A method and system for dynamic distribution of axle control power of a city rail vehicle based on real-time adhesion estimation

Abstract

This invention discloses a method and system for dynamic distribution of axle control power in urban rail vehicles based on real-time adhesion estimation. The method includes: acquiring real-time operating state data of the vehicle; calculating the dynamic vertical load of each axle during braking by combining the vehicle's longitudinal dynamics model; constructing an extended state observer and using it to estimate the wheel-rail adhesion and wheel-rail contact state parameters of each axle in real time, and calculating the theoretical maximum available adhesion coefficient accordingly; applying an asymmetric rate of change constraint to the theoretical maximum available adhesion coefficient to obtain the safe adhesion limit for constraint; constructing a multi-objective optimization function for braking force distribution, including a braking force tracking error objective and an adhesion utilization rate objective, and weighting the two objectives using adaptive weighting coefficients; and solving the multi-objective optimization function to obtain the optimal braking force for each axle. This invention can significantly improve the adhesion utilization rate and active safety performance of urban rail vehicles under complex road conditions, effectively preventing wheel slippage.

Description

Technical Field

[0001] This invention belongs to the field of braking control technology for urban rail transit vehicles, specifically relating to a method and system for dynamic distribution of axle control power for urban rail vehicles based on real-time adhesion estimation. Background Technology

[0002] With the acceleration of urbanization, urban rail transit systems (such as subways and light rail) have developed rapidly due to their large capacity and high efficiency. The continuous increase in train speed and density places higher demands on the active safety performance of trains, especially the reliability of the braking system. As the core braking method for urban rail vehicles, the performance of the electro-pneumatic braking system directly affects the stopping accuracy and operational safety of the train.

[0003] Currently, in the field of brake anti-slip control for urban rail vehicles, existing technical solutions mainly focus on optimizing the Wheel Slide Protection (WSP) system. WSP systems typically function as a reactive safety mechanism, passively triggering only when the wheel slip ratio exceeds a preset threshold or a slipping trend is detected. This involves controlling the exhaust valve to rapidly reduce brake cylinder pressure and restore wheel-rail adhesion. However, this "post-slip intervention" control logic inherently suffers from hysteresis. Frequent inflation and deflation not only consume a large amount of compressed air but can also lead to unexpected increases in braking distance and scratches on the wheel-rail surface. To overcome the limitations of this passive anti-slip mechanism and reduce the triggering frequency of WSP, a more precise axle control strategy needs to be introduced at the front-end distribution of braking force to achieve proactive prevention of slippage.

[0004] However, existing control strategies have significant limitations in terms of braking force distribution. On the one hand, engineering applications commonly employ a proportional distribution mode based on static load, which calculates braking force solely based on the static axle load measured by air spring pressure, neglecting the dynamic axle load transfer effect caused by longitudinal deceleration during vehicle braking. This results in insufficient braking force distribution to the front axle when the load is increased, while excessive braking force distribution to the rear axle when the load is decreased. This distribution bias makes the rear axle prone to prematurely entering a skidding state before reaching the road adhesion limit, thus forcing the triggering of the rear-end WSP system.

[0005] On the other hand, while existing technologies have proposed optimized allocation methods based on empirical adhesion coefficients or preset adhesion-slip curves, these strategies inherently rely on offline historical data or idealized assumptions. Because the wheel-rail interface state is highly time-varying and uncertain due to environmental factors such as rain, snow, humidity, and pollutants, fixed empirical adhesion parameters are insufficient to accurately represent the current actual physical limits. When actual road conditions deteriorate, allocation strategies based on empirical values ​​often fail due to overestimation, failing to achieve truly optimal allocation.

[0006] Therefore, the existing braking force distribution strategy is essentially a coarse-grained distribution based on open-loop control logic. Lacking a real-time perception and feedback mechanism for time-varying wheel-rail adhesion limits, the distribution algorithm can only output control commands based on ideal or static parameter assumptions, unable to determine whether the current distribution value is approaching the physical limit. This blind distribution method prevents the system from actively limiting braking force output when facing low or variable adhesion conditions, making the anti-skid protection device the only safety line. Once the distributed braking force exceeds the actual adhesion capacity, the system can only passively wait for wheel slippage before the wheel slippage protection device (WSP) intervenes with a lag. This distribution mode, which overly relies on rear-end anti-skid protection and lacks front-end proactive adaptation capabilities, not only limits the overall vehicle braking efficiency but also increases the risk of wheel-rail abrasion. Summary of the Invention

[0007] This invention provides a method and system for dynamic allocation of axle control power for urban rail vehicles based on real-time adhesion estimation. It aims to reconstruct the wheel-rail contact state in real time by extending the state observer and introducing an adaptive weight adjustment mechanism based on global adhesion utilization. Under a multi-objective optimization framework, it dynamically coordinates braking force tracking and anti-skid safety, thereby improving the adhesion utilization rate and active safety performance of urban rail vehicles under complex road conditions.

