Rail train operation braking energy consumption optimization method and system

Abstract

The invention discloses a braking energy consumption optimization method and system for rail train operation, and relates to the technical field of braking energy consumption optimization, and the method comprises the steps: loading a predetermined operation scheme of a target rail train, carrying out the station decomposition, and generating a plurality of node operation schemes; performing braking energy recovery potential prediction to obtain regenerative braking energy of a plurality of nodes; performing energy recovery scheduling on the plurality of braking energy recovery devices, and establishing a plurality of node braking energy recovery scheduling spaces; constructing a braking energy recovery evaluation assembly; and optimizing analysis is conducted on the multiple node braking energy recovery scheduling spaces, multiple node energy recovery optimizing strategies are obtained and encrypted and sent to the train operation control end, and the train operation control end conducts braking energy recovery on the target rail train according to the multiple braking energy recovery devices and the multiple node energy recovery optimizing strategies. The technical problem that in the prior art, the energy utilization rate is low in the energy recovery process is solved, and the technical effect of improving the energy utilization rate is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of braking energy consumption optimization, and in particular to a method and system for optimizing braking energy consumption during rail train operation. Background Art

[0002] With the increasing demand for rail transportation, the braking energy consumption optimization technology of rail trains has received extensive attention and application in the field of rail transportation. However, with the increase in train speed, the extension of operating lines and the continuous increase in energy costs, rail train energy consumption management faces greater challenges. Traditional braking energy recovery methods have many shortcomings in energy consumption optimization, including low energy recovery efficiency and inaccurate prediction of recovery potential, which leads to a large amount of kinetic energy waste in the energy recovery process and makes it difficult to achieve efficient energy utilization. Summary of the invention

[0003] The present application provides a method and system for optimizing the braking energy consumption of rail train operation, which is used to solve the technical problem that in the prior art, kinetic energy is wasted during energy recovery, resulting in low energy utilization.

[0004] In view of the above problems, the present application provides a method and system for optimizing the braking energy consumption of rail train operation.

[0005] In a first aspect of the present application, a method for optimizing braking energy consumption of a rail vehicle is provided, the method comprising: According to the train operation control terminal, a predetermined operation plan of the target rail train is loaded, and the predetermined operation plan is site-decomposed to generate multiple node operation plans, wherein the target rail train includes multiple braking energy recovery devices; based on the multiple node operation plans, the braking energy recovery potential of the target rail train is predicted according to the ARIMA-LSTM combination model to obtain multiple node regenerative braking energies; based on the multiple node regenerative braking energies, energy recovery scheduling is performed on the multiple braking energy recovery devices respectively, and multiple node braking energy recovery scheduling spaces are established; based on the multivariate braking energy recovery evaluation factors, a braking energy recovery evaluation component is constructed, wherein the multivariate braking energy recovery evaluation factors include braking energy recovery efficiency, braking energy recovery safety and braking energy recovery stability; based on the braking energy recovery evaluation component, optimization analysis is performed on the multiple node braking energy recovery scheduling spaces respectively according to the recovery scheduling optimization number constraint to obtain multiple node energy recovery optimization strategies; the multiple node energy recovery optimization strategies are encrypted and sent to the train operation control terminal, and the train operation control terminal performs braking energy recovery on the target rail train according to the multiple braking energy recovery devices and the multiple node energy recovery optimization strategies.

