Train energy-saving operation control method and system based on optimization algorithm

Abstract

The invention discloses a train energy-saving operation control method and system based on an optimization algorithm, and aims to significantly reduce the train operation energy consumption and improve the operation efficiency through optimizing a target speed curve and precise speed tracking control. According to the method, firstly, an energy-saving target speed curve is generated through an atomic orbit search and particle swarm hybrid optimization algorithm (AOS-PSO), and constraint conditions such as traction energy consumption, running time deviation, passenger comfort and parking precision are comprehensively considered in combination with a multi-target optimization model. Secondly, a combined controller based on an extended state observer (ESO) and sliding mode control (SMC) is designed, external disturbance is estimated and compensated in real time, and the precision and stability of speed tracking are ensured. The method has the advantages of being high in global optimization capacity, high in dynamic adaptability, remarkable in energy-saving effect and the like, is suitable for complex urban rail transit scenes, and provides an effective solution for train energy-saving operation.

Description

Technical Field

[0001] The present invention relates to the technical field of rail transit control, and particularly to a train energy-saving operation control method and system based on an optimization algorithm. Background Art

[0002] Traditional train control methods mainly aim at punctuality and safety, but there are the following problems:

[0003] 1. High energy consumption: Fixed driving curves are difficult to adapt to complex road conditions, and redundant acceleration or braking leads to energy waste.

[0004] 2. Poor dynamic adaptability: Piecewise optimization strategies are prone to falling into local optima and it is difficult to globally coordinate the energy consumption of multi-section operations.

[0005] 3. Insufficient tracking accuracy: Existing speed tracking control methods (such as PID) are sensitive to disturbances, prone to overshoot and chattering, which affect the energy-saving effect.

[0006] In the prior art, multi-objective optimization methods usually convert to a single objective through weighting, but the weight selection depends on experience and lacks the dynamic adaptability to actual operating conditions (such as gradients, speed limits). In addition, the chattering problem in speed tracking control limits the actual application effect of the energy-saving optimization curve. Therefore, there is an urgent need for an energy-saving control method that combines global optimization and high-precision tracking. Summary of the Invention

[0007] The object of the present invention is to provide a train energy-saving operation control method and system based on an optimization algorithm, which generates an energy-saving target speed curve through a hybrid optimization algorithm, and combines disturbance compensation and robust control strategies to achieve precise tracking. On the premise of ensuring the punctual operation of the train, through optimizing the target speed curve and precise speed tracking control, the energy consumption is significantly reduced and the operation efficiency is improved, thereby ensuring the operation stability and passenger comfort while reducing the energy consumption.

[0008] The technical solution for achieving the object of the present invention is as follows:

[0009] A train energy-saving operation control method based on an optimization algorithm, comprising:

[0010] Step 1: Real-time collect train operation data, including position, speed, line gradient, speed limit information and external disturbances;

[0011] Step 2: Construct a multi-objective optimization model with the minimum traction energy consumption and the minimum operation time deviation as the objectives, and the constraint conditions include the maximum speed, braking distance, passenger comfort and parking accuracy;

[0012] Step 3: Solve the multi-objective optimization model by using an atomic orbital search and particle swarm hybrid optimization algorithm to generate a global energy-saving target speed curve as the standard value;

[0013] Step 4: Design a combined controller based on the extended state observer and sliding mode control to achieve the tracking of the target speed curve;

[0014] Step 5: Take the real-time collected data as feedback, and based on the combined controller, update the control command every Δt through the rolling horizon optimization strategy to adapt to the real-time environmental changes.

[0015] Furthermore, the data acquisition and preprocessing in the above Step 1 include:

[0016] 5) Obtain the real-time position and speed information of the train through in-vehicle GPS or track sensors;

[0017] 6) Extract the line information of gradients, curve radii, and speed limit sections from the line database;

[0018] 7) Monitor external disturbances in real time through sensors;

[0019] 8) Clean, denoise, and normalize the collected data.