[0008] To achieve the above technical objectives, the present invention adopts the following technical solution:

[0009] A method for dynamic allocation of axle control power for urban rail vehicles based on real-time adhesion estimation includes:

[0010] Acquire real-time operating status data of the vehicle, including vehicle speed, vehicle longitudinal deceleration, wheel-to-wheel angular velocity of each axle of the bogie, braking torque, and longitudinal braking force required by the whole vehicle.

[0011] Based on the vehicle's longitudinal dynamics model, the dynamic vertical loads on each axle during braking are calculated.

[0012] An extended state observer is constructed based on the vehicle longitudinal dynamics model and the LuGre friction model. The wheel-rail adhesion and wheel-rail contact state parameters of each axle are estimated in real time using the extended state observer, and the theoretical maximum usable adhesion coefficient is calculated accordingly.

[0013] By imposing an asymmetric rate of change constraint on the theoretical maximum usable adhesion coefficient, a safe adhesion limit for constraint is obtained;

[0014] A multi-objective optimization function for braking force distribution is constructed, including the braking force tracking error objective and the adhesion utilization rate objective, and the two objectives are weighted by adaptive weight coefficients;

[0015] Under the premise of satisfying actuator constraints and safety adhesion limit constraints, the multi-objective optimization function is solved to obtain the optimal braking force for each axis.

[0016] Furthermore, the specific method for calculating dynamic vertical loads is as follows:

[0017] First, based on the dynamic equilibrium of the car body's nose-dive, calculate the vertical load at the center pin of the front bogie. Vertical load at the center pin of the rear bogie And the longitudinal force of the car body on the front bogie. and longitudinal forces of the rear bogie 、:

[0018]

[0019] in, For vehicle body mass; It is the acceleration due to gravity; This represents the longitudinal deceleration amplitude; The height of the vehicle's center of gravity; This is the center distance between the front and rear bogies; For the first Braking force of the shaft; For the mass of a single bogie; , These represent the front axle and rear axle of the front bogie, respectively. , These represent the front and rear axles of the rear bogie, respectively.

[0020] Subsequently, based on the vertical load and longitudinal force at the bogie center pin, the dynamic vertical load of each axle and wheelset is obtained according to the moment balance equation. :

[0021]

[0022] in, , These are the dynamic vertical loads for axes 1 and 2, respectively. , These are the dynamic vertical loads of the 3rd and 4th axes, respectively; This refers to the bogie wheelbase. The height of the connection point between the car body and the bogie from the rail surface; The height of the bogie's center of gravity above the rail surface; This is the internal load redistribution coefficient, used to characterize the contribution ratio of the braking-induced pitching torque to the internal axle load redistribution of the bogie.

[0023] Furthermore, the extended state observer is constructed using the following method:

[0024] Based on the wheel-rail adhesion characteristics described by the LuGre friction model, slip velocity and internal friction state are selected as system state variables, and unknown wheel-rail contact condition parameters are also included. The nonlinear term is defined as the extended state. , construct the first Augmented state-space model of the axis:

[0025]

[0026] in, The first The sliding speed and internal friction state of the shaft; A function characterizing the nonlinear features of wheel-rail friction; The radius of the wheel; For the first The wheel's angular velocity; For the first The equivalent moment of inertia of the shaft; This is the contact spot deformation distribution coefficient; These are the contact surface bristle stiffness coefficient and bristle damping coefficient in the LuGre friction model, respectively. The overall damping coefficient is given, and it satisfies... ,in It is the coefficient of viscous friction; The dynamic coefficients are related to the dynamic vertical load and satisfy the following conditions: ,in For the first Dynamic vertical load on the shaft; For the first Braking torque of the shaft, To slow down the vehicle; For extended state The derivative;

[0027] Design of linear extended state observers for each axis based on augmented state-space model:

[0028]

[0029] in, For the first The state estimation vector of the axis, The observer gain matrix is... For the first The measured value of the shaft's sliding speed; The system coefficient matrix is ​​determined by the augmented state-space model; For including vehicle deceleration The known input vector; For the first The nonlinear disturbance compensation vector of the axis.

