Train braking power distribution parameter dynamic optimization method based on SSA algorithm

Abstract

This invention provides a dynamic optimization method for train braking power distribution parameters based on the SSA algorithm, relating to the field of train braking control. The method includes the following steps: S1: Obtaining train operating status information; S2: Constructing an objective function for electro-pneumatic composite braking performance; S3: Importing the European standard braking model formula into the objective function; S4: Initializing the parameters of the sparrow search algorithm; S5: Optimizing the electro-pneumatic parameters using the sparrow search algorithm; S6: Executing electro-pneumatic composite braking control based on the optimized electro-pneumatic parameters. This invention uses the sparrow search algorithm to appropriately select the hyperparameters of the braking process for European standard braking trains, improving the algorithm's prediction accuracy and further balancing the braking distance, ride comfort, and energy efficiency indicators of train braking, providing a theoretical basis for optimizing the braking distance of high-speed trains.

Description

Technical Field

[0001] This invention relates to the field of train braking control. Background Technology

[0002] With the rapid development of high-speed railway technology, the operating speed of EMU trains is constantly increasing, which places higher demands on the safety and reliability of train braking systems. Existing EMU trains generally adopt an electro-pneumatic hybrid braking system that combines electric braking and air braking to achieve coordinated output of braking force.

[0003] However, during actual operation of high-speed trains, the braking force distribution between electric and air braking exhibits significant nonlinear characteristics due to variations in train speed, load conditions, and track conditions. Existing braking control methods often employ fixed ratios or empirical rule-based distribution strategies, making it difficult to achieve global optimization of braking performance under complex conditions. This can easily lead to problems such as excessive braking distance or insufficient braking smoothness. Therefore, there is an urgent need for a control method capable of intelligently optimizing electro-pneumatic composite braking parameters based on train operating conditions to improve the overall performance of the high-speed train braking system.

[0004] In existing technology, domestic high-speed trains generally use the European standard braking method for braking control. This principle typically assumes that the actual deceleration varies with speed; initially, at high speeds, the deceleration is relatively small, and as the train speed decreases, the deceleration gradually increases. Since the maximum train deceleration rarely exceeds 1 m / s², this method is generally effective. 2 For ease of calculation and to consider real-time performance, the European standard braking method simplifies the braking curve under the most unfavorable braking conditions into several equivalent straight lines. When using the European standard braking method to divide the braking process, it is generally assumed that the speed range is... Divided into n segments, the corresponding velocity boundary points are respectively And the deceleration within each segment To simplify the deceleration to the most unfavorable braking curve, the braking distances for the corresponding intervals are as follows: s n To take into account braking idle time and initial velocity The additional braking distance is independent of the deceleration setting. Therefore, the braking distance s of the maximum commonly used braking curve SBD consists of two parts: the idle distance and the actual braking distance, as shown in the formula:

[0005] (1) ;

[0006] In the formula, The idle travel distance refers to the time from when the braking command is issued to when the braking force is actually generated. Inside, the train was at its initial speed The distance traveled; the part to the right of the plus sign in the formula is the actual braking distance, which is divided into speed segments, with each segment having a corresponding deceleration. The sum of the calculated braking distances follows the displacement formula for uniformly decelerated motion. . Summary of the Invention

[0007] To overcome the problems of excessive braking distance and insufficient ride comfort in existing electro-pneumatic braking control systems, this invention provides a dynamic optimization method for train braking power distribution parameters based on the SSA algorithm.

[0008] The technical solution adopted by the present invention to achieve the above objectives is as follows:

[0009] A dynamic optimization method for train braking force distribution parameters based on the SSA algorithm is characterized by the following steps:

[0010] S1: Obtain train operation status information

[0011] The operating status information includes, but is not limited to: current operating speed, braking mode, and initial braking parameters, wherein the initial braking parameters include the braking force distribution coefficient between electric braking and air braking, and the electric-air braking switching threshold.

[0012] S2: Constructing the objective function for electro-pneumatic hybrid braking performance

[0013] Based on three performance indicators—braking distance, braking smoothness, and energy efficiency—an objective function f(x) for electro-pneumatic hybrid braking performance is constructed.

[0014] (2) ;

[0015] In the formula, D represents braking distance, P represents ride comfort, E represents energy efficiency, and w1, w2, and w3 are the weighting coefficients of the corresponding indicators.

[0016] S3: Import the European standard braking model into the objective function.