[0020] Furthermore, the objective function of the multi-objective optimization model in the above Step 2 is:

[0021]

[0022] where: v i is the speed of the train at the i-th Δt; F(v i ) is the traction force of the train when the speed is v i , and Δt is the time interval.

[0023] Furthermore, the constraint conditions of the multi-objective optimization model in the above Step 2 include:

[0024] (1) Operation safety constraint

[0025]

[0026] where: v(x0) is the speed of the train at the starting point; v(x f ) is the speed of the train at the end point; v max (x) is the speed limit of the train at this position;

[0027] (2) Arrival on-time constraint

[0028]

[0029] where: t B is the actual arrival time of the train; is the planned arrival time of the train; Δt B is the allowable time error range;

[0030] (3) Parking accuracy constraint

[0031]

[0032] Where: x(t stop ) is the position of the train when it stops; x station is the platform position; v(t stop ) is the speed of the train when it stops; ε max is the maximum value of the parking error;

[0033] (4) Passenger comfort constraint:

[0034]

[0035] Where: a(t) is the acceleration of the train at time t; j(t) is the jerk of the train at time t; v(t) is the speed of the train at time t; a max is the maximum allowable acceleration threshold; j max is the maximum allowable jerk threshold; Δv max is the maximum allowable speed change.

[0036] Furthermore, the hybrid optimization algorithm in step 3 specifically includes the steps of:

[0037] 6) Initialize the particle swarm and randomly generate the positions and velocities of the particles;

[0038] 7) Calculate the fitness value of each particle and update the individual optimal solution and the global optimal solution;

[0039] 8) Update the positions and velocities of the particles through the particle swarm optimization algorithm for global search;

[0040] 9) Perform local fine optimization of the candidate solutions generated by the particle swarm optimization using the atomic orbital search and dynamically adjust the positions of the candidate solutions;

[0041] 10) Determine whether the convergence condition is satisfied. If it is satisfied, output the optimal solution; otherwise, return to step 2).

[0042] Furthermore, the update formula of the particle swarm optimization algorithm is:

[0043]

[0044] Where: v i (t) is the velocity of the i-th particle at the t-th iteration; x i(t) is the current position of the i-th particle at the t-th iteration; w is the inertia weight, controlling the retention and decay of the particle velocity; c1 and c2 are learning factors; r1 and r2 are random numbers within the range of [0, 1].

[0045] Further, the atomic orbital search algorithm dynamically adjusts the candidate solution position as follows:

[0046]

[0047] In the formula: is the position of the i-th candidate solution in the k-th layer before update; is the position of the i-th candidate solution in the k-th layer after update; LE is the candidate solution with the lowest energy in the system; BS is the bound state of the atom; α i , β i , γ i is a random number vector uniformly distributed in the range of (0, 1).

[0048] Further, step 4 specifically includes:

[0049] Step 4.1: Real-time estimate the system disturbance through an extended state observer and compensate it;

[0050] Step 4.2: Construct a sliding mode surface and adopt an improved exponential reaching law to suppress chattering;

[0051] Step 4.3: Generate a control command by combining the observed values of the extended state observer and dynamically adjust the traction / braking force.

[0052] Further, the rolling horizon optimization strategy in step 5 includes:

[0053] Update the control command every Δt, and Δt is dynamically adjusted according to the train position, and is shortened to 5 seconds when approaching the station;

[0054] Through real-time data acquisition and optimization algorithms, dynamically adjust the target speed curve and control command to adapt to the complex and changeable operating environment.

[0055] A train energy-saving operation control system based on an optimization algorithm, comprising:

[0056] An acquisition unit that real-time acquires train operation data, including position, speed, track gradient, speed limit information, and external disturbances;

[0057] A multi-objective optimization model construction unit that constructs a multi-objective optimization model with the objectives of minimizing traction energy consumption and minimizing running time deviation, and the constraint conditions include maximum speed, braking distance, passenger comfort, and parking accuracy;

[0058] The controller design unit designs a combined controller based on an extended state observer and sliding mode control to achieve the tracking of the target speed curve;

[0059] The control unit updates the control command every Δt through a rolling horizon optimization strategy to adapt to the real-time environment changes.