[0006] The second aspect of the present application provides a braking energy consumption optimization system for rail train operation, the system comprising: A node operation plan generation module, wherein the node operation plan generation module loads the predetermined operation plan of the target rail train according to the train operation control terminal, and decomposes the predetermined operation plan by station to generate multiple node operation plans, wherein the target rail train includes multiple braking energy recovery devices; a recovery potential prediction module, wherein the recovery potential prediction module predicts the braking energy recovery potential of the target rail train based on the multiple node operation plans according to the ARIMA-LSTM combination model to obtain multiple node regenerative braking energies; an energy recovery scheduling module, wherein the energy recovery scheduling module performs energy recovery scheduling on the multiple braking energy recovery devices respectively based on the multiple node regenerative braking energies, and establishes multiple node braking energy recovery scheduling spaces; an evaluation component construction module, wherein the evaluation component construction module The component construction module constructs a braking energy recovery evaluation component based on multiple braking energy recovery evaluation factors, wherein the multiple braking energy recovery evaluation factors include braking energy recovery efficiency, braking energy recovery safety and braking energy recovery stability; an optimization analysis module, which is based on the braking energy recovery evaluation component and performs optimization analysis on the multiple node braking energy recovery scheduling spaces according to the recovery scheduling optimization number constraints to obtain multiple node energy recovery optimization strategies; a braking energy recovery module, which encrypts the multiple node energy recovery optimization strategies and sends them to the train operation control terminal, and the train operation control terminal performs braking energy recovery on the target rail train according to the multiple braking energy recovery devices and the multiple node energy recovery optimization strategies.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: The present application loads a predetermined operation plan of a target rail train according to a train operation control terminal, and decomposes the predetermined operation plan by station to generate multiple node operation plans, wherein the target rail train includes multiple braking energy recovery devices; based on the multiple node operation plans, the braking energy recovery potential of the target rail train is predicted according to the ARIMA-LSTM combination model to obtain multiple node regenerative braking energies; based on the multiple node regenerative braking energies, energy recovery scheduling is performed on multiple braking energy recovery devices respectively, and multiple node braking energy recovery scheduling spaces are established; based on multiple braking energy recovery evaluation factors, a braking energy recovery evaluation component is constructed, wherein the multiple braking energy recovery evaluation factors include braking energy recovery efficiency, braking energy recovery safety and braking energy recovery stability; based on the braking energy recovery evaluation component, optimization analysis is performed on multiple node braking energy recovery scheduling spaces respectively according to the recovery scheduling optimization number constraint to obtain multiple node energy recovery optimization strategies; the multiple node energy recovery optimization strategies are encrypted and sent to the train operation control terminal, and the train operation control terminal performs braking energy recovery on the target rail train according to the multiple braking energy recovery devices and the multiple node energy recovery optimization strategies. The present invention solves the technical problem that kinetic energy is wasted in the energy recovery process of the prior art, resulting in low energy utilization. The operation plan is loaded by the train operation control end, and the model is used to predict the braking energy recovery potential. The energy recovery device of each node is scheduled and optimized, and the scheduling space is optimized based on the evaluation component. The optimization strategy is transmitted to the control end to realize efficient and stable energy recovery of rail trains and achieve the technical effect of improving energy utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0009] Figure 1 A schematic diagram of a process flow of a method for optimizing braking energy consumption of a rail train in accordance with an embodiment of the present application; Figure 2 A schematic diagram of the structure of a braking energy consumption optimization system for rail train operation provided in an embodiment of the present application.

[0010] Explanation of the reference numerals: node operation plan generation module 11, recovery potential prediction module 12, energy recovery scheduling module 13, evaluation component construction module 14, optimization analysis module 15, braking energy recovery module 16. DETAILED DESCRIPTION

[0011] The present application provides a method and system for optimizing the braking energy consumption of rail train operation, aiming to solve the technical problem that in the prior art, kinetic energy is wasted in the energy recovery process, resulting in low energy utilization. The operation plan is loaded through the train operation control end, and the braking energy recovery potential is predicted by the model. The energy recovery device of each node is scheduled and optimized, and the scheduling space is optimized based on the evaluation component. The optimization strategy is transmitted to the control end to realize efficient and stable energy recovery of rail trains and achieve the technical effect of improving energy utilization.

[0012] 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, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0013] It should be noted that any variations of the terms "include" and "have" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or inherent to these processes, methods, products or devices.

[0014] Embodiment 1, as Figure 1 As shown, the present application provides a method for optimizing braking energy consumption of rail train operation, the method comprising: Step S100: According to the train operation control terminal, a predetermined operation plan of a target rail train is loaded, and the predetermined operation plan is decomposed into stations to generate multiple node operation plans, wherein the target rail train includes multiple braking energy recovery devices.

[0015] In the embodiment of the present application, the train operation control terminal loads the predetermined operation plan before starting the operation, that is, the planned route, speed, stop sites and operation instructions to be executed. After loading the predetermined operation plan, it is decomposed into sites, and the entire operation process is divided into multiple relatively independent operation nodes according to the sites. Each node represents the process of the train arriving at the next site from one site, specifically including the acceleration, deceleration, uniform speed and other operations of the section. Through this process, multiple node operation plans are generated.

