Virtual marshalling train energy-saving optimization method, equipment and product

Abstract

The invention discloses a virtual marshalling train energy-saving optimization method, device and product, and relates to the technical field of virtual marshalling train energy saving.The method comprises the steps that line data and historical train operation data of trains of different models are obtained, and train operation energy consumption of the trains of different models is calculated; based on a virtual marshalling technology and a mixed integer programming method, according to the train operation energy consumption, constructing a train operation diagram model; the train working diagram model comprises constraint conditions and a target function, the constraint conditions comprise a train timetable constraint, a passenger flow volume constraint and a train coupling constraint, and the target function aims at balancing passenger waiting time and train operation energy consumption; solving the train working diagram model to obtain a train working diagram; and according to the train historical operation data, the train working diagram and the line data, adopting a double-depth Q network algorithm to generate an optimal train operation speed curve. Optimization of train energy consumption is realized by applying a virtual marshalling technology, a mixed integer programming method and reinforcement learning.

Description

Technical Field

[0001] The present application relates to the field of energy-saving technology for virtual train formation, and in particular to a method, device and product for optimizing energy saving for virtual train formation. Background Art

[0002] The rail transit system is the foundation for the survival and function of the entire urban system. Urban rail transit is increasingly valued by countries around the world for its advantages such as high speed, large capacity, safety and energy saving. At present, my country's urban rail transit network is developing rapidly. With the rapid expansion of the urban rail transit network, the problem of energy consumption has become increasingly severe and has become a key challenge that needs to be solved urgently. Therefore, how to effectively improve energy efficiency and reduce energy consumption has become a key issue in the optimization of the rail transit system, which is of great significance to the sustainable development of the city and the achievement of the carbon emission peak target.

[0003] There are many factors that make up the energy consumption of train operation, but they can be generally divided into three categories: traction energy consumption, auxiliary energy consumption and station energy consumption. In practical applications, traction energy consumption occupies a dominant position in the energy consumption of train operation. In view of the key role of traction energy consumption in train operation energy consumption and its operability, the aim is to adjust the train operation strategy by optimizing traction energy consumption to achieve energy-saving effects.

[0004] In order to meet the needs of train energy conservation, there are usually two solutions: one is to optimize the train speed curve, and the other is to optimize the train operation schedule. Both of these have been hot topics in the rail transit industry in recent years. Although the above studies provide good theoretical support for the energy-saving operation of urban rail transit, these studies are currently based on fixed marshaling, that is, only large trains or small trains are used to run the line. In recent years, with the rapid development of intelligent technology, virtual marshaling technology has gradually been introduced into the field of rail transit. Virtual marshaling technology can use large trains and small trains to run the line at the same time, which is one of the key means to improve train operation efficiency and line capacity. On the basis of virtual marshaling technology, how to further optimize the operating energy consumption of trains, reduce energy consumption and improve energy efficiency has become an important issue in the current research of the rail transit field. Summary of the invention

[0005] The purpose of this application is to provide a virtual marshaling train energy-saving optimization method, equipment and product, which can optimize the energy consumption of trains by using virtual marshaling technology, mixed integer programming method and reinforcement learning.

[0006] To achieve the above objectives, this application provides the following solutions:

[0007] In a first aspect, the present application provides a virtual train marshaling energy-saving optimization method, comprising:

[0008] Acquire line data and historical train operation data of trains of different models, and calculate the train operation energy consumption of trains of different models according to the line data and the historical train operation data of trains of different models; the line data includes the distance between each station, the slope and the train speed limit, and the historical train operation data includes the current position, speed, acceleration and running resistance of the train;

[0009] Based on virtual marshaling technology and mixed integer programming method, a train operation diagram model is constructed according to the train operation energy consumption; the train operation diagram model includes constraints and an objective function, the constraints include train schedule constraints, passenger flow constraints and train coupling constraints, and the objective function is a function with the goal of balancing passenger waiting time and the train operation energy consumption;

[0010] Solving the train operation diagram model to obtain a train operation diagram;

[0011] According to the historical train operation data, the train operation diagram and the line data, a double-depth Q network algorithm is used to generate an optimal train operation speed curve.