[0060] Compared with the prior art, the remarkable advantages of the present invention are as follows:

[0061] (1) Strong global optimization ability: The AOS-PSO hybrid optimization algorithm is adopted to improve the optimization effect of the target speed curve; (2) High tracking accuracy: The ESO-SMC controller effectively reduces the speed deviation and improves the operation stability; (3) Strong robustness: It can adapt to different gradients, speed limit conditions and external disturbances, improving the applicability of the train's energy-saving operation; (4) By combining the optimization algorithm and the advanced control method, the present invention improves the energy-saving effect of the train operation, while ensuring the stability of speed tracking, providing theoretical support and practical reference for the energy-saving dispatching of rail transit. Brief Description of the Drawings

[0062] Figure 1 is the overall flowchart of the method of the present invention;

[0063] Figure 2 is the schematic diagram of the AOS-PSO hybrid optimization algorithm process;

[0064] Figure 3 is the block diagram of the ESO-SMC controller structure;

[0065] Figure 4 is the output target speed curve graph;

[0066] Figure 5 is the speed error comparison graph between ESO-SMC and the traditional control method. Detailed Embodiments

[0067] Combined with Figure 1 , the present invention provides a train energy-saving operation control method based on an optimization algorithm, aiming to significantly reduce the train operation energy consumption through optimizing the target speed curve and precise speed tracking control, while ensuring the train's punctuality rate, passenger comfort and operation safety. The following will describe in detail the specific embodiments of the present invention, including steps such as data acquisition, optimization model construction, algorithm implementation, controller design and simulation verification.

[0068] 1. Data Acquisition and Preprocessing

[0069] The first step in train energy-saving operation control is to collect in real time the relevant data of train operation. These data include but are not limited to:

[0070] 1) Train position and speed: Obtain the real-time position and speed information of the train through on-board GPS or track sensors.

[0071] 2) Track information: Including gradient, curve radius, speed limit sections, etc. This information is usually stored in the train's track database.

[0072] 3) External disturbances: Such as wind resistance, wheel-rail adhesion-slip force, etc. These disturbances will affect the running state of the train.

[0073] 4) Train parameters: Including train mass, traction characteristics, braking characteristics, etc. These parameters are the basis for constructing the train dynamics model.

[0074] After data collection, preprocessing is required, including data cleaning, denoising, and normalization, to ensure the accuracy and consistency of the data.

[0075] 2. Construct a multi-objective optimization model

[0076] In order to generate an energy-saving target speed curve, the present invention constructs a multi-objective optimization model. The objective function is to minimize the traction energy consumption, and the constraint conditions include running safety, arrival punctuality, parking accuracy, and passenger comfort.

[0077] 2.1 Objective function

[0078] Minimize traction energy consumption:

[0079] Traction energy consumption is the main energy consumption source during train operation. The calculation formula for traction energy consumption is:

[0080]

[0081] In the formula: v i —— The speed of the train at the i-th Δt (km / h);

[0082] F(v i )—— The traction force of the train at a speed of v i (kN).

[0083] The constraint conditions are:

[0084] (1) Running safety

[0085] The running safety of the train is the primary consideration index. Once the train exceeds the speed limit, the ATP system will activate the overspeed protection mechanism and automatically trigger emergency braking. When the train is in automatic driving, it must strictly abide by the section speed limit to avoid overspeed. The expression of the running safety constraint is as follows:

[0086]

[0087] Where: v(x0) —— the speed of the train at the starting point;

[0088] v(x f ) —— the speed of the train at the end point;

[0089] v max (x) —— the speed limit of the train at this position.