[0016] The target rail train includes multiple braking energy recovery devices, which rely on multiple braking energy recovery devices to recover braking energy. These devices convert the excess kinetic energy generated during the braking process into electrical energy, which is stored or reused. The braking energy recovery device includes an electric braking system and an ultracapacitor energy storage device. The electric braking system converts kinetic energy into electrical energy through generator braking, which is then fed back to the train power grid. The ultracapacitor energy storage device uses the fast charging and discharging characteristics of ultracapacitors to quickly store the electrical energy generated by braking and release it when needed to support subsequent train acceleration or other high-energy consumption requirements.

[0017] Step S200: Based on the multiple node operation plans, the braking energy recovery potential of the target rail train is predicted according to the ARIMA-LSTM combined model to obtain the regenerative braking energy of multiple nodes.

[0018] In an embodiment of the present application, firstly, the load of the target rail train is predicted based on multiple node operation schemes to generate predicted load data for each node. Next, the historical braking energy recovery records are retrieved to construct a sample node operation scheme set, a sample node train load set, and a sample node train recovery braking energy set. Based on these sample data, the ARIMA model and the LSTM model are jointly learned to build a braking energy recovery potential prediction channel. This channel can capture the energy recovery potential of the train during braking, identify and predict the regenerative braking energy of each node. Finally, multiple node operation schemes and predicted loads are input into this channel to obtain the regenerative braking energy of multiple nodes.

[0019] Furthermore, in the method provided in the embodiment of the application, based on the multiple node operation schemes, the braking energy recovery potential of the target rail train is predicted according to the ARIMA-LSTM combined model to obtain the regenerative braking energy of multiple nodes, and further includes: Based on the multiple node operation plans, load prediction is performed on the target rail train to obtain multiple node train predicted loads; braking energy recovery records are retrieved for the target rail train to obtain a sample node operation plan set, a sample node train load set and a sample node train recovery braking energy set; the ARIMA-LSTM combined model includes an ARIMA model and an LSTM model; based on the sample node operation plan set, the sample node train load set and the sample node train recovery braking energy set, the ARIMA model and the LSTM model are jointly learned to build a braking energy recovery potential prediction channel; the multiple node operation plans and the multiple node train predicted loads are input into the braking energy recovery potential prediction channel to obtain the regenerative braking energy of the multiple nodes.

[0020] In the embodiment of the present application, the load of the rail train is first predicted based on the operation schemes of multiple nodes. The historical data mean method is used here to predict the load. Specifically, the load data of the same period and section in the past seven days are extracted from the historical database. Then the obtained seven-day load data is averaged to obtain the predicted load of each node. In this way, the predicted load data of multiple nodes are generated.

[0021] Next, in order to generate the training data set for the ARIMA-LSTM model, the braking energy recovery records of the target rail train are retrieved, and the sample node operation plan set, sample node train load set, and sample node train recovery braking energy set of the target rail train are obtained from the historical database. The sample node operation plan set records the actual operation plan of the target rail train at each historical node. The sample node train load set corresponds to the actual load data of the train at each historical node. The sample node train recovery braking energy set is the energy data recovered during the actual braking process of the corresponding node.

[0022] Subsequently, in the process of predicting the braking energy recovery potential of rail trains, the ARIMA-LSTM combined model is used. The model consists of an ARIMA model and an LSTM model, and the energy recovery potential is predicted through joint learning. Specifically, the ARIMA and LSTM models are first jointly trained based on the sample node operation plan set, the sample node train load set, and the sample node train recovery braking energy set. The ARIMA model takes the sample node operation plan set and the sample node train load set as input, and the sample node train recovery braking energy set as output. The ARIMA model is supervised and trained to obtain the first recovery potential prediction model for load trend prediction. The LSTM model also takes the sample node operation plan set and the sample node train load set as input, and the sample node train recovery braking energy set as output. The LSTM model is trained to enable it to identify and capture nonlinear relationships in the data, thereby generating the second recovery potential prediction model.

[0023] After the ARIMA and LSTM models are trained, the outputs of the two models are integrated into a fusion training data set, and further training and testing are performed in combination with the sample node train recovery braking energy set to obtain a recovery potential prediction fusion model. Finally, the ARIMA and LSTM models are used as the first-level nodes, the fusion model is used as the second-level node, and these nodes are connected to construct a braking energy recovery potential prediction channel.

[0024] Finally, the operation plans and predicted load data of multiple nodes are input into the prediction channel, and the regenerative braking energy of each node is output to obtain the regenerative braking energy of multiple nodes.