[0012] In a second aspect, the present application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the energy-saving optimization method for virtual train formation described above.

[0013] In a third aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the energy-saving optimization method for virtual train formation described above.

[0014] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0015] The present application provides a method, device and product for energy-saving optimization of virtual marshaling trains. The present application optimizes the train operation diagram through a mixed integer programming method, which can balance the waiting time of passengers and the energy consumption of the train while ensuring the timetable constraints. Moreover, the present application optimizes the train operation speed curve according to the historical operation data of the train and the train operation diagram by using a dual-depth Q network algorithm to minimize energy consumption and meet other constraints. Reinforcement learning can handle dynamically changing environments to ensure that the train operation speed is always optimal. It can be seen that the accurate calculation and optimization of train energy consumption is achieved through the comprehensive use of virtual marshaling technology, mixed integer programming methods and deep reinforcement learning. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 This is an application environment diagram of a virtual train marshaling energy-saving optimization method in one embodiment of the present application;

[0018] Figure 2 A schematic diagram of a flow chart of a virtual train energy-saving optimization method provided in one embodiment of the present application;

[0019] Figure 3 It is a schematic diagram of the functional modules of the dual-depth Q network;

[0020] Figure 4 is a schematic diagram of the optimal train speed curve;

[0021] Figure 5 Schematic diagram of fixed marshaling coupling and virtual marshaling coupling strategies with different weight ratios;

[0022] Figure 6a Schematic diagram of passenger waiting time performance of fixed marshaling coupling and virtual marshaling coupling strategies in non-peak hours under different instances;

[0023] Figure 6b Schematic diagram of train energy consumption of fixed marshaling coupling and virtual marshaling coupling strategies in off-peak hours under different instances;

[0024] Figure 7a Schematic diagram of passenger waiting time performance of fixed marshaling coupling and virtual marshaling coupling strategies during peak hours under different instances;

[0025] Figure 7b Schematic diagram of train energy consumption under fixed marshaling coupling and virtual marshaling coupling strategies during peak hours under different instances;

[0026] Figure 8 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0028] The train operation diagram model includes three parts: trains, passenger flow and energy consumption. The optimization of the train operation diagram not only needs to consider basic conditions such as stations, lines, and undercarriage resources, but also involves multiple constraints such as marshaling operations, energy consumption limits, safe intervals, and dynamic passenger flows. The introduction of virtual marshaling strategies is likely to bring nonlinear constraints and expand the scale of constraints, which will bring huge challenges to model solving. These complex constraints are often intertwined. How to strike a balance between these conditions during optimization to ensure efficient train operation while ensuring passenger experience and system safety is a major challenge in model solving. In addition, how to maintain the accuracy of the model while ensuring computational efficiency is also one of the difficulties that need to be overcome.

[0029] The passenger flow of urban rail transit has significant dynamic spatiotemporal characteristics, and passenger flow fluctuations have a direct impact on train scheduling and marshaling decisions. In actual operation, the passenger flow in different time periods and at different stations varies greatly. How to dynamically respond to passenger flow demand in the operation diagram optimization and reasonably adjust the train marshaling form is also one of the core difficulties of model optimization. At the same time, the uncertainty of passenger flow increases the complexity of the model, requiring both rapid response and ensuring the stability and robustness of the model during optimization.

[0030] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0031] The energy-saving optimization method for virtual train formation provided in the embodiment of the present application can be applied to Figure 1In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the line data and the historical train operation data of different models of trains to the server 104, and the server 104 calculates the train operation energy consumption of different models of trains based on the line data and the historical train operation data of different models of trains; based on the virtual marshaling technology and the mixed integer programming method, a train operation diagram model is constructed according to the train operation energy consumption; the train operation diagram model is solved to obtain the train operation diagram; and then the double-depth Q network algorithm is used to generate the optimal train operation speed curve based on the historical train operation data, the train operation diagram and the line data. The server 104 can feedback the obtained optimal train operation speed curve to the terminal 102. In addition, in some embodiments, the energy-saving optimization method for virtual train formation can also be implemented independently by the server 104 or the terminal 102. For example, the terminal 102 can directly process the line data and the historical train operation data of trains of different models, or the server 104 can obtain the line data and the historical train operation data of trains of different models from the data storage system and process them.