[0017] The braking curve formula from the European standard braking method model is incorporated into the objective function. The formula for the European standard braking method model is as follows:

[0018] (3) ;

[0019] In the formula, D is the braking distance, v is the initial velocity, μ is the braking friction coefficient, g is the gravitational acceleration, and δ is the braking force distribution coefficient;

[0020] S4: Initialize the parameters of the sparrow search algorithm

[0021] Using the braking force distribution coefficient and switching threshold of electric braking and air braking as optimization variables, the population parameters of the sparrow search algorithm are initialized. The corresponding parameters include setting the population size, maximum number of iterations, number of discoverers, number of followers, and proportion of vigilants.

[0022] S5: Optimization of electro-pneumatic parameters using the sparrow search algorithm

[0023] A global search is conducted using a discoverer, follower, and watchdog mechanism to find the optimal combination of braking parameters. In each iteration, the fitness of each sparrow is evaluated based on the objective function value, and the sparrow's position information is updated to ultimately obtain the optimal electro-pneumatic composite braking parameters.

[0024] S6: Perform electro-pneumatic composite braking control based on the optimized electro-pneumatic parameters.

[0025] Based on the above parameter optimization results, the optimal electro-pneumatic combined braking parameters are output, and electro-pneumatic combined braking control is executed according to these optimized parameters to achieve a reasonable ratio of electric braking and air braking, thereby optimizing the train braking process.

[0026] Specifically, in step S4, the optimization variables in the high-speed rail scenario include: the number of intervals (discrete), the boundary speed (real number), and the coefficient (real number). These variables are encoded using a hybrid encoding.

[0027] Among them, binary encoding is used to encode the number of discrete variable intervals, with each binary bit representing the activation status of a speed interval. Bitwise operations are used to quickly adjust the number of intervals. Real number encoding is used to encode the boundary speed and coefficients of continuous variables. It is stored directly in real number form, preserving the precision of the parameters and facilitating subsequent iterative optimization.

[0028] During initialization, the algorithm generates an initial population that meets engineering constraints based on the actual operating data of the high-speed rail line.

[0029] Specifically, the improvements include the following engineering constraints:

[0030] 1. Safety constraints: In the formula S ATP S is the braking distance of the train after the automatic train protection system is activated. train This is the actual braking distance of the train, measured by bench testing.

[0031] 2. Deceleration constraint: Measured range of high-speed rail composite braking: 1.2 m / s² 2 ≤a min,i ≤1.5 m / s 2 0.9≤k i ≤0.95; where a min,iLet k be the minimum deceleration in the i-th velocity range, and k be the reduction factor. An upper limit is set for the reduction factor to prevent insufficient safety redundancy.

[0032] 3. Speed ​​range constraint: 350 = v0 > v1 > v2 > ... > v n =0, the interval number n∈[3,6].

[0033] Specifically, step S5 includes the following sub-steps:

[0034] Step 5-1: Global search by the discoverer, used for speed optimization of section boundaries in high-speed rail scenarios.

[0035] Step 5-1-1: The discoverer generates a new boundary velocity value according to formula (4);

[0036] (4) ;

[0037] In the formula, The j-th dividing velocity is the i-th discoverer in the t-th generation; `T` is a random number that controls the search step size; `T` is the maximum number of iterations of the algorithm.

[0038] Step 5-1-2: Calculate the new interval according to formula (5). This was calculated by the Automatic Train Protection (ATP) system:

[0039] (5) ;

[0040] In the formula, v is the current dividing velocity. , where a is the braking deceleration, and S is the braking deceleration. ATP The braking distance of the train after the automatic train protection system is activated. Then discard the solution directly and reset the discoverer's position to the previous generation's optimal position, where S train This refers to the train's actual braking distance.

[0041] Step 5-1-3: Retain the optimal boundary velocity combination that satisfies the constraints and pass it on to the follower;

[0042] Step 5-2: Local Development of Followers

[0043] Algorithm formula:

[0044] (6) ,

[0045] (7) ;

[0046] In the formula, Optimization variables for followers , The optimal variable for the discoverer `randn()` is a vector of all 1s; `rand()` is a random number following a standard normal distribution; `rand()` is a random number following a uniform distribution in the interval [0,1]; `N` is the correlation parameter involved in the optimization, representing the variables. or The total quantity;

[0047] By setting N and dividing the threshold, the algorithm can balance the efficiency and accuracy of local development, and the control parameter values ​​eventually converge to a better combination of control parameters.