[0030] Furthermore, the wheel-rail adhesion and wheel-rail contact state parameters of each axle are estimated in real time using an extended state observer, and the theoretical maximum usable adhesion coefficient is calculated accordingly, including using the state estimation vector output by the observer. The estimated values ​​of each state component in the equation, combined with the definition of the extended state, are used to algebraically solve for the unknown wheel-rail contact condition parameters. Its inverse formula is:

[0031]

[0032] in, The first one is estimated by the observer. Shaft sliding speed, internal friction state, and expansion state; The Stribeck effect function is defined with the estimated slip velocity as the independent variable; the inverse solution process assumes the existence of a small slip. and Under the conditions;

[0033] Subsequently, the wheel-rail contact condition parameters obtained by inverse kinematics were used. Reconstruct the current wheel-rail adhesion-slip characteristic curve to obtain the steady-state adhesion coefficient. The parsing expression:

[0034]

[0035] Since the steady-state wheel-rail adhesion force is equal to the steady-state adhesion coefficient The expression, when multiplied by the dynamic vertical load, directly establishes the relationship between the contact state parameters and the macroscopic wheel-rail adhesion. In engineering calculations, the viscous friction term, which has a negligible impact on the peak point, is ignored. By solving the gradient equation of the steady-state adhesion coefficient with respect to the slip velocity... The optimal slip velocity that maximizes the adhesion coefficient was calculated. ;

[0036] Finally, the optimal slip velocity is obtained. Substituting back into the steady-state adhesion coefficient expression, the theoretical maximum usable adhesion coefficient at the current moment can be obtained. ;in, It is the first The sliding speed of the shaft, Let represent the steady-state adhesion coefficient of the i-th axis.

[0037] Furthermore, by imposing an asymmetric rate of change constraint on the theoretically maximum available adhesion coefficient, the first constraint used for the constraint is obtained. Safety adhesion limit of the shaft :

[0038]

[0039] in, This is the original deviation. This represents the safe stickiness limit for the current calculation cycle. This represents the safe adhesion limit of the previous calculation cycle; For the saturation limiting function, its specific operation rules are as follows: when When, the output value is ;when When, the output value is ;when When, the output value is ; and These are the descent rate threshold and ascent rate threshold of the safe adhesion limit, respectively, and satisfy the following conditions: This allows the adhesion limit to decrease rapidly when slippage is detected and increase slowly when adhesion is restored; the final output safe adhesion limit is... , For safety factors.

[0040] Furthermore, a multi-objective function including braking force tracking error target and adhesion utilization rate target. , is represented as:

[0041]

[0042] in, These represent the dynamic tracking error target and the adhesion utilization rate target, respectively. for The corresponding weighting coefficients, This is the longitudinal braking force required for the entire vehicle; For the first Braking force distributed across the shaft, The vector formed by the braking forces assigned to all axes; for The corresponding allocation coefficient vector; For the first Vertical load on the shaft; For the first Safety adhesion limit of the shaft.

[0043] Furthermore, the weighting coefficients The adaptive adjustment mechanism is as follows:

[0044] The global adhesion utilization index is calculated based on the ratio of total braking demand force to total available adhesive capacity of the vehicle. :

[0045]

[0046] in, For the longitudinal braking force required by the whole vehicle, For the first The safe adhesion limit of the shaft, For the first Vertical load on the shaft;

[0047] Construct a nonlinear mapping relationship based on the Sigmoid function, according to the global adhesion utilization index. Dynamically calculate the weighting coefficients of the adhesion utilization target :

[0048]

[0049] in, This is the conversion rate parameter, used to control the sensitivity of weight switching; The global adhesion utilization threshold;

[0050] Calculate the weighting coefficients of the dynamic tracking error target based on normalization constraints. :

[0051] .

[0052] Furthermore, a quadratic programming algorithm is employed to solve the multi-objective function. This yields the braking force distributed across all axes.

[0053] A dynamic power distribution system for axle control of urban rail vehicles based on real-time adhesion estimation is characterized by comprising a sensing unit, a braking control unit, and a braking execution unit.

[0054] The sensing unit is used to: collect real-time vehicle operating status data, including vehicle speed, wheel speed of each axle, total braking demand signal, and axle load signal;

[0055] The braking control unit is used to: execute the dynamic distribution method for axle control power of urban rail vehicles based on real-time adhesion estimation as described above, and calculate the optimal braking force distribution command for each axle in real time based on the data collected by the sensing unit.

[0056] The braking actuator is used to: receive the optimal braking force distribution command output by the braking control unit, and generate corresponding braking friction force through the basic braking device corresponding to each shaft.