[0025] Furthermore, in the method provided in the embodiment of the application, based on the sample node operation scheme set, the sample node train load set and the sample node train recovery braking energy set, the ARIMA model and the LSTM model are jointly learned to build a braking energy recovery potential prediction channel, and further includes: Taking the sample node operation scheme set and the sample node train load set as input information and the sample node train recovery braking energy set as output information, the ARIMA model is supervised trained to obtain a first recovery potential prediction model; the LSTM model is supervised trained according to the sample node operation scheme set, the sample node train load set and the sample node train recovery braking energy set to obtain a second recovery potential prediction model; the output data of the first recovery potential prediction model and the second recovery potential prediction model are collected to obtain a fusion training data set; training and testing are performed according to the fusion training data set and the sample node train recovery braking energy set to obtain a recovery potential prediction fusion model; the first recovery potential prediction model and the second recovery potential prediction model are used as the first-level nodes for braking energy recovery potential prediction, and the recovery potential prediction fusion model is used as the second-level node for braking energy recovery potential prediction; the first-level node for braking energy recovery potential prediction and the second-level node for braking energy recovery potential prediction are connected to generate the braking energy recovery potential prediction channel.

[0026] In the embodiment of the present application, the ARIMA model is first supervised and trained with the sample node operation scheme set and the sample node train load set as input data and the sample node train recovery braking energy set as output data. The ARIMA model generates a first recovery potential prediction model by identifying trends and seasonal changes in the time series, which is used to predict the energy recovery trend of the train during braking.

[0027] Next, the LSTM model is trained, using the same sample node operation plan set and train load set as input, and the sample node train recovery braking energy set as output. The LSTM model can handle nonlinear and complex time series features, and by learning the relationship between load and energy recovery, it generates a second recovery potential prediction model to capture the short-term fluctuations and nonlinear characteristics between train load and energy recovery.

[0028] After the ARIMA and LSTM models are trained, the output data of the two models are collected to form a fusion training data set. Then, a fusion model is trained using the fusion training data set as input and the sample node train recovery braking energy set as the target output. The fusion model is trained using a multi-layer perceptron (MLP) model to achieve an optimal balance between the outputs of different models. The prediction effect of the fusion model is tested and verified through cross-validation to ensure that it has a high generalization ability for the energy recovery potential of different nodes, thereby obtaining a recovery potential prediction fusion model.

[0029] When constructing the braking energy recovery potential prediction channel, the ARIMA and LSTM models are set as the first-level nodes to capture the trend and nonlinear characteristics in the data respectively, and the recovery potential prediction fusion model is used as the second-level node to integrate the output of the first-level node to improve the overall prediction accuracy. The connection between the first-level node and the second-level node constitutes the braking energy recovery potential prediction channel.

[0030] Step S300: Based on the regenerative braking energies of the multiple nodes, respectively perform energy recovery scheduling on the multiple braking energy recovery devices to establish multiple node braking energy recovery scheduling spaces.

[0031] In an embodiment of the present application, based on the regenerative braking energy of multiple nodes, multiple braking energy recovery devices are scheduled for energy recovery node by node to construct a braking energy recovery scheduling space for multiple nodes. Specifically, the regenerative braking energy of each node, such as the qth node, is first extracted, and the energy storage characteristic parameters of each recovery device are collected to generate an energy storage characteristic data set. Then, the regenerative braking energy of the node is used as the scheduling target, and energy recovery decisions are made for multiple recovery devices in combination with the energy storage characteristic data set to form a corresponding scheduling space.

[0032] After constructing the scheduling space of each node, check whether the number of decisions in the scheduling space of the node meets the preset quantity constraints. If the constraints are met, the scheduling space is added to the overall multi-node scheduling space; if not, the scheduling scheme is expanded based on the scheduling target and energy storage characteristic data. Finally, through the above process, the braking energy recovery scheduling space of multiple nodes is obtained.