[0032] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, and portable wearable devices of the Internet of Things. The server 104 may be implemented as an independent server or a server cluster composed of multiple servers, or may be a cloud server.

[0033] In an exemplary embodiment, Figure 2 As shown, a virtual train marshaling energy-saving optimization method is provided. The method is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or can be executed by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used as an example to illustrate, including the following steps 201 to 204. Among them:

[0034] Step 201, obtain line data and historical train operation data of different models of trains, and calculate the train operation energy consumption of different models of trains based on the line data and the historical train operation data of different models of trains; the line data includes the distance between each station, the slope and the train speed limit, and the historical train operation data includes the train's current position, speed, acceleration and running resistance.

[0035] The wheel-rail force sensor, traction force sensor and aerodynamic resistance sensor are integrated on the train to calculate the running resistance of the train under different speeds and conditions. In addition, the acceleration sensor, speed sensor and positioning device installed on the train can collect the acceleration, speed and position data series of the train. After defining the time interval as a specific value, after sorting, the train historical operation data table of the train formation within M time steps can be obtained as follows:

[0036] Table 1. Train historical operation data table

[0037]

[0038] Step 202, based on virtual marshaling technology and mixed integer programming (Mixed Integer Programming, MIP), a train operation diagram model is constructed according to the train operation energy consumption; the train operation diagram model includes constraints and objective functions, the constraints include train schedule constraints, passenger flow constraints and train coupling constraints, and the objective function is a function with the goal of balancing passenger waiting time and train operation energy consumption.

[0039] Step 203, solving the train operation diagram model to obtain the train operation diagram.

[0040] Step 204, based on the historical train operation data, the train operation diagram and the line data, a dual deep Q-network (DQN) algorithm is used to generate an optimal train operation speed curve.

[0041] By implementing the above steps 201 to 204, the present application optimizes the train operation diagram through a mixed integer programming method, which can balance the waiting time of passengers and the energy consumption of the train while ensuring the timetable constraints. Moreover, the present application optimizes the train operation speed curve according to the historical operation data of the train and the train operation diagram by using a dual-depth Q network algorithm to minimize energy consumption and meet other constraints. Reinforcement learning can handle dynamically changing environments to ensure that the train operation speed is always optimal. It can be seen that through the comprehensive use of virtual marshaling technology, mixed integer programming methods and deep reinforcement learning, accurate calculation and optimization of train energy consumption are achieved.

[0042] Further, the calculation of the train operation energy consumption of different models of trains according to the line data and the historical train operation data of different models of trains involved in step 201 is described in detail, specifically, the train operation energy consumption of different models of trains is calculated according to the following formula:

[0043]

[0044] Among them, v i (0,ξ m ) is the mass at station i, ξm The speed of the m-type train at t = 0; The time spent at the station When it arrives at station i+1, its mass is ξ m The speed of the m-type train; v i (t,ξ m ) is the speed of the train at station i at time t; is the running time of train k at station i; m is the mass of the m-type train; V i min (t) is the minimum speed limit at station i at time t; S is the set of stations; M is the set of virtual marshaling train types; K is the set of train numbers.

[0045]

[0046] Where v(t+1) is the speed of the train at time t+1; x(t+1) is the distance the train travels at time t+1; v(t) is the speed of the train at time t; x(t) is the distance the train travels at time t; λ B (t,ξ m ) is the ratio of the train's braking force output per unit time; λ F (t,ξ m ) is the ratio of the train's output traction force per unit time; B[v i (t,ξ m )] is the speed of the m-type train at station i, which is v i (t,ξ m ) under braking force; r[v i (t,ξ m )] is the speed of the m-type train at station i, which is v i (t,ξ m ) basic resistance under g i (t) is the slope of station i at time t, i.e., the gradient force; F[v i (t,ξ m )] is the speed of the m-type train at station i, which is v i (t,ξ m ) under traction.