[0090] (2) Arrival punctuality

[0091] Arrival punctuality is an important indicator of train operation efficiency and passenger experience. The ATS system ensures that trains arrive at each station according to the planned time by monitoring and dispatching the running status of trains in real time. If a train fails to reach its destination within the specified time window, it will not only affect the coordination of the overall operation plan but also cause passengers to be late. Therefore, strict time constraints must be observed during the train operation to ensure punctual arrival. The expression of the arrival punctuality constraint is as follows:

[0092]

[0093] Where: t B —— the actual arrival time of the train;

[0094] —— the planned arrival time of the train;

[0095] Δt B —— the allowable time error range.

[0096] (3) Stopping accuracy

[0097] Stopping accuracy is crucial for the efficient operation of urban rail transit, directly related to the safety of passengers and the efficiency of getting on and off the train. Under the control of the ATO system, the train can achieve accurate stopping by dynamically adjusting the speed and braking to ensure that the train stops at the predetermined platform position. If the stopping position error exceeds the allowable range, it will not only affect the access of passengers but also pose a risk to the normal operation of platform facilities such as platform screen doors. According to relevant regulations, the train stopping error under non-special conditions should be controlled within ±30 cm. The expression of the arrival punctuality constraint is as follows:

[0098]

[0099] Where: x(t stop ) —— the position of the train when it stops;

[0100] x station —— the platform position;

[0101] v(t stop ) —— the speed of the train when it stops;

[0102] ε max ——The maximum value of the parking error.

[0103] (4) Passenger comfort

[0104] During the train operation, the passenger comfort is one of the important considerations in operation optimization. Excessive acceleration or jerk (the rate of change of acceleration) will bring discomfort to passengers and even affect their safety. To ensure that the passenger riding experience meets international standards, the comfort of train operation should meet the requirements of the ISO-2631 standard, which determines the comfort limit values for human exposure to different vibration environments by evaluating the sensitivity of the human body to acceleration and vibration. Therefore, during operation, the train should strictly control the changes in acceleration and jerk, and avoid sudden speed fluctuations and frequent working condition switches to ensure a smooth feeling for passengers during the journey. The expression of the passenger comfort constraint and the evaluation index table are as follows:

[0105]

[0106] In the formula: a(t) —— The acceleration of the train at time t;

[0107] j(t) —— The jerk of the train at time t;

[0108] v(t) —— The speed of the train at time t;

[0109] a max —— The maximum allowable acceleration threshold;

[0110] j max —— The maximum allowable jerk threshold;

[0111] Δv max —— The maximum allowable speed change.

[0112] Table 1 Passenger comfort evaluation index table

[0113]

[0114] 3. AOS-PSO hybrid optimization algorithm to generate an energy-saving target speed curve.

[0115] Combined with Figure 2 , to solve the above multi-objective optimization model, the present invention proposes a hybrid optimization algorithm based on Atomic Orbital Search (AOS) and Particle Swarm Optimization (PSO). This algorithm combines the global search ability of PSO and the local fine optimization ability of AOS, and can find the global optimal solution in complex multi-objective optimization problems.

[0116] 3.1 Particle Swarm Optimization (PSO)

[0117] The PSO algorithm searches for the optimal solution by simulating the foraging behavior of a bird flock and using the search of the particle swarm in the solution space. Each particle represents a candidate solution. The particle swarm optimization (PSO) is used for global exploration, and the update formula is:

[0118]

[0119] In the formula: v i (t) —— The velocity of the i-th particle at the t-th iteration;

[0120] x i (t) —— The current position of the i-th particle at the t-th iteration;

[0121] w —— Inertia weight, controlling the maintenance and decay of the particle velocity;

[0122] c1, c2 —— Learning factors;

[0123] r1, r2 —— Random numbers within the range of [0, 1].

[0124] 3.2 Atomic Orbital Search (AOS)

[0125] The AOS algorithm dynamically adjusts the position of the candidate solution by simulating the transition behavior of electrons in atomic orbitals. The atomic orbital search algorithm (AOS) is used for local development, and the candidate solution update rule is:

[0126]

[0127] In the formula: —— The position of the i-th candidate solution in the k-th layer before update;

[0128] —— The position of the i-th candidate solution in the k-th layer after update;

[0129] LE —— The candidate solution with the lowest energy in the system;

[0130] BS —— The binding state of the atom;

[0131] α i , β i , γ i —— A random number vector uniformly distributed in the range of (0, 1).