[0048] when i N / 2 represents followers with lower rankings: this indicates that they have poor fitness and will directly generate new solutions based on the discoverer's optimal position to quickly improve performance;

[0049] When i ≤ N / 2, it represents a higher-ranking follower: based on the discoverer's optimal position, it will perform a local search in combination with its own current position to finely discover the optimal solution;

[0050] Suitable for high-speed rail operating conditions:

[0051] Step 5-2-1: The follower fine-tunes the interval boundary points based on the interval determined by the discoverer. The step size range is [0.01-0.05].

[0052] Step 5-2-2: Correct the deceleration value by combining the braking data collected in real time by the vehicle's onboard sensors. ;

[0053] Step 5-2-3: Calculate the adjusted Preserve the parameter combination that makes it smaller and satisfies the constraints;

[0054] Step 5-3: Vigilant hazard monitoring and parameter correction to cope with sudden operating conditions.

[0055] Algorithm formula:

[0056] (8) ,

[0057] (9) ;

[0058] In the formula, x best,j (t) represents the optimal parameters for the population; β∼N(0,1) represents the normal distribution step size factor; f i This represents the current individual objective function value.

[0059] f bestThe optimal fitness value of the population in the current iteration is the optimal result of the objective function for all sparrow individuals, corresponding to the optimal value of the optimization objective "minimum braking distance error";

[0060] x i,j (t+1) represents the j-th optimization variable of the i-th vigilant in the t+1-th generation, and represents the correction parameter (such as braking deceleration rate, boundary speed, etc.) when the working condition changes suddenly.

[0061] In the formula, if f i >f best This indicates that the current braking deceleration rate has deviated from the optimal safety benchmark and needs to be corrected immediately; if f i =f best This indicates that the current parameters are still in a safe and optimal state and do not need to be adjusted.

[0062] Suitable for high-speed rail operating conditions:

[0063] Step 5-3-1: The carriage collects the adhesion coefficient μ at a frequency of 500Hz from the braking node (the watchdog). If μ is detected to decrease from 0.35 to 0.28 (a decrease of more than 20%), emergency braking is immediately triggered from the braking node.

[0064] Step 5-3-2: Adjust the value of a in the corresponding interval according to the correction results of formulas (8) and (9). min,i For example, in the interval [350, 290], a min,i From 1.35 → 1.08 m / s 2 ;

[0065] Step 5-3-3: Synchronize and update a safe Recalculate S ATP Ensure a safety redundancy of ≥5%;

[0066] The above time consumption has been corrected to ≤10ms, meeting the real-time requirements of high-speed rail.

[0067] Step 5-4: Convergence Judgment and Optimal Parameter Output

[0068] Step 5-4-1: Calculate the optimal objective function value for each generation of the population. ;

[0069] Step 5-4-2: If If convergence is determined, the iteration is terminated;

[0070] Step 5-4-3: Output the optimal parameter combination and write it into the vehicle ATP system parameter library.

[0071] The beneficial effects of this invention are as follows: This invention uses SSA (Sparrow Search Algorithm) to appropriately select the hyperparameters of the braking process of European standard braking trains, which improves the prediction accuracy of the algorithm and further balances the braking distance, ride comfort and energy efficiency of the train braking, providing a theoretical basis for the optimization of braking distance of high-speed trains. Attached Figure Description

[0072] Figure 1 This is a comparison chart of braking distances between the traditional European standard braking method and the SSA optimization scheme of this invention.

[0073] Figure 2 This is a deceleration distribution diagram for each velocity range after optimization using the SSA algorithm.

[0074] Figure 3 This is the optimized parameter response curve when the operating conditions change abruptly (adhesion coefficient decreases by 20%).

[0075] Figure 4 This is the SSA algorithm iterative convergence curve (operating conditions of a 350km / h high-speed train). Detailed Implementation

[0076] The present invention will be further explained and described below with reference to the accompanying drawings and embodiments.

[0077] The core of the dynamic optimization method for train braking power distribution parameters based on the SSA algorithm of the present invention is to use the sparrow search algorithm to predict the braking distance under various working conditions.

[0078] The sparrow search algorithm is a swarm intelligence optimization algorithm that simulates the foraging and anti-predation behaviors of sparrows. Its core consists of three types of individuals: discoverers (explorers), followers, and vigilants. Its behavioral logic is highly compatible with the parameter optimization requirements of the high-speed rail braking model (see Table 1 for details).

[0079] Table 1. Mapping relationship between each component of the Sparrow Algorithm and the existing high-speed rail braking model.

[0080]

[0081] Based on the original European standard braking model and with the minimum braking distance of ATP as the objective, the discoverer-follower-watcher mechanism of the Sparrow Search Algorithm (SSA) is deeply integrated with the optimization of high-speed rail braking parameters. Through iterative optimization of the algorithm, the braking distance is minimized while meeting the safety constraints of the European standard.