[0033] Furthermore, in the method provided in the embodiment of the application, based on the regenerative braking energy of the multiple nodes, the multiple braking energy recovery devices are respectively scheduled for energy recovery, and multiple node braking energy recovery scheduling spaces are established, which also includes: According to the regenerative braking energy of the multiple nodes, extract the regenerative braking energy of the qth node, wherein q is a positive integer; collect the energy storage characteristic parameters of the multiple braking energy recovery devices to obtain an energy storage characteristic data set; take the regenerative braking energy of the qth node as the energy recovery scheduling target, make energy recovery decisions for the multiple braking energy recovery devices according to the energy storage characteristic data set, and obtain the braking energy recovery scheduling space of the qth node; collect the decision quantity parameters of the braking energy recovery scheduling space of the qth node to obtain the decision quantity of the qth space; judge whether the decision quantity of the qth space satisfies the decision quantity constraint; if the decision quantity of the qth space satisfies the decision quantity constraint, add the braking energy recovery scheduling space of the qth node to the braking energy recovery scheduling spaces of the multiple nodes; if the decision quantity of the qth space does not meet the decision quantity constraint, based on the decision quantity constraint, make a decision expansion on the braking energy recovery scheduling space of the qth node according to the energy recovery scheduling target and the energy storage characteristic data set.

[0034] In the embodiment of the present application, the node currently being processed, for example, the regenerative braking energy of the qth node, is first extracted from the regenerative braking energy data of multiple nodes, and is used as the energy recovery scheduling target of the node. Through the query operation, the specific energy value of the qth node is directly obtained from the regenerative braking energy data set. Where q is a positive integer.

[0035] Next, the energy storage characteristic parameters of all the braking energy recovery devices of the node are obtained from the preset database to form an energy storage characteristic data set including information such as the energy storage capacity, charging rate, and current energy storage level of each device.

[0036] After obtaining the scheduling target and energy storage feature data set of the qth node, energy allocation is performed through the constrained optimization algorithm to construct the braking energy recovery scheduling space of the qth node. Here, the energy demand of the qth node is taken as the core optimization target, and a series of constraints are set for each recovery device on this basis. Specifically, first, it is ensured that the allocated energy of each recovery device does not exceed its maximum energy storage capacity. This constraint ensures that the energy storage device will not be overloaded during the energy recovery process. Secondly, the constraint of the charging rate ensures that the charging rate of each device is within the allowable range of the device to prevent the device from overheating or damage due to excessive charging rate. In addition, considering the current energy storage status of each device, the energy allocation is limited according to the existing energy storage level to avoid energy waste caused by full energy storage. These constraints are pre-set by technical experts. Under the framework of the above optimization targets and constraints, this information is input into the constrained optimization algorithm. Through the algorithm's solution process, the optimal energy allocation scheme that meets all constraints is searched under the defined conditions. This process uses optimization methods such as gradient descent or interior point method. Through these solution techniques, the best solution for energy allocation is found under multiple constraints. After the solution is completed, multiple qualified recovery scheduling decisions are generated, each decision represents an energy allocation combination, and these combinations together constitute the braking energy recovery scheduling space of the qth node.

[0037] Then the decision quantity parameter in the scheduling space of the qth node is calculated, that is, the total number of feasible recovery scheduling schemes in the scheduling space, which is obtained by statistical calculation. The decision quantity parameter is compared with the preset decision quantity constraint to ensure that the number of schemes in the scheduling space is sufficient to cope with different energy requirements. If the decision quantity meets the constraint requirements, the scheduling space of the qth node is added to the overall multi-node braking energy recovery scheduling space, which means that the energy recovery scheduling of the node has been successfully completed.

[0038] If the number of decisions of the qth node does not meet the preset requirements, the heuristic expansion method is used to expand the decision-making of the scheduling space of the node. During the expansion process, the number of decisions is increased by generating new recycling scheduling combinations, such as using genetic algorithms or simulated annealing algorithms to generate more allocation schemes to ensure sufficient scheduling scheme support under diverse recycling needs. Through these expansion technologies, the number and diversity of scheduling schemes are increased, thereby maintaining high flexibility in different energy recovery scenarios.

[0039] Through the above process, the braking energy recovery scheduling space including all nodes is gradually constructed. The scheduling space of each node contains multiple recovery scheduling decisions.

[0040] Step S400: constructing a braking energy recovery evaluation component based on multiple braking energy recovery evaluation factors, wherein the multiple braking energy recovery evaluation factors include braking energy recovery efficiency, braking energy recovery safety and braking energy recovery stability.

[0041] In the embodiment of the present application, the multivariate braking energy recovery evaluation factors include braking energy recovery efficiency, braking energy recovery safety and braking energy recovery stability. First, the importance analysis is performed based on these three evaluation factors to obtain the key coefficients of each factor, and then the weights of each factor are assigned based on these key coefficients to construct a weighted network, namely, the braking energy recovery evaluation weighted network.