[0047]

[0048] r(v)=a1+a2v+a3v 2 ;

[0049] Wherein, F(v) is the traction force when the speed of the train is v; B(v) is the braking force when the speed of the train is v; is the maximum traction; is the maximum traction power; v z Breaking speed limits for traction; is the maximum braking force; is the maximum braking power; v b is the breaking speed limit of braking; r(v) is the basic resistance when the train speed is v, including friction resistance and air resistance; a1, a2, a3 are the Davis equation coefficients representing the train parameter characteristics.

[0050]

[0051] Among them, E m is the total energy consumption of the train, which is composed of the cumulative traction energy consumption generated between stations; x is the running distance of the train between stations.

[0052] Furthermore, the train schedule constraints in step 202 include train section running time constraints and train arrival and departure time constraints.

[0053] The train interval running time constraints are as follows:

[0054]

[0055] in, is the departure time of train k+1 at station i; is the departure time of train k at station i; T min It is the safe departure time between consecutive trains.

[0056] The time constraints for train arrival and departure are as follows:

[0057]

[0058] in, is the arrival time of train k at station i; is the departure time of train k at station i-1; is the running time of train k departing from station i-1.

[0059] Furthermore, the train coupling constraints in step 202 are as follows:

[0060]

[0061] Among them, y k,m is a decision variable, ensuring that the train can only choose one type of formation.

[0062] Furthermore, the passenger flow constraints in step 202 include the number of passengers getting off the train, the number of waiting passengers, the remaining passenger capacity, the number of passengers getting on the train, and the number of passengers on the train.

[0063] The passengers getting off are:

[0064]

[0065] Among them, pa k,i are passengers getting off train k at station i, pr k,i-1 are the remaining passengers who did not board train k at station i-1; A i is a ratio.

[0066] The number of waiting passengers is:

[0067]

[0068] Among them, pw k,i is the number of passengers waiting for train k at station i; pw k-1,i is the number of passengers who failed to board train k-1 at station i; pb k-1,i is the number of passengers arriving at station i after the previous train k-1 left; is the remaining passenger getting-off rate in the car at station i at time t; is a slack variable. When train k departs at [t, t+1], it is the offset of the departure time from station i relative to t. If train k does not depart, its value is 0. is a slack variable. When train k-1 departs at [t, t+1], it is the offset of the departure time from station i relative to t. If train k-1 does not depart, its value is 0. is a binary variable, which is 1 if train k departs from station i at time t, otherwise it is 0; is a binary variable, which is 1 if train k-1 departs from station i at time t, and 0 otherwise.

[0069] The remaining passenger capacity is:

[0070]

[0071] Among them, pc k,i is the number of passengers that train k can accommodate at station i, i.e., the train capacity; C m is the initial no-load capacity of train k.

[0072] The number of passengers on board is:

[0073] pb k,i =min{pw k,i ,pc k,i},i∈S,k∈K;

[0074] Among them, pb k,i is the number of passengers boarding train k at station i.

[0075] The number of passengers on the train is:

[0076]

[0077] Further, the objective function in step 202 includes a first sub-objective function and a second sub-objective function.

[0078] The first sub-objective function is:

[0079]

[0080] The second sub-objective function is:

[0081]

[0082] The objective function is:

[0083] F=w1·F1+w2·F2;

[0084] Among them, w1 and w2 are the weights of the dual-objective optimization method.

[0085] Furthermore, in step 203, the train operation diagram model is solved to obtain the train operation diagram, and a Gurobi solver or a CPLEX solver may be used, which supports mixed integer programming (MIP) and is applicable to the train operation diagram problem.