[0082] Table 2 lists the high-speed rail braking control nodes and the required quantitative indicators for the SSA algorithm in this embodiment.

[0083] Table 2. Mapping relationship between SSA algorithm and control nodes of high-speed rail braking system:

[0084]

[0085] The core optimization logic of the algorithm is to iteratively search for the optimal parameter combination that minimizes braking distance redundancy and satisfies safety constraints in the solution space through global exploration by the discoverer and local development by the follower, while dynamically adjusting parameters under sudden changes in operating conditions through real-time monitoring by the vigilant.

[0086] Specifically, this embodiment includes the following steps:

[0087] Step 1: Population Encoding and Initialization (Binary + Real Number Mixed Encoding)

[0088] In the high-speed rail scenario, the optimization variables include the number of intervals (discrete), the boundary speed (real number), and the coefficients (real number). Multiple variables are encoded using a hybrid approach.

[0089] The binary encoding is used to encode the number of discrete variable intervals. Each binary bit represents the activation status of a speed interval, and the number of intervals can be quickly adjusted through bitwise operations.

[0090] Real number encoding is used to encode the boundary speed and coefficients of continuous variables. It is stored directly in real number form, preserving the precision of the parameters and facilitating subsequent iterative optimization.

[0091] During initialization, the algorithm generates an initial population that meets engineering constraints based on the actual operating data of the high-speed rail line, laying the foundation for subsequent iterations.

[0092] Step 1-1: Determine the optimization objective and constraints for ATP braking distance.

[0093] (1) Core optimization objective: Calculate braking distance using ATP Minimize the objective function, i.e. At the same time, it must meet the safety constraint of the actual braking distance of the train, and the safety redundancy rate (which can be adjusted according to the line grade) must be met. ).

[0094] (2) Engineering constraints include: the speed interval boundary point must meet the high-speed rail speed level; the interval deceleration must meet the vehicle braking system hardware limitations; the algorithm iteration time is ≤500ms (to meet the real-time requirements of onboard ATP).

[0095] Step 1-2: Screening key optimization variables for ATP braking distance

[0096] Extracting the pair from the ATP braking distance calculation model Three core variables that significantly affect safe braking distance are the optimization targets of SSA:

[0097] (1) Velocity interval boundary variable: set as (n is the number of speed intervals - 1), for example, when high-speed rail commonly uses 3 intervals, the boundary variable is... (This is the dividing line between low and medium speeds) This marks the dividing line between medium and high speeds. The value range is [0, 350] km / h;

[0098] (2) Interval deceleration correction coefficient: set as , is the correction factor for the basic deceleration of the European standard braking method, with a value range of [0.8, 1.1] (to ensure that the deceleration is within the hardware allowable range);

[0099] (3) Minimum deceleration threshold in the interval: set as The range of values ​​is This is a core safety parameter for ATP calculations. See [link to ATP braking distance optimization results] for details. Figure 1 .

[0100] Steps 1-3: Design a variable encoding method adapted to SSA

[0101] Step 1-3-1: Individual Dimensions and Segmented Coding

[0102] Individual dimension definition: Each sparrow individual corresponds to a complete set of ATP braking parameters, with dimension D = 3n, where n is the number of velocity ranges. For example, when there are 3 velocity ranges, n = 3, so the individual dimension D = 9, and the parameter sequence is as follows:

[0103]

[0104] in, : Boundary speed, used to divide the speed range into 3 speed ranges; Braking coefficient for each interval; : Minimum braking deceleration rate for each interval.

[0105] The segmented coding structure organizes different types of braking parameters into functional segments, which not only ensures the integrity of the parameters, but also allows the algorithm to clearly identify the physical meaning of each parameter during iteration, facilitating subsequent constraint verification and engineering reconstruction.

[0106] Step 1-3-2: Normalization and Denormalization Processing

[0107] To make the parameters fit the search space of the sparrow search algorithm, we first need to normalize each variable:

[0108] (10) ;

[0109] In the formula, x represents the original engineering parameters. x represents the upper and lower limits of x; x' is the normalized value mapped to the interval [0,1].

[0110] Then perform reverse normalization: After the iteration is complete, use the reverse formula of equation (10) to restore the normalized value to the actual engineering value:

[0111] (11) ;

[0112] The results obtained in this way can be directly applied to the parameter configuration of the ATP system.

