current braking rate of the train is greater than the target braking rate, then reduce the applied wheel and axle braking torque to approach the target braking rate according to the preset algorithm.

[0044] Step S46: If the current braking rate of the train is less than the target braking rate, the wheel and axle torque applied by the train command is set to Torque. t+1 .

[0045] The beneficial effects of this invention are as follows: By measuring the wheel axle braking torque and the vehicle body braking rate, this invention uses a gradient descent method to make the train's braking rate approach the maximum braking rate that the track can provide in real time, thereby preventing train slippage; this invention can adapt to all environmental types, whether dry, wet, sandblasted, or oil-sprayed, without the need to add sensors for external environmental factors, reducing subway operating costs; when braking under relatively stable and harsh track surface conditions, it can prevent train slippage in advance; it can obtain the maximum track friction force in real time, accurately, quickly, and adaptively, improving subway operating efficiency. Attached Figure Description

[0046] Figure 1 This is a flowchart of a subway train slippage prevention method based on gradient descent according to the present invention;

[0047] Figure 2 This is a flowchart illustrating the measurement of train wheel axle braking torque and overall braking rate of the train body in this invention.

[0048] Figure 3 This is a flowchart illustrating the specific process by which the braking system in this invention determines the braking torque to be output according to a sequence of conditions.

[0049] Figure 4 It is a curve showing the relationship between the rail surface adhesion coefficient and the creep rate;

[0050] Figure 5 This is a schematic diagram of the extreme points and partitions of the adhesion coefficient-creep ratio relationship;

[0051] Figure 6 This is a graph showing the gradient relationship between the adhesion coefficient and the creep rate in the adhesion region in this invention.

[0052] Figure 7 This is a graph showing the relationship between the adhesion coefficient and the creep rate gradient in the sliding region in this invention.

[0053] Figure 8 This is a schematic diagram showing the region where the extreme points exist in the adhesion coefficient-creep rate gradient relationship in this invention, and the upper and lower limits of the creep rate in the region;

[0054] Figure 9 This is a schematic diagram of the creep rate values ​​of the sliding region and the adhesive region in the same adhesion coefficient-creep rate gradient relationship in this invention. Detailed Implementation

[0055] The technical solution of the present invention will now be clearly and completely described in conjunction with the accompanying drawings. In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0056] The invention will now be further described with reference to the accompanying drawings.

[0057] Creep refers to the slight sliding that occurs when there is a longitudinal force between a wheel and a rail due to frictional deformation between the wheel and rail. It is a state between complete adhesion and complete sliding.

[0058] Creep rate = (vehicle speed - wheel speed) / vehicle speed, which indicates the degree of wheel slippage.

[0059] There is a positive correlation between wheel and axle braking torque and creep rate.

[0060] The relationship between the rail surface adhesion coefficient and the creep rate is as follows: Figure 4 As shown, although the adhesion coefficient-creep rate curves vary with the changes in rail surface conditions, the trend of the adhesion coefficient-creep rate curves under different rail surface conditions is the same. As the creep rate increases, the adhesion coefficient first rises to an extreme point and then decreases.

[0061] like Figure 5 As shown, the extreme points correspond to the maximum adhesion coefficient and optimal creep rate under the current rail surface conditions. Trains operating near these extreme points can utilize the maximum frictional force provided by the track. Dividing the area by the extreme points, the left side represents the adhesion region, and the right side represents the slip region.

[0062] When a train is running in the adhesion zone, as the wheel-axle braking torque increases, the creep rate between the wheel and rail increases, and the braking force provided by the rail surface also increases. Within this zone, the wheel-axle braking torque is positively correlated with the rail surface friction, and the train will not slip.

[0063] If the wheel-axle braking torque continues to increase, the wheel-rail adhesion state will exceed its extreme point and enter the slippage region. As the creep rate increases, the braking force provided by the rail surface will decrease. In this region, the wheel-axle braking torque and the rail surface friction are negatively correlated. With the increase of the wheel-axle braking torque, the wheel and rail will gradually become unstable and slip, until they lock up completely, seriously affecting driving safety.

[0064] Subway track conditions are complex and highly variable. For example, the junction between tunnels and open tracks, severe weather such as rain, snow, and dust storms, maintenance operations such as sandblasting and oiling of vehicles, accumulation of debris like fallen leaves, and changes in train speed all affect the adhesion coefficient-creep rate curve, altering the maximum friction force the track can provide. Changes in track conditions and their impacts are difficult to fully observe, measure, and estimate. Therefore, real-time acquisition of the extreme points of the adhesion coefficient-creep rate curve under current track conditions is crucial for preventing slippage, improving operational efficiency, and ensuring train safety.

[0065] like Figure 1 As shown, a gradient descent-based method for preventing subway train slippage includes:

[0066] Step S1: Measure the braking torque of the train wheel axles and the overall braking rate of the train body.