[0057] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0058] 1. This invention, based on a real-time adhesion estimation-based dynamic distribution method for axle control power in urban rail vehicles, represents a technological leap from passive anti-skid to active adhesion utilization. By constructing an extended state observer, it utilizes the vehicle's existing wheel speed and braking torque signals to reconstruct the unmeasurable wheel-rail contact state in real time and calculate the maximum available adhesion coefficient, eliminating the need for expensive radar or optical sensors. Furthermore, an asymmetric rate-of-change constraint mechanism is introduced, allowing the estimated value to rapidly decrease when an adhesion drop is detected, and limiting the rate of increase when adhesion recovers. This mechanism effectively eliminates safety hazards caused by system execution delays and signal noise, ensuring that braking force output is actively limited before the WSP anti-skid system is triggered, preventing wheel lock-up.

[0059] 2. This invention, based on a real-time adhesion estimation-based dynamic allocation method for axle control power in urban rail vehicles, solves the problem that traditional fixed-weight strategies cannot simultaneously address braking demand and anti-skid safety. This invention innovatively introduces a global adhesion utilization index as a scheduling variable and employs an adaptive weight adjustment mechanism based on the Sigmoid function. This mechanism can intelligently decide based on current operating conditions: under good road conditions with sufficient adhesion, it automatically increases the braking force tracking weight to ensure precise control of the urban rail train's deceleration; under low-adhesion road conditions such as rain or snow, or in emergency braking conditions, it automatically yields to anti-skid safety weights. This dynamic trade-off strategy endows the control system with strong environmental adaptability, ensuring that the vehicle can achieve optimal braking performance under all weather conditions.

[0060] 3. This invention, based on a real-time adhesion estimation-based dynamic axle control power allocation method for urban rail vehicles, overcomes the limitations of static axle load allocation and significantly improves the overall vehicle adhesion utilization rate. The invention explicitly considers the dynamic axle load transfer effect caused by longitudinal deceleration during vehicle braking in the allocation model, calculating the dynamic vertical load of each axle in real time. Through a multi-objective optimization algorithm, braking force is preferentially allocated to the front axle due to inertial load increase, reducing allocation to the rear axle due to load reduction. This capacity-based allocation strategy effectively avoids wasting the adhesion capacity of the front axle and premature slippage of the rear axle due to overload, thereby minimizing braking distance while ensuring no slippage.

[0061] 4. This invention presents a dynamic allocation method for axle control power of urban rail vehicles based on real-time adhesion estimation. The extended state observer and quadratic programming algorithm employed are computationally efficient and suitable for real-time operation in train braking control units with limited computing power. This method does not require complex offline training or massive historical data support, exhibits strong robustness to changes in vehicle parameters, and effectively overcomes the problem of allocation failure caused by inaccurate empirical adhesion parameters in existing technologies. Attached Figure Description

[0062] Figure 1This is a diagram illustrating the overall architecture of the dynamic power allocation method for axle control of urban rail vehicles based on real-time adhesion estimation as described in this invention.

[0063] Figure 2 This is a principle block diagram of the dynamic power allocation method for axle control of urban rail vehicles based on real-time adhesion estimation as described in this invention.

[0064] Figure 3 This is a flowchart of the dynamic power allocation method for axle control of urban rail vehicles based on real-time adhesion estimation as described in this invention. Detailed Implementation

[0065] The embodiments of the present invention will be described in detail below. These embodiments are based on the technical solutions of the present invention and provide detailed implementation methods and specific operation processes to further explain the technical solutions of the present invention.

[0066] Example 1

[0067] like Figure 1 The diagram shows the overall architecture of the dynamic power distribution method for axle control of urban rail vehicles based on real-time adhesion estimation as described in this invention. The system mainly includes a sensing unit, a braking control unit, a braking execution unit, and the urban rail vehicle as the controlled object. The sensing unit acquires real-time operating status data of the vehicle and includes wheel speed sensors for collecting the wheel angular velocity of each axle, an inertial measurement unit for collecting the longitudinal deceleration of the vehicle, and a load sensor for collecting air spring pressure to calculate the static axle load. The braking control unit, as the core algorithm module of the system, after receiving data from the sensing unit, generates the optimal target braking force command for each axle by combining the overall braking requirements of the vehicle with its internal dynamic axle load calculation module, adhesion estimation module based on an extended state observer, and adaptive weighted braking force optimal allocation module. The braking execution unit is a physical actuator containing a basic braking device corresponding to each axle. After receiving the braking force distribution command from the braking control unit, this unit acts on the basic braking device through a corresponding actuation mechanism to accurately generate the required braking friction force. The actuation mechanism can be an electro-pneumatic conversion valve for adjusting the air brake cylinder pressure or a motor driving the all-electric brake.