[0042] On the basis of the weighted network, three network models for predicting different evaluation factors are further trained to obtain the braking energy recovery efficiency prediction network, the braking energy recovery safety prediction network and the braking energy recovery stability prediction network. These networks are set as primary nodes to evaluate the performance of each factor respectively. Then these primary nodes are connected to the secondary nodes, which are weighted networks and serve as comprehensive evaluation output nodes. Through the connection of the primary nodes and the secondary nodes, the braking energy recovery evaluation component is finally formed.

[0043] Furthermore, in the method provided in the embodiment of the application, a braking energy recovery evaluation component is constructed based on multiple braking energy recovery evaluation factors, and further includes: Importance evaluation is performed according to the multivariate braking energy recovery evaluation factors to obtain multiple key coefficients of the evaluation factors; weights are assigned to the multivariate braking energy recovery evaluation factors according to the multiple key coefficients of the evaluation factors to establish a braking energy recovery evaluation weighted network; based on the multivariate braking energy recovery evaluation factors, a braking energy recovery efficiency prediction network, a braking energy recovery safety prediction network and a braking energy recovery stability prediction network are trained; the braking energy recovery efficiency prediction network, the braking energy recovery safety prediction network and the braking energy recovery stability prediction network are used as the first-level nodes of braking energy recovery evaluation; the braking energy recovery evaluation weighted network is used as the second-level node of braking energy recovery evaluation; the first-level node of braking energy recovery evaluation and the second-level node of braking energy recovery evaluation are connected to generate the braking energy recovery evaluation component.

[0044] In the embodiment of the present application, the importance of the multivariate braking energy recovery evaluation factors is first evaluated to determine the relative importance of each factor. The multivariate linear regression method is used here to determine the degree of influence of each factor on the overall energy recovery effect of the system through regression analysis. Specifically, the historical data of the three factors of recovery efficiency, safety and stability are used as independent variables, and the overall energy recovery performance, such as recovery rate, is used as the dependent variable for multivariate linear regression analysis. After the regression model is generated, the regression coefficient of each factor is obtained. The larger the regression coefficient, the greater the influence of the factor on the system performance. Finally, the key coefficient of each factor is determined based on the regression coefficient. For example, if the regression coefficient of recovery efficiency is 0.7, safety is 0.2, and stability is 0.1, then the importance of recovery efficiency is the highest and the key coefficient is relatively large. Through this process, multiple key coefficients of evaluation factors are obtained.

[0045] Next, the multivariate braking energy recovery evaluation factors are weighted according to the key coefficients of multiple evaluation factors, and a braking energy recovery evaluation weighted network is established on this basis. The construction of the weighted network is based on the weighted summation method, and the key coefficients are directly used as the weight parameters of the weighted network to ensure that each factor is reflected according to its importance in the final comprehensive score. For example, the key coefficient of recovery efficiency is 0.7, which is used as the weight of the recovery efficiency factor in the weighted network; similarly, the weights of safety and stability factors are 0.2 and 0.1 respectively. This weighted network reasonably integrates the influence of each factor in the comprehensive evaluation through weight allocation.

[0046] After the weighted network is established, three independent prediction networks are trained separately, namely the braking energy recovery efficiency prediction network, the braking energy recovery safety prediction network and the braking energy recovery stability prediction network. Each network is used to predict the performance of the corresponding factor separately. These networks are all neural network models. During the training process, the historical data of each factor is input into the corresponding network for training. Through the back propagation algorithm and the gradient descent method, the parameters of the network are optimized and adjusted so that the network can accurately predict the recovery efficiency, safety and stability scores. The output of each prediction network represents the factor score under different conditions, which provides a basis for the subsequent comprehensive evaluation.

[0047] After that, the above three prediction networks are used as first-level nodes, and each first-level node outputs the recovery efficiency, safety and stability scores respectively. Then, these scores are input into the second-level node brake energy recovery evaluation weighted network. As a second-level node, the role of the weighted network is to weight the score of each factor according to its weight to output a comprehensive evaluation score. Specifically, the weighted network multiplies the scores output by the first-level nodes by the corresponding weights, for example, the recovery efficiency score is multiplied by 0.7, the safety score is multiplied by 0.2, and the stability is multiplied by 0.1, and these results are added together to obtain an overall evaluation score. This score reflects the comprehensive performance of the recovery device.

[0048] By connecting the primary node and the secondary node, the construction of the braking energy recovery evaluation component is finally completed.