[0086] Further, the process involved in step 204 is described in detail, wherein the optimal train speed curve is generated by using a double-depth Q network algorithm according to the historical train operation data, the train operation diagram and the line data. Figure 3 As shown, specifically including:

[0087] Step 2041, based on the historical train operation data, train operation diagram, line data and the obtained reward function, a train operation reinforcement learning environment is constructed; the distance between each station, the slope and the train speed limit are used as the static data of the line, and the current position, speed, acceleration and running resistance of the train are used as the dynamic data of the train operation. During the train operation process, the intelligent agent interacts with the train operation reinforcement learning environment, and the environment generates a new train operation state, reward value and state value function to feed back to the intelligent agent. The intelligent agent continuously evaluates and improves the strategy through the value function, selects the maximum action value function, feeds back the maximum action value to the train operation reinforcement learning environment, and continuously updates the operating condition value through such a closed-loop structure, and finally selects the optimal operating condition action to generate the optimal train speed curve.

[0088] In this embodiment, the state set of the train is set to:

[0089]

[0090] Among them, x represents the current position of the train, and Xi represents the distance between the two stations. This ratio represents the relative position of the train at the current moment; v represents the speed of the train at the current moment, V i min represents the minimum permissible speed of train i. This ratio reflects the relative speed of the train at the current moment to its minimum permissible speed, which can indicate whether the train is running at a low speed; t represents the current moment, and T p V stands for the planned running time, which usually refers to the time required for a train to run between two adjacent stations as planned. This ratio represents the current relative time and can reflect the position of the train on the timeline. i represents the train speed limit, and V max Represents the maximum permissible speed of the train. This ratio reflects the relative speed of the train's current speed relative to its maximum permissible speed, and can indicate whether the train is running at high speed or close to its speed limit.

[0091] The action set of the train is set to:

[0092] a={0,1,2,3};

[0093] Among them, 0, 1, 2, and 3 represent traction, cruising, coasting, and braking, respectively, which are the possible output actions of the intelligent agent (i.e., the on-board controller).

[0094] The reward function of the train is set as:

[0095]

[0096] At each current state of the train, the train operation reinforcement learning environment will give a reward value for the action selected by the agent to evaluate the quality of the action. Among them, w3·ΔT represents the punctuality reward, which reflects the difference between the train arrival time and the target time. ΔT represents the difference between the actual arrival time of the train and the target arrival time (which may be positive or negative). When the train arrives ahead of time, ΔT is negative; when the train arrives late, ΔT is positive. The w3 weight is a positive number, so when the train deviates from the target time, whether it is early or late, it will get a negative reward (i.e., penalty). Represents energy-saving bonus, which reflects the rate of change of energy consumption during the operation of the train. It represents the rate of change of energy consumption over time, which can be positive or negative (but usually energy consumption is increasing, so this term is usually negative). The weight w4 is a positive number, so when energy consumption increases, a negative reward (i.e., penalty) will be obtained. If the train takes energy-saving measures that result in a reduction in energy consumption (although this may be less common in actual situations), a positive reward will be obtained. stands for comfort bonus, which reflects the change in train acceleration, that is, the rate of change of acceleration, which affects the comfort of passengers. represents the absolute value of the rate of change of acceleration, and dt is the time interval. i -x) represents the position accuracy reward, which reflects the position accuracy of the train in a specific section. i represents the target position of the train in interval i, x represents the current position of the train, and the sign of the weight w6 depends on the optimization goal: if you want the train to be rewarded when it is close to the target position, w6 is positive; if you want the train to be punished when it is far away from the target position, w6 is negative. Here w6 will be set to a positive number to encourage the train to maintain an accurate position in a specific interval.

[0097] Step 2042: construct a train speed curve optimization network based on the double-depth Q network.

[0098] Step 2043: Train the train speed curve optimization network based on the train operation reinforcement learning environment to obtain a trained train speed curve optimization network.

[0099] Furthermore, the train speed curve optimization network includes an evaluation network and a target network; the determination process of the trained train speed curve optimization network is:

[0100] Initialize the evaluation network and the target network, which is the process of giving initial values ​​to the weights and biases in the deep neural network or policy model. Good network parameter initialization is crucial to the convergence speed and final performance of the algorithm. Through proper initialization, the gradient vanishing or exploding problem can be avoided and the stability of the training process can be improved. Xavier initialization is used to select a suitable initial value range according to the number of layers of the network and the characteristics of the activation function to ensure that the signal can be effectively transmitted during forward propagation and back propagation.