[0068] This embodiment is based on a dynamic allocation method for axle control power of urban rail vehicles using real-time adhesion estimation, referencing... Figure 2 , Figure 3 As shown, it includes the following steps:

[0069] Step 1: Obtain real-time operating status data of the vehicle from the sensor module, including vehicle speed, vehicle longitudinal deceleration, wheel-set angular velocity of each axle of the bogie, braking torque, and longitudinal braking force required by the whole vehicle.

[0070] Step 2: Based on the current vehicle operating data and the vehicle's longitudinal dynamics model, calculate the dynamic vertical load during braking. Braking deceleration causes the vehicle body to nod, resulting in axle load transfer. Calculate the dynamic vertical load for each axle. The calculation method is as follows:

[0071] First, based on the dynamic equilibrium of the car body's nose-dive, calculate the vertical load at the center pins of the front and rear bogies. (Front bogie) and (Rear bogie), and the longitudinal force exerted by the car body on the front bogie. and longitudinal forces of the rear bogie :

[0072]

[0073] in, Vehicle mass (including passengers); It is the acceleration due to gravity; This represents the longitudinal deceleration amplitude; The height of the vehicle's center of gravity; This is the center distance between the front and rear bogies; For the first Braking force of the shaft; For the mass of a single bogie. , These represent the front axle and rear axle of the front bogie, respectively. , These represent the front and rear axles of the rear bogie, respectively.

[0074] Subsequently, based on the stress analysis of the bogie, the moment balance equation was solved to obtain the dynamic vertical load of each wheelset. :

[0075]

[0076] in, , These are the dynamic vertical loads on the first axle (front axle) and the second axle (rear axle) of the front bogie, respectively. , These are the dynamic vertical loads of the third (front) axle and the fourth (rear) axle of the rear bogie, respectively. This refers to the bogie wheelbase. The height of the connection point (center pin) between the car body and the bogie from the rail surface; The height of the bogie's center of gravity above the rail surface; This is the internal load redistribution coefficient, used to characterize the contribution ratio of the braking-induced pitching torque to the internal axle load redistribution of the bogie.

[0077] Step 3: Based on the current wheel speed, vehicle speed and braking torque, reconstruct the wheel-rail contact state in real time through the extended state observer, and calculate the theoretical maximum available adhesion coefficient accordingly.

[0078] Based on the wheel-rail adhesion characteristics described by the LuGre friction model, the slip velocity is selected. and internal friction state As a system state variable, it will include unknown wheel-rail contact condition parameters. The nonlinear term is defined as the extended state. , construct the first Augmented state-space model of the axis:

[0079]

[0080] in, The first The sliding speed and internal friction state of the shaft; A function characterizing the nonlinear features of wheel-rail friction; For the first The wheel's angular velocity; These are system parameters related to the vehicle and LuGre model parameters; For the first Braking torque of the shaft, To slow down the vehicle; For the first Axis extension state The derivative; based on this model, a linear extended state observer is designed:

[0081]

[0082] in, For the state estimation vector, The observer gain matrix is... For measurement output; The system coefficient matrix is ​​determined by the augmented state-space model; Given a known input vector containing the vehicle's deceleration; This is a nonlinear disturbance compensation vector, whose elements contain the observer's estimate of the nonlinear frictional characteristics in the LuGre model, used to offset the nonlinear terms in the system's state equations.

[0083] In addition, the wheel-rail adhesion and wheel-rail contact state parameters of each axle are estimated in real time using an extended state observer, and the theoretical maximum usable adhesion coefficient is calculated based on this, including:

[0084] First, the state estimation vector output by the observer is used. The estimated values ​​of each state component in the equation, combined with the definition of the extended state, are used to algebraically solve for the unknown wheel-rail contact condition parameters. Its inverse formula is:

[0085]

[0086] in, The first one is estimated by the observer. Shaft sliding speed, internal friction state, and expansion state; The Stribeck effect function is defined with the estimated slip velocity as the independent variable; the inverse solution process assumes the existence of a small slip. and It is carried out under the following conditions.