[0049] Step S500: Based on the braking energy recovery evaluation component, the plurality of node braking energy recovery scheduling spaces are respectively optimized and analyzed according to the recovery scheduling optimization number constraint to obtain a plurality of node energy recovery optimization strategies.

[0050] In an embodiment of the present application, based on the braking energy recovery evaluation component, the braking energy recovery scheduling space of each node is optimized and analyzed respectively to generate an energy recovery optimization strategy for each node. In this process, each node, such as the scheduling space of the qth node, is optimized independently in turn. Specifically, an initial scheduling scheme, i.e., the mth scheme, is randomly extracted from the scheduling space of the qth node, and the energy recovery performance of the scheme is calculated using the evaluation component to obtain its evaluation coefficient. Subsequently, the next scheduling scheme, i.e., the m+1th scheme, is extracted from the scheduling space, and its evaluation coefficient is calculated and compared with the previous scheme, from which the current winning scheme is selected as the new benchmark.

[0051] Repeat this process in the scheduling space of each node, continuously extract new solutions from the scheduling space for evaluation and comparison, and gradually optimize the performance of the scheduling strategy. During the iteration process, according to the preset recycling scheduling optimization number constraint, find the best scheduling solution within the limited number of iterations. When the optimization number reaches the constraint condition, the optimal solution of the node is determined as its energy recovery optimization strategy. This optimization process is performed on the scheduling space of all nodes in turn, and finally an independent energy recovery optimization strategy is generated for each node, and these strategies are summarized to obtain multiple node energy recovery optimization strategies.

[0052] Furthermore, in the method provided in the embodiment of the application, the optimization analysis is performed on the multiple node braking energy recovery scheduling spaces respectively according to the recovery scheduling optimization number constraint to obtain multiple node energy recovery optimization strategies, and further includes: According to the q-node braking energy recovery scheduling space, the m-th braking energy recovery scheduling decision is randomly extracted, wherein m is a positive integer; based on the braking energy recovery evaluation component, the m-th braking energy recovery scheduling decision is subjected to recovery prediction evaluation to obtain the m-th braking energy recovery evaluation coefficient; according to the q-node braking energy recovery scheduling space, the m+1-th braking energy recovery scheduling decision is extracted, and the m+1-th braking energy recovery evaluation coefficient is calculated in combination with the braking energy recovery evaluation component; based on the m-th braking energy recovery evaluation coefficient and the m+1-th braking energy recovery evaluation coefficient, the m-th braking energy recovery scheduling decision and the m+1-th braking energy recovery scheduling decision are identified as superior to obtain the current superior recovery scheduling decision; based on the braking energy recovery evaluation component, the current superior recovery scheduling decision is continuously iterated and optimized according to the q-node braking energy recovery scheduling space until the number of iterative optimizations meets the recovery scheduling optimization number constraint, and the q-th node energy recovery optimization strategy is generated; and the q-th node energy recovery optimization strategy is added to the multiple node energy recovery optimization strategies.

[0053] In an embodiment of the present application, when optimizing the braking energy recovery scheduling space of the qth node, first randomly extract an mth braking energy recovery scheduling decision, where m is a positive integer, and use the braking energy recovery evaluation component to perform a recovery prediction evaluation on the decision to obtain the mth braking energy recovery evaluation coefficient. Specifically, the operation plan of the qth node is extracted according to the operation plans of multiple nodes, and the target rail train is modeled to obtain the target train model. Based on the model and the mth scheduling decision, a braking energy recovery simulation is performed to obtain the energy recovery performance data of the mth decision. Subsequently, these data are input into the first-level node of the evaluation component to calculate the recovery efficiency prediction coefficient, the recovery safety prediction coefficient and the recovery stability prediction coefficient. These coefficients reflect the energy recovery performance of each item of the mth scheme, and are comprehensively weighted at the second-level node of the evaluation component to output a comprehensive score, namely the mth braking energy recovery evaluation coefficient.

[0054] After obtaining the evaluation coefficient of the mth solution, continue to extract the next scheduling decision from the scheduling space of the qth node, that is, the m+1th scheduling decision, and calculate its m+1th braking energy recovery evaluation coefficient through the same process. Compare the evaluation coefficients of the mth and m+1th solutions, and record the scheduling solution with the higher score as the current winning recovery scheduling decision.