[0101] Based on the train operation reinforcement learning environment, the current state of the train is determined; the current state includes the current relative position, the current first relative speed, the current relative time and the current second relative speed.

[0102] The current state is input into the evaluation network to determine the current action; the evaluation network calculates the expected future return (i.e., Q value) of each action based on the current state and action list, and then selects an action to execute based on the greedy strategy (or ε-greedy strategy), which is usually the action with the highest Q value (or randomly selects an action with a certain probability for exploration).

[0103] According to the current action, the train operation reinforcement learning environment is used to calculate the current reward value and determine the state at the next moment.

[0104] The current moment reward value and the next moment state are input into the target network to calculate the target Q value. According to the current moment state, current moment action, target Q value and loss function, the evaluation network and target network are updated; specifically, the evaluation network predicts a Q value based on the current state and the selected action, and the loss function is calculated using the difference between the target Q value and the Q value predicted by the evaluation network, as follows:

[0105] yloss=E[(r+γmaxQ(s,b';θ-)-Q(s,b;θ)) 2 ];

[0106] Among them, r is the reward value, γ is the discount factor, b represents all actions predicted by the DQN network, b' represents all actions output by the DQN network, θ represents the parameters used by the DQN network for prediction, and θ - represents the parameters of the target network, and Q(s,b;θ) is the Q value predicted by the evaluation network.

[0107] Repeat the above steps until a predetermined number of iterations is reached or the loss function converges to a predetermined threshold, and a trained train speed curve optimization network is obtained.

[0108] Step 2044, based on the trained train speed curve optimization network, obtain the optimal train speed curve, such as Figure 4 shown.

[0109] To evaluate the performance of the energy-saving optimization method for virtual train formation, three sets of experiments were conducted. The first set of experiments compared the fixed coupling strategy and the virtual formation coupling strategy. The second set of experiments studied the impact of the weight coefficient in the objective function to understand the trade-off between energy consumption and service quality. The last set of experiments evaluated the impact of passenger demand changes on the performance of the energy-saving optimization method for virtual train formation.

[0110] Evaluation of different train formation numbers: The performance of the fixed coupling strategy and the virtual marshaling coupling strategy are compared. Specifically, a total of 20 instances are constructed, and the number of trains K increases from 14 to 52. For the fixed coupling strategy, a long coupled train formation is used; while for the virtual marshaling coupling strategy, different types of train formations are allowed. In each instance, Gurobi (version 10.0.1) is used to solve the model to achieve the optimal solution (the error does not exceed 0.5%).

[0111] Evaluation of different train adjustment strategies: In the objective function of the train operation diagram model, the weight factors w1 and w2 represent the contribution of total delay and train operation cost, respectively. In order to study the impact of different strategies on train coupling, the number of trains is set to 14 and the weight ratio w 1 / w2 is set to 0.05, 0.06, 0.07, 0.10, 0.14, 0.25, 0.5, and 1 respectively. Figure 5 The performance of fixed and virtual marshaling coupling strategies in these eight examples is shown, including passenger waiting time and overall energy consumption of the train. Since train formations have different preferences in performance, different train adjustment strategies need to be adopted according to limited resources and passenger demand.

[0112] Evaluation of different passenger flow demands: In order to quantitatively evaluate the effect of different passenger demands, a total of 20 instances are constructed, and the number of trains K increases from 14 to 52. In each instance, we set w1 = 1 and w2 = 10 in the objective function. Figures 6 and 7 show the performance of fixed marshaling coupling and virtual marshaling coupling strategies under different passenger demands (off-peak and peak periods, respectively) in these 20 instances, including passenger waiting time and train energy consumption.