[0113] Steps 1-4: Configure the core algorithm parameters of SSA

[0114] The key parameters of the Sparrow Search algorithm are set based on three main requirements in high-speed rail operation (which need to be optimized through trial calculations): Determine the engineering requirements for high-speed rail ATP optimization, including: constraints of step 1 (speed range, deceleration range, iteration time); core optimization objectives (minimize ATP braking distance, safety redundancy rate ≥ 5%); and engineering implementation requirements (such as real-time performance and adaptability to operating conditions). The specific steps are as follows:

[0115] Step 1-4-1: Fix basic parameters: First, preset a set of initial parameters (such as population size 20, maximum number of iterations 50, role ratio 2:7:1, α=0.5).

[0116] Step 1-4-2: Adjust control variables one by one: Keep other parameters unchanged, only adjust the population size, and test values ​​of 10, 20, 30, and 40. Test each set of parameters 5 times, and record the average iteration time and the optimal value of the objective function (ATP braking distance).

[0117] Step 1-4-3: Select the optimal combination: Select parameters that have "short iteration time and good objective function optimization effect" (in the high-speed rail scenario, the overall performance is best when the population size is 20~30).

[0118] Step 1-4-4: Scenario adaptation and correction: Combine the specific scenario constraints (the high-speed rail ATP requires an iteration time of ≤500ms) and fine-tune the parameters (set the maximum number of iterations to 50~80).

[0119] Steps 1-5: Initialization of the sparrow population

[0120] Based on the above variable range and encoding rules, an initial population is generated using a combination of random initialization and Latin hypercube sampling.

[0121] Step 1-5-1: First, generate 70% of the initial individuals within the variable range using Latin hypercube sampling to ensure the homogeneity of the population;

[0122] Step 1-5-2: Randomly generate another 30% of the individuals to increase population diversity;

[0123] Step 1-5-3: Perform constraint verification on the generated initial individuals, remove individuals that do not meet the hard constraints, and regenerate until the population size reaches the set value N.

[0124] Step 2: Iterative Optimization Process

[0125] Step 2-1: Global search by the discoverer (corresponding to the speed optimization of section boundaries in the high-speed rail scenario)

[0126] (12) ;

[0127] In the formula, The j-th dividing velocity is the i-th discoverer in the t-th generation; is a random number that controls the search step size; T is the maximum number of iterations (total number of iterations) of the algorithm.

[0128] High-speed rail adaptation operation:

[0129] 1. The discoverer generates a new boundary velocity value according to formula (12);

[0130] 2. Calculate the new interval. The formula is calculated by the Automatic Train Protection (ATP) system:

[0131] (13) ;

[0132] In the formula, v is the current dividing velocity. , where a is the braking deceleration, and S is the braking deceleration. ATP The braking distance of the train after the automatic train protection system is activated. Then discard the solution directly and reset the discoverer's position to the previous generation's optimal position, where S train This is the actual braking distance of the train;

[0133] 3. Retain the optimal boundary velocity combination that satisfies the constraints and pass it on to the follower.

[0134] Step 2-2: Local Development of Followers

[0135] Algorithm formula:

[0136] (14) ,

[0137] (15) ;

[0138] In the formula, Optimization variables for followers , The optimal variable for the discoverer. It is a vector of all 1s; N corresponds to the correlation parameters involved in the optimization, representing the variables. or The total number.

[0139] By setting N and dividing the threshold, the algorithm can balance the efficiency and accuracy of local development, allowing different followers to adopt different update strategies, and finally converge to a better combination of control parameters.

[0140] when i N / 2 (representing followers ranked lower): This indicates that their fitness is poor, and they will directly generate new solutions based on the discoverer's optimal position to quickly improve performance.

[0141] When i ≤ N / 2 (representing a higher-ranked follower): it will perform a local search based on the discoverer's optimal position and its own current position to finely discover the optimal solution.

[0142] High-speed rail adaptation operation:

[0143] 1. Followers fine-tune the interval boundaries determined by the discoverers. ;

[0144] 2. Correct the deceleration value by combining braking data collected in real time by onboard sensors. ;

[0145] 3. Calculate the adjusted... , retain Smaller parameter combinations that satisfy the constraints.

[0146] Steps 2-3: Hazard monitoring and parameter correction by vigilant personnel (responding to sudden changes in operating conditions)

[0147] Algorithm formula:

[0148] (16) ,

[0149] (17) ;

[0150] In the formula, x best,j (t) represents the optimal parameters for the population; β∼N(0,1) represents the normal distribution step size factor; f i This represents the current individual objective function value.