[0067] like Figure 2 As shown, step S1 includes:

[0068] Step S11: Install a wheel axle torque sensor on the train's brake axle to measure the train's wheel axle braking torque.

[0069] Step S12: Install an accelerometer on the train body to measure the overall braking rate of the train body, wherein the braking rate of the train body is proportional to the adhesion coefficient between the wheel and the rail.

[0070] Step S2: Determine the learning rate of the algorithm based on the creep rate.

[0071] In the embodiments, such as Figure 6 , 7 As shown, on the adhesion coefficient-creep rate curve, the gradient descent method is used to control the creep rate to approach the extreme point, as shown in the following formula:

[0072]

[0073] In the formula, C t+1 Let C be the creep rate at time t+1. t Let μ be the creep rate at time t. t Let ΔT be the adhesion coefficient at time t. t Let ΔT be the time difference between time t and time t-1. t+1 Let ΔC be the time difference between time t and time t+1. t C t With C t-1 The difference, Δμ t For μ t and μ t-1 The difference, where α is the learning rate;

[0074] Since the vehicle body braking rate is directly proportional to the adhesion coefficient, and the wheel axle braking torque is positively correlated with the creep rate, and considering the adhesion coefficient-creep rate curve, the vehicle body braking rate-wheel axle braking torque curve also has an extreme point, corresponding to the optimal wheel axle braking torque and the maximum braking rate under the current rail surface conditions. From the above formula, the control formula for wheel axle braking torque is:

[0075]

[0076] In the formula, Torque t+1 Torque is the wheel and axle braking torque applied by the command at time t+1. t Let ΔBrakeRate be the wheel and axle braking torque applied by the command at time t. t ΔTorque is the difference between the braking rate measured by the accelerometer at time t and time t-1. t This is the difference between the wheel axle braking torque measured by the wheel axle torque sensor at time t and time t-1.

[0077] In the embodiments, such as Figure 8 As shown, the extreme points of different rail surface conditions are distributed within a creep rate region. If the creep rate is small, there is no stress-induced deformation on the wheel-rail contact surface, and the wheel-rail friction has not yet reached its extreme value. Conversely, if the wheel-rail friction is to reach its extreme value, a certain creep rate is required. If the creep rate is large, significant sliding has already occurred between the wheel and rail, and the extreme point cannot appear in this region. Therefore, we can conclude that:

[0078] When the creep rate is small or large, a larger learning rate can be used to enable the system to converge quickly to the region where the extreme point may exist.

[0079] In regions where extreme points may exist, a smaller learning rate is used to prevent the system from exhibiting large oscillations and non-convergence around the extreme points.

[0080] This can be further described by the following formula:

[0081]

[0082] In the formula, CreepRate is the current creep rate, and CrLimit is the creep rate. Low CrLimit is the lower limit of creep rate. High α is the upper limit of creep rate. ascend α peak and α descend Both are learning rates, and α descend >α ascend >α peak >0.

[0083] Step S3: Based on the algorithm learning rate, wheel axle braking torque value, and acceleration value, iteratively control the approximation of the maximum braking rate.

[0084] In this embodiment, since the operating cycle of the braking system is fixed, there is no need to consider the difference in control time intervals. Therefore, during braking, the wheel axle braking torque is calculated to approximate the maximum braking rate, and the formula is as follows:

[0085]

[0086] when At that time, ΔTorque t Set as It is typically set to 2-5 times the smallest unit of system calculation precision to prevent ΔTorque. t When the value approaches 0, the precision of the division decreases, causing the above expression to not converge.

[0087] Step S4: Based on the target braking rate and the iterative value, implement the corresponding braking torque to achieve the train's target braking rate or converge to the extreme point, thereby preventing slippage.

[0088] In this embodiment, during the train braking process, the applied wheel axle braking torque gradually increases from zero, and the train's creep rate also gradually increases.

[0089] During the initial braking phase, the train operates in the adhesion region of the adhesion coefficient-creep rate curve. During this phase, the train will control the increase of wheel and axle braking torque until the target braking rate of the train is reached or converges to the extreme point, thereby preventing slippage.

[0090] As the train brakes and moves, the rail surface conditions may change, causing the wheel-rail state to slip within the original wheel-axle braking torque. The system will control the wheel-axle braking torque to decrease until it converges to the extreme point, thereby preventing slippage.

[0091] Within the slip region, the system cannot stop converging towards the extreme point simply because the target braking rate has been reached. Instead, it must first converge to the extreme point, then compare the maximum braking rate with the target braking rate to decide whether to continue reducing the wheel and axle braking torque. This is because, to achieve the same braking rate, the creep rate in the slip region is higher than that in the adhesion region. If the train's wheel-rail condition remains within the slip region, it will cause more wheel-rail wear, as follows: Figure 9 As shown.