[0113] The present application also provides an application scenario, which applies the above-mentioned virtual marshaling train energy-saving optimization method. Specifically: the virtual marshaling train energy-saving optimization method provided in this embodiment can be applied in the virtual marshaling train energy-saving optimization scenario. The virtual marshaling train energy-saving optimization scenario includes a train running speed curve optimization link and a train running speed curve application link; the train running speed curve optimization link is used to obtain the optimal train running speed curve based on the line data and the historical train running data of trains of different models; the train running speed curve application link is used to make the train run according to the optimal train running speed curve; the virtual marshaling train energy-saving optimization method belongs to the train running speed curve optimization link.

[0114] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 8 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store processing data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for energy-saving optimization of virtual marshaling trains is implemented.

[0115] Those skilled in the art will understand that Figure 8The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0116] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0117] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0118] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0119] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0120] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0121] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0122] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0123] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A virtual train energy-saving optimization method, characterized in that: The virtual train assembly energy saving optimization method comprises: Acquire line data and historical train operation data of trains of different models, and calculate the train operation energy consumption of trains of different models according to the line data and the historical train operation data of trains of different models; the line data includes the distance between each station, the slope and the train speed limit, and the historical train operation data includes the current position, speed, acceleration and running resistance of the train; Based on virtual marshaling technology and mixed integer programming method, a train operation diagram model is constructed according to the train operation energy consumption; the train operation diagram model includes constraints and an objective function, the constraints include train schedule constraints, passenger flow constraints and train coupling constraints, and the objective function is a function with the goal of balancing passenger waiting time and the train operation energy consumption; Solving the train operation diagram model to obtain a train operation diagram; According to the historical train operation data, the train operation diagram and the line data, a double-depth Q network algorithm is used to generate an optimal train operation speed curve.

2. The energy-saving optimization method for virtual train formation according to claim 1, characterized in that: The train operation energy consumption of trains of different models is calculated according to the line data and the train historical operation data of the trains of different models, and the specific process is: The train operation energy consumption of different types of trains is calculated according to the following formula; Among them, v i (0,ξ m ) is the mass ξ at station i m The speed of the m-type train at t = 0; The time spent at the station When it arrives at station i+1, its mass is ξ m The speed of the m-type train; v i (t,ξ m ) is the speed of the train at station i at time t; is the running time of train k at station i; m M-type train quality; is the minimum speed limit at station i at time t; S is the set of stations; is the set of virtual marshaling train types; K is the set of train numbers; Where v(t+1) is the speed of the train at time t+1; x(t+1) is the distance the train travels at time t+1; v(t) is the speed of the train at time t; x(t) is the distance the train travels at time t; λ B (t,ξ m ) is the ratio of the train's braking force output per unit time; λ F (t,ξ m ) is the ratio of the train's output traction force per unit time; B[v i (t,ξ m )] is the speed of the m-type train at station i, which is v i (t,ξ m ) under braking force; r[v i (t,ξ m )] is the speed of the m-type train at station i, which is v i (t,ξ m ) basic resistance under g i (t) is the slope of station i at time t, i.e., the gradient force; F[v i (t,ξ m )] is the speed of the m-type train at station i, which is v i (t,ξ m ) under traction; r(v)=a1+a2v+a3v 2 ; Wherein, F(v) is the traction force when the speed of the train is v; B(v) is the braking force when the speed of the train is v; is the maximum traction; is the maximum traction power; v z Breaking speed limits for traction; is the maximum braking force; is the maximum braking power; v b is the breaking speed limit of the brake; r(v) is the basic resistance when the train speed is v, including friction resistance and air resistance; a1, a2, a3 are representative parameters of the train The Davis equation coefficients of sex; Among them, E m is the total energy consumption of the train, which is composed of the cumulative traction energy consumption generated between stations; x is the running distance of the train between stations.

3. The energy-saving optimization method for virtual train formation according to claim 1, characterized in that: The train schedule constraints include train interval running time constraints and train arrival and departure time constraints; The train interval running time constraints are as follows: in, is the departure time of train k+1 at station i; is the departure time of train k at station i; T min To provide safe departure time between consecutive trains; The time constraints for train arrival and departure are as follows: in, is the arrival time of train k at station i; is the departure time of train k at station i-1; is the running time of train k departing from station i-1; The train coupling constraints are as follows: Among them, y k,m is a decision variable, ensuring that the train can only choose one type of formation.