[0151] f best It represents the optimal fitness value of the population in the current iteration. It is the optimal result of the objective function for all sparrow individuals and corresponds to the optimal value of optimization objectives such as "minimizing braking distance error".

[0152] x i,j(t+1) represents the j-th optimization variable of the i-th vigilant in the t+1-th generation, which represents the correction parameter (such as braking deceleration rate, boundary speed, etc.) when the working condition changes abruptly.

[0153] In the formula, if f i >f best This indicates that the current parameters (such as braking deceleration rate) have deviated from the optimal safety benchmark and need to be corrected immediately. If f i =f best This indicates that the current parameters are still in a safe and optimal state and do not need to be adjusted.

[0154] High-speed rail adaptation operation (taking a 20% decrease in adhesion coefficient in rainy or snowy weather as an example):

[0155] 1. The carriage collects the adhesion coefficient μ at a frequency of 500Hz from the node (watcher). When it detects that μ changes from 0.35 to 0.28 (a decrease of 20%), it immediately triggers the anti-predation behavior in the sparrow search algorithm.

[0156] 2. Adjust 'a' in the corresponding interval according to the formula. min,i For example, in the interval [350, 290], a min,i From 1.62 → 1.35 m / s 2 ;

[0157] 3. Synchronously update a safe That is, the safe deceleration value, recalculate S ATP Ensure a safety redundancy of ≥5%;

[0158] 4. Correction time ≤ 10ms, meeting the real-time requirements of high-speed rail.

[0159] After adjusting the deceleration value according to formulas (16) and (17), the a value in each velocity range is... i For details on the (deceleration) distribution, please refer to [link / reference]. Figure 2 .

[0160] For the parameter response curve of high-speed rail during sudden changes in operating conditions (taking a 20% decrease in viscosity coefficient during rain or snow as an example), please refer to [reference needed]. Figure 3 .

[0161] Steps 2-4: Convergence Judgment and Optimal Parameter Output

[0162] 1. Calculate the optimal objective function value for each generation of the population. ;

[0163] 2. If If convergence is determined, the iteration is terminated;

[0164] 3. Output the optimal parameter combination and write it into the vehicle-mounted ATP system parameter library.

[0165] For the iterative convergence curve of the SSA algorithm (taking 350km / h as an example), please refer to... Figure 4 .

[0166] Key details of engineering adaptation:

[0167] Integrating SSA into the final stage of ATP braking distance optimization hinges on achieving safe output, efficient transmission, dynamic adaptation, and fault fallback of optimized parameters. Specifically, this includes:

[0168] 1. Optimal parameter set selection and output: After the algorithm converges, the parameters are selected based on dry conditions, rain / snow, strong winds, and slope (uphill gradient). downhill slope Typical working conditions such as these are selected to meet the requirements. Multiple sets of braking parameters are combined under safety redundancy constraints. Taking a 350km / h high-speed train as an example, the speed range boundary point under dry operating conditions is set as... Corresponding deceleration reduction factor Minimum deceleration The minimum deceleration under rain and snow conditions is adjusted to This forms a working condition-parameter mapping table.

[0169] 2. Low-latency transmission and writing: 5G-R communication technology is adopted, with a transmission rate set to ≥100Mbps and end-to-end transmission latency controlled to ≤20ms. After the parameter set is transmitted to the on-board ATP system, it is verified by CRC-32 to ensure data integrity, and then written to the "optimized parameter area" of the braking parameter library. The writing time is ≤5ms. Within 10ms after the transmission is completed, a "parameter update successful" signal is fed back to the trackside server to form a closed loop.

[0170] 3. Dynamic adaptation to multiple working conditions: Supports both manual selection and automatic recognition modes. Drivers can manually switch working conditions through the HMI interface. At the same time, the system collects data such as viscosity coefficient and wind speed at a frequency of 500Hz through on-board sensors. When the viscosity coefficient is ≤0.2 or the wind speed is ≥25m / s, the corresponding optimized parameters are automatically called. The switching time is ≤10ms, ensuring that the braking parameters are matched in real time when the working conditions change.

[0171] 4. Actual vehicle verification and calibration: Actual vehicle braking tests were conducted on straight tracks, 20‰ uphill slopes, and 30‰ downhill slopes, as well as under conditions such as dry (adhesion coefficient 0.35) and rain / snow (adhesion coefficient 0.28). The ATP braking distance under optimized parameters was recorded as 4463.5m, and the actual braking distance of the train was recorded as 4250m. The safety redundancy reached 5.02%, and the braking response time was ≤150ms. After fine-tuning the parameter k_2 to 0.90 for insufficient braking smoothness in a certain section, the smoothness index was improved by 12%, and an ATP report was generated.