[0087] Subsequently, the wheel-rail contact condition parameters obtained by inverse kinematics were used. Reconstruct the current wheel-rail adhesion-slip characteristic curve to obtain the steady-state adhesion coefficient. The parsing expression:

[0088]

[0089] Since the steady-state wheel-rail adhesion force is equal to the steady-state adhesion coefficient The expression, when multiplied by the dynamic vertical load, directly establishes the relationship between the contact state parameters and the macroscopic wheel-rail adhesion. In engineering calculations, the viscous friction term, which has a negligible impact on the peak point, is ignored. By solving the gradient equation of the steady-state adhesion coefficient with respect to the slip velocity... The optimal slip velocity that maximizes the adhesion coefficient was calculated. ;

[0090] Finally, the optimal slip velocity is obtained. Substituting back into the steady-state adhesion coefficient expression, the theoretical maximum usable adhesion coefficient at the current moment can be obtained. ;in, It is the first The sliding speed of the shaft, Let represent the steady-state adhesion coefficient of the i-th axis.

[0091] Step 4: Apply an asymmetric rate of change constraint to the theoretical maximum usable adhesion coefficient to obtain the safe adhesion limit. :

[0092]

[0093] in, This is the original deviation. This represents the safe stickiness limit for the current calculation cycle. This represents the safe adhesion limit of the previous calculation cycle; For the saturation limiting function, its specific operation rules are as follows: when When, the output value is ;when When, the output value is ;when When, the output value is ; and These are the descent rate threshold and ascent rate threshold of the safe adhesion limit, respectively, and satisfy the following conditions: This allows the adhesion limit to decrease rapidly when slippage is detected and increase slowly when adhesion is restored; the final output safe adhesion limit is... , The safety factor is (e.g., 0.98).

[0094] Step 5, refer to Figure 3 As shown, adaptive weighting coefficients for the braking force tracking error target and the adhesion utilization rate target are calculated based on load requirements and adhesion limits.

[0095] First, calculate the global adhesion utilization index. :

[0096]

[0097] in, The total braking force is given by the denominator, which is the total available adhesive capacity of the vehicle. This indicates the extent to which current braking demands are approaching physical limits.

[0098] Then, based on this global adhesion utilization index The weights are dynamically adjusted using the Sigmoid function:

[0099]

[0100] in, As the weight of the braking force tracking target, Weighting for anti-slip safety objectives; For conversion rate parameters, The threshold value is used. Time (sufficient adhesion) The system tends to precisely track braking commands because the risk of skidding is extremely low at this time; when or Time (insufficient adhesion), As the number of wheels increases rapidly, the system tends to minimize adhesion utilization, prioritizing anti-skid safety and preventing wheel lock-up.

[0101] Step 6: Based on the calculated adaptive weights and safe adhesion limit, establish a target including braking force tracking. and safety adhesion utilization target Unified cost function :

[0102]

[0103] Among them, the first item This indicates the braking force tracking error, used to ensure deceleration performance; the second item This represents the sum of squares of the adhesion utilization rates of each axis, used to guide the operating point away from the adhesion saturation region. This is done while satisfying the actuator's maximum braking force constraint ( Under the constraints of the objective function and the rate of change, the optimal braking force for each axis is obtained by solving the minimum value of the objective function using a quadratic programming algorithm. .

[0104] Step 7: Based on the optimal braking force distribution value obtained in Step 6, generate pressure commands for the braking control units of each axle, drive the corresponding brake cylinders to charge and vent, and realize the braking control of the urban rail vehicle.

[0105] This invention inputs vehicle operating status data into an extended state observer to calculate the maximum adhesion boundary in real time and eliminates the impact of time delay through asymmetric limiting. Simultaneously, it uses global adhesion utilization as a scheduling variable and dynamically adjusts the optimization objective through an adaptive weighting mechanism. While meeting braking requirements in real time, it maintains the adhesion utilization of each axle within a safe range, achieving optimal allocation under dynamic axle load transfer and effectively preventing wheel slippage on low-adhesion road surfaces. This eliminates the need for frequent intervention from the backend WSP, protecting the wheel-rail interface.

[0106] Example 2

[0107] This embodiment provides a dynamic power distribution system for axle control of urban rail vehicles based on real-time adhesion estimation, including a sensor module, a braking control unit, and a braking execution unit. The sensor module includes wheel speed sensors, acceleration sensors, and load sensors to collect vehicle speed, wheel angular velocity, longitudinal acceleration, and air spring pressure. The braking control unit is used to: calculate the optimal braking force distribution command for each axle in real time based on the data collected by the sensor module and using the dynamic braking force distribution method for urban rail vehicles based on real-time adhesion estimation described in Embodiment 1. The braking execution unit is used to: receive the optimal braking force distribution command and drive the basic braking device of each axle to generate the corresponding braking force.