4. The energy-saving optimization method for virtual train formation according to claim 1, characterized in that: The passenger flow constraints include passengers getting off the train, the number of waiting passengers, the remaining passenger capacity, the number of passengers boarding the train, and the number of passengers on the train; The passengers getting off the bus are: Among them, pa k,i are passengers getting off train k at station i, pr k,i-1 are the remaining passengers who did not board train k at station i-1; A i is a ratio; The number of waiting passengers is: Among them, pw k,i is the number of passengers waiting for train k at station i; pw k-1,i is the number of passengers who failed to board train k-1 at station i; pb k-1,i is the number of passengers arriving at station i after the previous train k-1 left; is the remaining passenger getting-off rate in the car at station i at time t; is a slack variable. When train k departs at [t, t+1], it is the offset of the departure time from station i relative to t. If train k does not depart, its value is 0. is a binary variable, which is 1 if train k departs from station i at time t, otherwise it is 0; is a binary variable, which is 1 if train k-1 departs from station i at time t, otherwise it is 0; The remaining passenger capacity is: Among them, pc k,i is the number of passengers that train k can accommodate at station i, i.e., the train capacity; C m is the initial no-load capacity of train k; The number of passengers on board is: pb k,i =min{pw k,i ,pc k,i },i∈S,k∈K; Among them, pb k,i is the number of passengers boarding train k at station i; The number of passengers on the train in question is:

5. The energy-saving optimization method for virtual train formation according to claim 1, characterized in that: The objective function includes a first sub-objective function and a second sub-objective function; The first sub-objective function is: The second sub-objective function is: The objective function is: F=w1·F1+w2·F2; Among them, w1 and w2 are the weights of the dual-objective optimization method.

6. The energy-saving optimization method for virtual train formation according to claim 1, characterized in that: According to the historical train operation data, the train operation diagram and the line data, a double-depth Q network algorithm is used to generate an optimal train operation speed curve, specifically including: Constructing a train operation reinforcement learning environment according to the train historical operation data, the train operation diagram, the line data and the obtained reward function; Construct a train speed curve optimization network based on a double-depth Q network; Training a train speed curve optimization network based on the train operation reinforcement learning environment to obtain a trained train speed curve optimization network; Based on the trained train speed curve optimization network, the optimal train speed curve is obtained.

7. The energy-saving optimization method for virtual train formation according to claim 6 is characterized in that: The train speed curve optimization network includes an evaluation network and a target network; the train speed curve optimization network is trained based on the train operation reinforcement learning environment to obtain a trained train speed curve optimization network, specifically including: Initializing the evaluation network and the target network; Based on the train operation reinforcement learning environment, determining the current state of the train; the current state includes the current relative position, the current first relative speed, the current relative time and the current second relative speed; Inputting the current state into the evaluation network to determine the current action; According to the current moment action, the current moment reward value is calculated using the train operation reinforcement learning environment, and the next moment state is determined; Input the current moment reward value and the next moment state into the target network to calculate the target Q value; Update the evaluation network and the target network according to the current state, the current action, the target Q value and the loss function; The above steps are repeated until a predetermined number of iterations is reached or the loss function converges to a predetermined threshold, thereby obtaining a trained train speed curve optimization network.

8. The energy-saving optimization method for virtual train formation according to claim 6 is characterized in that: The reward function is: Among them, w3·ΔT represents the punctuality reward; Rewards for energy saving; represents the comfort reward, dt is the time interval; w6·(X i -x) represents the position accuracy reward; w3, w4, w5 and w6 are the weights of the reward function.

9. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the energy-saving optimization method for a virtual train assembly as described in any one of claims 1 to 8.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the energy-saving optimization method for virtual train formation described in any one of claims 1 to 8 is implemented.

Images