[0132] 3.3 Hybrid Optimization Process

[0133] 1) Initialization: Randomly generate a particle swarm and initialize the positions and velocities of the particles.

[0134] 2) Fitness evaluation: Calculate the fitness value of each particle and update the individual optimal solution and the global optimal solution.

[0135] 3) PSO global search: Update the positions and velocities of particles through the PSO algorithm for global search.

[0136] 4) AOS local optimization: Perform local fine optimization on the candidate solutions generated by PSO and dynamically adjust the positions of the candidate solutions.

[0137] 5) Convergence detection: Determine whether the convergence condition is satisfied. If it is satisfied, output the optimal solution; otherwise, return to step 2.

[0138] 4. ESO-SMC Controller Design

[0139] Combined with Figure 3 , in order to accurately track the energy-saving target speed curve, the present invention designs a combined controller based on an extended state observer (ESO) and sliding mode control (SMC) to achieve the tracking of the target speed curve. Among them, the extended state observer (ESO) is used to estimate the system disturbance in real time and compensate; the sliding mode control (SMC) constructs a sliding mode surface to make the system state slide along the sliding mode surface. The specific tracking process includes:

[0140] Step 4.1: Estimate the system disturbance in real time through ESO and compensate.

[0141] Step 4.2: Construct a sliding mode surface and adopt an improved exponential reaching law to suppress chattering.

[0142] Step 4.3: Generate a control command by combining the ESO observation value and dynamically adjust the traction / braking force.

[0143] The dynamic equation of the extended state observer (ESO) is:

[0144]

[0145] where x1 is the current position of the train;

[0146] x2 is the current speed of the train;

[0147] F u —— The train control input (traction force or braking force);

[0148] m is the total mass of the train;

[0149] d(t) is the external unknown disturbance and satisfies |d(t)| ≤ D, where the constant D > 0;

[0150] a, b, c are the basic resistance coefficients during train operation;

[0151] y is the system output value.

[0152] The sliding mode control law is:

[0153]

[0154] Step 5: Take the real-time collected data as feedback. Based on the combined controller, update the control instruction every Δt through the rolling horizon optimization strategy to adapt to the real-time environmental changes.

[0155] The present invention generates an energy-saving target speed curve through the AOS-PSO hybrid optimization algorithm, and combines with the ESO-SMC controller to achieve precise tracking, significantly reducing the train operation energy consumption. It is applicable to urban rail transit trains, improving the operation efficiency and passenger comfort. This method has the advantages of strong global optimization ability, strong dynamic adaptability, and significant energy-saving effect, is applicable to complex urban rail transit scenarios, and has broad application prospects.

[0156] This embodiment also provides a train energy-saving operation control system based on an optimization algorithm, including:

[0157] An acquisition unit that collects train operation data in real time, including position, speed, line gradient, speed limit information, and external disturbances;

[0158] A multi-objective optimization model construction unit that constructs a multi-objective optimization model with the minimum traction energy consumption and the minimum operation time deviation as the objectives, and the constraint conditions include maximum speed, braking distance, passenger comfort, and parking accuracy;

[0159] A controller design unit that designs a combined controller based on the extended state observer and sliding mode control to achieve the tracking of the target speed curve;

[0160] A control unit that updates the control instruction every Δt through the rolling horizon optimization strategy to adapt to the real-time environment change.

[0161] The present invention will be further described below in conjunction with the embodiments and the drawings.

[0162] Embodiment

[0163] To verify the effectiveness of the present invention, the following takes a certain urban rail transit line as an example to describe in detail the specific implementation process and effect of the present invention. The line is about 33 kilometers long, contains 18 stations, the train is a 6-car B-type train, and the maximum running speed is 80 km / h. The simulation experiment is based on the actual operation data of this line, including line gradient, speed limit information, train traction characteristics, braking characteristics, etc.