[0108] The above embodiments are preferred embodiments of this application. Those skilled in the art can make various changes or improvements based on them. Without departing from the overall concept of this application, these changes or improvements should fall within the scope of protection claimed in this application.

Claims

1. A method for dynamic allocation of axle control power for urban rail vehicles based on real-time adhesion estimation, characterized in that, include: Acquire real-time operating status data of the vehicle, including vehicle speed, vehicle longitudinal deceleration, wheel-to-wheel angular velocity of each axle of the bogie, braking torque, and longitudinal braking force required by the entire vehicle. Based on the vehicle's longitudinal dynamics model, the dynamic vertical loads on each axle during braking are calculated. An extended state observer is constructed based on the vehicle longitudinal dynamics model and the LuGre friction model. The wheel-rail adhesion and wheel-rail contact state parameters of each axle are estimated in real time using the extended state observer, and the theoretical maximum usable adhesion coefficient is calculated accordingly. By imposing an asymmetric rate of change constraint on the theoretical maximum usable adhesion coefficient, a safe adhesion limit for constraint is obtained; A multi-objective optimization function for braking force distribution is constructed, including the braking force tracking error objective and the adhesion utilization rate objective, and the two objectives are weighted by adaptive weight coefficients; Under the premise of satisfying actuator constraints and safety adhesion limit constraints, the multi-objective optimization function is solved to obtain the optimal braking force for each axis.

2. The method for dynamic allocation of axle control power for urban rail vehicles based on real-time adhesion estimation according to claim 1, characterized in that, Specific methods for calculating dynamic vertical loads: First, based on the dynamic equilibrium of the car body's nose-dive, calculate the vertical load at the center pin of the front bogie. Vertical load at the center pin of the rear bogie And the longitudinal force of the car body on the front bogie. and longitudinal forces of the rear bogie : ； in, For vehicle body mass; It is the acceleration due to gravity; This represents the longitudinal deceleration amplitude; The height of the vehicle's center of gravity; This is the center distance between the front and rear bogies; For the first Braking force of the shaft; For the mass of a single bogie; , These represent the front axle and rear axle of the front bogie, respectively. , These represent the front and rear axles of the rear bogie, respectively. Subsequently, based on the vertical load and longitudinal force at the bogie center pin, the dynamic vertical load of each axle and wheelset is obtained according to the moment balance equation. : ； in, , These are the dynamic vertical loads for axes 1 and 2, respectively. , These are the dynamic vertical loads of the 3rd and 4th axes, respectively; This refers to the bogie wheelbase. The height of the connection point between the car body and the bogie from the rail surface; The height of the bogie's center of gravity above the rail surface; This is the internal load redistribution coefficient, used to characterize the contribution ratio of the braking-induced pitching torque to the internal axle load redistribution of the bogie.

3. The method for dynamic allocation of axle control power for urban rail vehicles based on real-time adhesion estimation according to claim 1, characterized in that, The extended state observer is constructed using the following method: Based on the wheel-rail adhesion characteristics described by the LuGre friction model, slip velocity and internal friction state are selected as system state variables, and unknown wheel-rail contact condition parameters are also included. The nonlinear term is defined as the extended state. , construct the first Augmented state-space model of the axis: ； in, The first The sliding speed and internal friction state of the shaft; A function characterizing the nonlinear features of wheel-rail friction; The radius of the wheel; For the first The wheel's angular velocity; For the first The equivalent moment of inertia of the shaft; This is the contact spot deformation distribution coefficient; These are the contact surface bristle stiffness coefficient and bristle damping coefficient in the LuGre friction model, respectively. The overall damping coefficient, and satisfying ,in It is the coefficient of viscous friction; The dynamic coefficients are related to the dynamic vertical load and satisfy the following conditions: ,in For the first Dynamic vertical load on the shaft; For the first Braking torque of the shaft, To slow down the vehicle; For extended state The derivative; Design of linear extended state observers for each axis based on augmented state-space model: ； in, For the first The state estimation vector of the axis, The observer gain matrix is... For the first The measured value of the shaft's sliding speed; The system coefficient matrix is ​​determined by the augmented state-space model; For including vehicle deceleration The known input vector; For the first The nonlinear disturbance compensation vector of the axis.