[0172] 5. Operation and Maintenance Iterative Updates: The trackside server collects over 100,000 braking operation data points from the onboard ATP every month. Analysis shows that optimized parameters reduce braking distance by an average of 3.2%, and the safety redundancy compliance rate is 99.8%. When the addition of wheel-rail materials causes the adhesion coefficient to increase to 0.40, the SSA algorithm is re-run to generate a new parameter set, which is then updated remotely via 5G-R. The update time is ≤30 seconds, and the parameters from the last 3 versions are retained to support fault rollback.

[0173] 6. Fault-safe fallback: When the SSA algorithm times out (>100ms) or fails to converge, the onboard ATP automatically switches to traditional European standard parameters (e.g., average deceleration of 0.95m / s² in the [80,0]km / h range) within 10ms; if 5G-R transmission is interrupted, the currently effective parameters are used, and parameter synchronization is completed within 1s after transmission is restored; if the operating condition sensor fails, it switches to conservative parameters for rain and snow conditions by default, and simultaneously sends audible and visual alarms to the driver and uploads fault codes to the dispatch center to ensure train braking safety.

[0174] Tables 3 and 4 are quantitative verification tables for the high-speed rail scenario of 350km / h after applying the solution of the present invention; Table 5 is a comparison table of the optimization effect of this embodiment.

[0175] Table 3 Basic Parameter Table

[0176]

[0177] Table 4 Key data during the SSA algorithm iteration process

[0178]

[0179] Table 5 Comparison of Optimization Results

[0180]

[0181] This invention has been described through embodiments. Those skilled in the art will understand that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the invention. Furthermore, under the teachings of this invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the invention. Therefore, this invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the protection scope of this invention.

Claims

1. A dynamic optimization method for train braking force distribution parameters based on the SSA algorithm, characterized in that, Includes the following steps: S1: Obtain train operation status information The operating status information includes, but is not limited to: current operating speed, braking mode, and initial braking parameters, wherein the initial braking parameters include the braking force distribution coefficient between electric braking and air braking, and the electric-air braking switching threshold. S2: Constructing the objective function for electro-pneumatic hybrid braking performance Based on three performance indicators—braking distance, braking smoothness, and energy efficiency—an objective function f(x) for electro-pneumatic hybrid braking performance is constructed. （1） ； In the formula, D represents braking distance, P represents ride comfort, E represents energy efficiency, and w1, w2, and w3 are the weighting coefficients of the corresponding indicators. S3: Import the European standard braking model into the objective function. The braking curve formula from the European standard braking method model is incorporated into the objective function. The formula for the European standard braking method model is as follows: （2） ； In the formula, D is the braking distance, v is the initial velocity, μ is the braking friction coefficient, g is the gravitational acceleration, and δ is the braking force distribution coefficient; S4: Initialize the parameters of the sparrow search algorithm Using the braking force distribution coefficient and switching threshold of electric braking and air braking as optimization variables, the population parameters of the sparrow search algorithm are initialized. The corresponding parameters include setting the population size, maximum number of iterations, number of discoverers, number of followers, and proportion of vigilants. S5: Optimization of electro-pneumatic parameters using the sparrow search algorithm A global search is conducted using a discoverer, follower, and vigilant mechanism to find the optimal combination of braking parameters; In each iteration, the fitness of each sparrow is evaluated based on the objective function value, and the sparrow's position information is updated to finally obtain the optimal electro-pneumatic combined braking parameters. S6: Perform electro-pneumatic composite braking control based on the optimized electro-pneumatic parameters. Based on the above parameter optimization results, the optimal electro-pneumatic combined braking parameters are output, and electro-pneumatic combined braking control is executed according to these optimized parameters to achieve a reasonable ratio of electric braking and air braking, thereby optimizing the train braking process.

2. The method for dynamic optimization of train braking force distribution parameters according to claim 1, characterized in that: In step S4, the optimization variables include: number of intervals: represented by discrete values; boundary speed: represented by real numbers; coefficients: represented by real numbers. The above variables use mixed encoding: Among them, binary encoding is used to encode the number of discrete variable intervals, with each binary bit representing the activation status of a speed interval. Bitwise operations are used to quickly adjust the number of intervals. Real number encoding is used to encode the boundary speed and coefficients of continuous variables. It is stored directly in real number form, preserving the precision of the parameters and facilitating subsequent iterative optimization. During initialization, the algorithm generates an initial population that meets engineering constraints based on the actual operating data of the high-speed rail line.