[0164] Taking a certain subway line as an example, the train travels from Station A to Station B, and the line contains 3 gradient sections and 2 curves. After implementing the method of the present invention:

[0165] 1) Energy consumption: The energy consumption per single interval decreases from 14.34 kW·h to 12.75 kW·h, with a decrease of 11.09%.

[0166] 2) Tracking error: The speed error range reduces from ±0.3 m / s to ±0.18 m / s.

[0167] 3) Computational efficiency: The time taken for a single optimization is 50 ms, meeting the requirements of real-time control.

[0168] The simulation diagrams are as Figure 4 and Figure 5 , and from the information on the line gradient changes in Figure 4 it can be seen that this section of the track successively passes through downhill sections with gradients of 0‰, -3.7‰, and -4.5‰ from the starting point, and then enters uphill sections with gradients of 1.8‰ and 2.9‰. To minimize energy consumption, after optimization by the AOS-PSO algorithm, the train adopts a specific energy-saving operation strategy in this interval: At the start, the train accelerates under the maximum traction condition until it approaches the speed limit of 75 km / h. Subsequently, the train enters the downhill section and switches to the coasting condition, where the potential energy is converted into kinetic energy to maintain the running speed. On the relatively flat section around 200 meters to 500 meters, the train adopts the cruise condition to maintain a constant speed. When entering the downhill section at around 800 meters, the train switches to the coasting condition again to reduce speed and save energy. Finally, when approaching the end point, the train switches to the maximum braking condition to decelerate. This energy-saving optimization strategy makes full use of the changes in the line gradient, maximizes the utilization of potential energy and kinetic energy, enables the train to better utilize kinetic energy and potential energy during operation, reduces traction energy consumption, and at the same time improves the comfort experience of passengers. By analyzing Figure 5 it can be seen that the PID and SMC algorithms have relatively large errors during the process of tracking the target speed curve. For example, when the train is at around 224 meters, after the train completes traction acceleration and switches to the coasting condition, there is a large speed gap between the actual speed curve and the energy-saving target speed curve. Similarly, there are also relatively large errors at the switching points between cruise and coasting, and between coasting and braking. In contrast, the ESO-SMC controller can make the actual target speed curve accurately approach the optimized energy-saving target speed curve throughout the entire operation process, and can also maintain a good control effect at the switching points of operating conditions.

[0169] The simulation results show that this method reduces the energy consumption per single interval by 11.09% and the overall energy consumption of multiple intervals by 2.93%, and controls the speed tracking error within ±0.18 m / s, significantly superior to traditional control methods.

[0170] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the appended claims.

Claims

1. A train energy-saving operation control method based on an optimization algorithm, characterized in that, Including: Step 1: Real-time collect train operation data, including position, speed, line gradient, speed limit information and external disturbances; Step 2: Construct a multi-objective optimization model with the objectives of minimizing traction energy consumption and minimizing running time deviation, and the constraint conditions include maximum speed, braking distance, passenger comfort and parking accuracy; Step 3: Use a hybrid optimization algorithm combining atomic orbital search and particle swarm optimization to solve the multi-objective optimization model, and generate a global energy-saving target speed curve as the standard value; Step 4: Design a combined controller based on an extended state observer and sliding mode control to achieve tracking of the target speed curve; Step 5: Take the real-time collected data as feedback, and based on the combined controller, through a rolling horizon optimization strategy, update the control instruction every Δt to adapt to the real-time environmental changes.

2. The method according to claim 1, wherein The data collection and preprocessing in Step 1 include: 1) Obtain the real-time position and speed information of the train through on-vehicle GPS or track sensors; 2) Extract the gradient, curve radius, and speed limit interval line information from the line database; 3) Real-time monitor external disturbances through sensors; 4) Clean, denoise and normalize the collected data.