[0092] Therefore, as Figure 3 As shown, the braking system will determine the appropriate braking torque based on the following conditions in sequence:

[0093] Step S41, the braking system determines Torque. t+1 Is it smaller than Torque?t ;

[0094] Step S42, if Torque t+1 <Torque t Then the wheel and axle braking torque applied by the train command is set to Torque. t+1 ;

[0095] Step S43, conversely, determine the magnitude of the train's current braking rate and the target braking rate;

[0096] Step S44: If the current braking rate of the train is equal to the target braking rate, then the applied wheel and axle braking torque is set to Torque. t ;

[0097] Step S45: If the current braking rate of the train is greater than the target braking rate, the applied wheel and axle braking torque is reduced to approach the target braking rate according to the preset algorithm.

[0098] Step S46: If the current braking rate of the train is less than the target braking rate, the wheel and axle torque applied by the train command is set to Torque. t+1 .

[0099] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for preventing subway train slippage based on gradient descent, characterized in that, include: Step S1: Measure the braking torque of the train wheel axles and the overall braking rate of the train body; Step S2: Determine the algorithm learning rate based on the creep rate; Step S3: Based on the algorithm learning rate, wheel axle braking torque value, and acceleration value, iteratively control the approximation of the maximum braking rate; Step S4: Based on the target braking rate and the iterative value, implement the corresponding braking torque to achieve the train's target braking rate or converge to the extreme point, thereby preventing slippage. In step S2, the gradient descent method is used on the adhesion coefficient-creep rate curve to control the creep rate to approach the extreme point, as shown in the following formula: ； In the formula, for The creep rate at any given time for The creep rate at any given time for The adhesion coefficient at any given time. for Time and The time difference between moments for Time and The time difference between moments for and The difference, for and The difference, The learning rate; Since the vehicle body braking rate is directly proportional to the adhesion coefficient, and the wheel axle braking torque is positively correlated with the creep rate, and considering the adhesion coefficient-creep rate curve, the vehicle body braking rate-wheel axle braking torque curve also has an extreme point, corresponding to the optimal wheel axle braking torque and the maximum braking rate under the current rail surface conditions. From the above formula, the control formula for wheel axle braking torque is: ； In the formula, for The wheel and axle braking torque applied at the specified time. for The wheel and axle braking torque applied at the specified time. The braking rate measured by the accelerometer Time and The difference in time, The wheel axle braking torque measured by the wheel axle torque sensor is in Time and The difference in time.

2. The method for preventing subway train slippage based on gradient descent according to claim 1, characterized in that, Step S1 includes: Step S11: Install a wheel axle torque sensor on the train's brake axle to measure the train's wheel axle braking torque; Step S12: Install an accelerometer on the train body to measure the overall braking rate of the train body, wherein the braking rate of the train body is proportional to the adhesion coefficient between the wheel and the rail.

3. The method for preventing subway train slippage based on gradient descent according to claim 1, characterized in that, In step S2, the extreme points of different rail surface conditions are distributed within a creep rate region. If the creep rate is small, no stress-induced deformation occurs on the wheel-rail contact surface, and the wheel-rail friction has not yet reached its extreme value. Conversely, if the wheel-rail friction is to reach its extreme value, a certain creep rate is required. If the creep rate is large, significant sliding has already occurred between the wheel and rail, and the extreme point cannot appear in this region. Thus, we obtain: When the creep rate is small or large, a larger learning rate can be used to enable the system to converge quickly to the region where the extreme point may exist. In regions where extreme points may exist, a smaller learning rate is used to prevent the system from exhibiting large oscillations and non-convergence around the extreme points. This can be further described by the following formula: ； In the formula, The current creep rate, This is the lower limit of the creep rate. This represents the upper limit of the creep rate. , and All are learning rates, and .

4. The method for preventing subway train slippage based on gradient descent according to claim 3, characterized in that, In step S3, since the operating cycle of the braking system is fixed, there is no need to consider the difference in control time intervals. Therefore, during braking, the wheel axle braking torque is calculated to approximate the maximum braking rate, and the formula is as follows: ； when hour, Set as , It is usually set to 2-5 times the smallest unit of system calculation precision to prevent... When the value approaches 0, the precision of the division decreases, causing the above expression to not converge.

5. The method for preventing subway train slippage based on gradient descent according to claim 4, characterized in that, In step S4, the braking system will determine the braking torque that should be output according to the following conditions in sequence: Step S41, Braking system determination Is it less than ; Step S42, if The wheel and axle braking torque applied by the train command is set to ; Step S43, conversely, determine the magnitude of the train's current braking rate and the target braking rate; Step S44: If the current braking rate of the train is equal to the target braking rate, then the wheel and axle braking torque to be applied is set to... ; Step S45: If the current braking rate of the train is greater than the target braking rate, then reduce the applied wheel and axle braking torque to approach the target braking rate according to the preset algorithm. Step S46: If the current braking rate of the train is less than the target braking rate, the wheel and axle torque applied by the train command is set to... .

Images