4. The method for dynamic allocation of axle control power for urban rail vehicles based on real-time adhesion estimation according to claim 1, characterized in that, The wheel-rail adhesion and wheel-rail contact state parameters of each axle are estimated in real time using an extended state observer, and the theoretical maximum usable adhesion coefficient is calculated based on this, including: Using the state estimation vector output by the observer The estimated values ​​of each state component in the equation, combined with the definition of the extended state, are used to algebraically solve for the unknown wheel-rail contact condition parameters. Its inverse formula is: ； in, The first one is estimated by the observer. Shaft sliding speed, internal friction state, and expansion state; The Stribeck effect function is defined with the estimated slip velocity as the independent variable; the inverse solution process assumes the existence of a small slip. and Under the conditions; Subsequently, the wheel-rail contact condition parameters obtained by inverse kinematics were used. Reconstruct the current wheel-rail adhesion-slip characteristic curve to obtain the steady-state adhesion coefficient. The parsing expression: ； Since the steady-state wheel-rail adhesion force is equal to the steady-state adhesion coefficient The expression, when multiplied by the dynamic vertical load, directly establishes the relationship between the contact state parameters and the macroscopic wheel-rail adhesion; in engineering calculations, the viscous friction term is neglected. By solving the gradient equation of the steady-state adhesion coefficient with respect to the slip velocity... The optimal slip velocity that maximizes the adhesion coefficient was calculated. ; Finally, the optimal slip velocity is obtained. Substituting these values ​​into the steady-state adhesion coefficient expression, the theoretical maximum usable adhesion coefficient at the current moment can be obtained. ;in, It is the first The sliding speed of the shaft, Let represent the steady-state adhesion coefficient of the i-th axis.

5. The method for dynamic allocation of axle control power for urban rail vehicles based on real-time adhesion estimation according to claim 1, characterized in that, Imposing an asymmetric rate of change constraint on the theoretically maximum available adhesion coefficient yields the first constraint. Safety adhesion limit of the shaft : ； in, This is the original deviation. This represents the safe stickiness limit for the current calculation cycle. This represents the safe adhesion limit in the previous calculation cycle; For the saturation limiting function, its specific operation rules are as follows: when When, the output value is ;when When, the output value is ;when When, the output value is ; and These are the descent rate threshold and ascent rate threshold of the safe adhesion limit, respectively, and satisfy the following conditions: This allows the adhesion limit to decrease rapidly when slippage is detected and increase slowly when adhesion is restored; the final output safe adhesion limit is... , For safety factors.

6. The method for dynamic allocation of axle control power for urban rail vehicles based on real-time adhesion estimation according to claim 1, characterized in that, A multi-objective function including braking force tracking error target and adhesion utilization rate target. , represented as: ； in, These represent the dynamic tracking error target and the adhesion utilization rate target, respectively. for The corresponding weighting coefficients, This is the longitudinal braking force required for the entire vehicle; For the first Braking force distributed across the shaft, The vector formed by the braking forces assigned to all axes; for The corresponding allocation coefficient vector; For the first Vertical load on the shaft; For the first Safety adhesion limit of the shaft.

7. The method for dynamic allocation of axle control power for urban rail vehicles based on real-time adhesion estimation according to claim 6, characterized in that, The weighting coefficient The adaptive adjustment mechanism is as follows: The global adhesion utilization index is calculated based on the ratio of total braking demand force to total available adhesive capacity of the vehicle. : ； in, For the longitudinal braking force required by the whole vehicle, For the first The safe adhesion limit of the shaft, For the first Vertical load on the shaft; Construct a nonlinear mapping relationship based on the Sigmoid function, according to the global adhesion utilization index. Dynamically calculate the weighting coefficients of the adhesion utilization target : ； in, This is the conversion rate parameter, used to control the sensitivity of weight switching; The global adhesion utilization threshold; Calculate the weighting coefficients of the dynamic tracking error target based on normalization constraints. : 。 8. The method for dynamic allocation of axle control power for urban rail vehicles based on real-time adhesion estimation according to claim 6, characterized in that, The objective function is minimized using a quadratic programming algorithm to obtain the optimal braking force for all axes.

9. A dynamic power distribution system for axle control of urban rail vehicles based on real-time adhesion estimation, characterized in that, It includes a sensing unit, a braking control unit, and a braking actuation unit; The sensing unit is used to: collect real-time vehicle operating status data, including vehicle speed, wheel speed of each axle, total braking demand signal, and axle load signal; The braking control unit is used to: execute the dynamic distribution method of axle control power of urban rail vehicles based on real-time adhesion estimation as described in any one of claims 1-8, and calculate the optimal braking force distribution command for each axle in real time based on the data collected by the sensing unit. The braking actuator is used to: receive the optimal braking force distribution command output by the braking control unit, and generate corresponding braking friction force through the basic braking device corresponding to each shaft.

Images