3. The method according to claim 1, wherein The objective function of the multi-objective optimization model in Step 2 is: where: v i is the speed of the train at the i-th Δt; F(v i ) is the traction force of the train when the speed is v i , and Δt is the time interval.

4. The method according to claim 3, wherein The constraint conditions of the multi-objective optimization model in Step 2 include: (1) Operation safety constraint Where: v(x0) is the speed of the train at the starting point; v(x f ) is the speed of the train at the ending point; v max (x) is the speed limit of the train at this position; (2) Arrival punctuality constraint where: t B is the actual arrival time of the train; is the planned arrival time of the train; Δt B is the allowable time error range; (3) Parking accuracy constraint where: x(t stop ) is the position of the train when it stops; x station is the platform position; v(t stop ) is the speed of the train when it stops; ε max is the maximum value of the parking error; (4) Passenger comfort constraint: Where: a(t) is the acceleration of the train at time t; j(t) is the jerk of the train at time t; v(t) is the speed of the train at time t; a max is the maximum allowable acceleration threshold; j max is the maximum allowable jerk threshold; Δv max is the maximum allowable speed change amount.

5. The method according to claim 1, wherein The specific steps of the hybrid optimization algorithm in Step 3 include: 1) Initialize the particle swarm, and randomly generate the positions and velocities of the particles; 2) Calculate the fitness value of each particle, and update the individual optimal solution and the global optimal solution; 3) Update the positions and velocities of the particles through the particle swarm optimization algorithm for global search; 4) Perform local fine optimization of the atomic orbital search on the candidate solutions generated by the particle swarm optimization, and dynamically adjust the positions of the candidate solutions; 5) Judge whether the convergence condition is satisfied. If satisfied, output the optimal solution; otherwise, return to Step 2).

6. The method according to claim 5, characterized in that The update formula of the particle swarm optimization algorithm is: Where: v i (t) is the velocity of the i-th particle at the t-th iteration; x i (t) is the current position of the i-th particle at the t-th iteration; w is the inertia weight, controlling the maintenance and decay of the particle velocity; c1 and c2 are learning factors; r1 and r2 are random numbers within the range of [0, 1].

7. The method according to claim 5, wherein The dynamic adjustment of the candidate solution position by the atomic orbital search algorithm is: Wherein: is the position of the ith candidate solution in the kth layer before update; is the position of the ith candidate solution in the kth layer after update; LE is the candidate solution with the lowest energy in the system; BS is the bound state of the atom; α i , β i , γ i is a random number vector uniformly distributed in the range of (0, 1).

8. The method according to claim 1, wherein Step 4 specifically includes: Step 4.1: Real-time estimate and compensate the system disturbance through an extended state observer; Step 4.2: Construct a sliding mode surface and adopt an improved exponential reaching law to suppress chattering; Step 4.3: Generate a control instruction by combining the observation value of the extended state observer, and dynamically adjust the traction / braking force.

9. The method according to claim 1, characterized in that, The rolling horizon optimization strategy in Step 5 includes: Update the control instruction every Δt, and Δt is dynamically adjusted according to the train position, and is shortened to 5 seconds when approaching the station; Dynamically adjust the target speed curve and control instruction through real-time data collection and optimization algorithm to adapt to the complex and changeable operation environment.

10. A train energy-saving operation control system based on an optimization algorithm for implementing the method according to any one of claims 1-9, characterized in that, Including: A collection unit that real-time collects train operation data, including position, speed, line gradient, speed limit information and external disturbances; A multi-objective optimization model construction unit that constructs a multi-objective optimization model with the objectives of minimizing traction energy consumption and minimizing running time deviation, and the constraint conditions include maximum speed, braking distance, passenger comfort and parking accuracy; A controller design unit that designs a combined controller based on an extended state observer and sliding mode control to achieve tracking of the target speed curve; The control unit updates the control instructions every Δt through a rolling horizon optimization strategy to adapt to the real-time environmental changes.

Images