3. The method for dynamic optimization of train braking force distribution parameters according to claim 2, characterized in that: The engineering constraints include: (1) Safety constraints: In the formula S ATP S is the braking distance of the train after the automatic train protection system is activated. train This is the actual braking distance of the train, measured by bench testing. (2) Deceleration constraint: The measured range of high-speed rail composite braking is 1.2 m / s 2 ≤a min,i ≤1.5 m / s 2 0.9≤k i ≤0.95; where a min,i Let k be the minimum deceleration in the i-th velocity range, and k be the reduction factor. (3) Speed ​​range constraint: 350 = v0 > v1 > v2 > ... > v n =0, the interval number n∈[3,6].

4. The method for dynamic optimization of train braking force distribution parameters according to claim 1, characterized in that: Step S5 includes the following sub-steps: Step 5-1: Global search by the discoverer, used for optimizing the speed at the section boundaries in high-speed rail scenarios. Step 5-1-1: The discoverer generates a new dividing velocity value according to formula (3); （3） ； In the formula, The j-th dividing velocity is the i-th discoverer in the t-th generation; `T` is a random number that controls the search step size; `T` is the maximum number of iterations of the algorithm. Step 5-1-2: Calculate the new interval according to formula (4). This is calculated by the Automatic Train Protection (ATP) system: （4） ； In the formula, v is the current dividing velocity. , where a is the braking deceleration, and S is the braking deceleration. ATP The braking distance of the train after the automatic train protection system is activated. Then discard the solution directly and reset the discoverer's position to the previous generation's optimal position. In the formula, S... train This is the actual braking distance of the train; Step 5-1-3: Retain the optimal boundary velocity combination that satisfies the constraints and pass it on to the follower; Step 5-2: Local Development of Followers Algorithm formula: （5） ， （6） ； In the formula, Optimization variables for followers , The optimal variable for the discoverer is a vector of all 1s; randn() is a random number following a standard normal distribution; rand() is a random number following a uniform distribution in the interval [0,1]; N is the correlation parameter involved in the optimization, representing the variables. or The total quantity; By setting N and dividing the threshold, the algorithm can balance the efficiency and accuracy of local development, and the control parameter values ​​eventually converge to a better combination of control parameters. Suitable for high-speed rail operating conditions: Step 5-2-1: The follower fine-tunes the interval boundary points based on the interval determined by the discoverer. The step size range is [0.01-0.05]. Step 5-2-2: Correct the deceleration value by combining the braking data collected in real time by the vehicle's onboard sensors. ; Step 5-2-3: Calculate the adjusted Preserve the parameter combination that makes it smaller and satisfies the constraints; Step 5-3: Vigilant hazard monitoring and parameter correction to cope with sudden operating conditions. Algorithm formula: （8） ， （9） ； In the formula, x best,j (t) represents the optimal parameters for the population; β∼N(0,1) represents the normal distribution step size factor; f i This represents the current individual objective function value. f best The optimal fitness value of the population in the current iteration is the optimal result of the objective function for all sparrow individuals, corresponding to the optimal value of the optimization objective "minimum braking distance error"; x i,j (t+1) represents the j-th optimization variable of the i-th vigilant in the t+1-th generation, which represents the correction parameters represented by braking deceleration rate and boundary speed when the corresponding working condition changes abruptly. In the formula, if f i >f best This indicates that the current braking deceleration rate has deviated from the optimal safety benchmark and needs to be corrected immediately; if f i =f best This indicates that the current parameters are still in a safe and optimal state and do not need to be adjusted. Suitable for high-speed rail operating conditions: Step 5-3-1: The carriage starts from the braking node, which is the watchdog in the sparrow algorithm, and collects the adhesion coefficient μ at a frequency of 500Hz. If the detection shows that μ drops by more than 20%, the emergency braking of the braking node is immediately triggered. Step 5-3-2: Adjust the value of a in the corresponding interval according to the correction results of formulas (8) and (9). min,i ; Step 5-3-3: Synchronize and update a safe That is, the safe deceleration value, and then recalculate S. ATP Ensure a safety redundancy of ≥5%; Correcting the above variables takes ≤10ms; Step 5-4: Convergence Judgment and Optimal Parameter Output Step 5-4-1: Calculate the optimal objective function value for each generation of the population. ; Step 5-4-2: If If convergence is determined, the iteration is terminated; Step 5-4-3: Output the optimal parameter combination and write it into the vehicle ATP system parameter library.

Images