[0055] In the scheduling space of the qth node, iterative optimization is performed based on the current winning solution. By continuously extracting new solutions, calculating their evaluation coefficients and comparing them with the current winning solution, the scheduling decision is gradually optimized, and the solution with the highest score is always retained. This process is carried out under the preset number of recycling scheduling optimization constraints. When the number of iterations reaches the upper limit, the current winning solution is determined as the energy recovery optimization strategy for the qth node.

[0056] Finally, the optimal strategy of the node is added to the energy recovery optimization strategy set of multiple nodes, and the same optimization analysis is performed on all nodes in turn to ensure that the energy recovery strategy with the best overall performance is found among multiple nodes.

[0057] Furthermore, in the method provided in the embodiment of the application, based on the braking energy recovery evaluation component, the mth braking energy recovery scheduling decision is subjected to recovery prediction evaluation to obtain the mth braking energy recovery evaluation coefficient, and further includes: According to the multiple node operation plans, extract the qth node operation plan; based on the target rail train, modeling is performed to obtain a target train model; based on the qth node operation plan, braking energy recovery simulation is performed on the target train model according to the mth braking energy recovery scheduling decision to obtain an mth decision energy recovery simulation data set; the mth decision energy recovery simulation data set is input into the first-level braking energy recovery evaluation node to obtain the mth recovery efficiency prediction coefficient, the mth recovery safety prediction coefficient and the mth recovery stability prediction coefficient; the mth recovery efficiency prediction coefficient, the mth recovery safety prediction coefficient and the mth recovery stability prediction coefficient are input into the second-level braking energy recovery evaluation node to output the mth braking energy recovery evaluation coefficient.

[0058] In the embodiment of the present application, firstly, the operation plan of the qth node is extracted according to the operation plans of multiple nodes. The operation plan of the qth node provides the operation conditions and task parameters of the node, laying the foundation for the subsequent energy recovery evaluation.

[0059] Next, we build a target train model based on the specific properties of the target track train. The model contains the physical characteristics of the train during braking, including mass, speed, braking force, energy recovery device configuration and other information, to ensure that the model can accurately reflect the energy conversion behavior of the train during operation.

[0060] Based on the operation plan of the qth node, the mth braking energy recovery scheduling decision is applied to simulate the braking energy recovery of the target train model, thereby generating an mth decision energy recovery simulation data set. This data set records the energy recovery performance under the mth scheduling decision, and contains multiple energy recovery indicators, such as recovery efficiency, safety and stability data, to provide specific performance data support for subsequent evaluation. After that, the generated mth decision energy recovery simulation data set is input into the first-level node of the braking energy recovery evaluation component. The first-level node calculates the performance of the scheme in terms of recovery efficiency, safety and stability, and outputs the corresponding mth recovery efficiency prediction coefficient, mth recovery safety prediction coefficient and mth recovery stability prediction coefficient. These prediction coefficients quantify the effects of the decision in terms of energy recovery efficiency, safety and operational stability, respectively, and provide sub-item support for comprehensive evaluation.

[0061] Finally, the mth recovery efficiency prediction coefficient, the mth recovery safety prediction coefficient and the mth recovery stability prediction coefficient are input into the secondary node of the evaluation component, which performs comprehensive weighted calculation and outputs the mth braking energy recovery evaluation coefficient.

[0062] Step S600: encrypting the multiple node energy recovery optimization strategies and sending them to the train operation control terminal, and the train operation control terminal performs braking energy recovery on the target rail train according to the multiple braking energy recovery devices and the multiple node energy recovery optimization strategies.

[0063] In the embodiment of the present application, the energy recovery optimization strategies of multiple nodes are first encrypted to ensure that the data is not intercepted or tampered with during transmission. The encryption process uses a symmetric encryption algorithm (such as AES) or an asymmetric encryption algorithm (such as RSA) to generate a key and convert the energy recovery strategy of each node into an encrypted data packet. The encrypted data packet ensures that it can only be decrypted and read at the authorized control end.

[0064] After encryption is completed, the encrypted data packet is sent to the train operation control end, and the data security during the transmission process is further guaranteed through a secure transmission protocol such as TLS or VPN channel. After arriving at the control end, the operation control end decrypts the data through the shared decryption key and restores the original multiple node energy recovery optimization strategy.

[0065] After the operation control end obtains and decrypts the data, it combines the actual configuration and status of multiple brake energy recovery devices, gradually applies the energy recovery strategy of each node, and performs brake energy recovery operations on the target rail train to ensure the optimal energy recovery effect at each node.

[0066] In the embodiments of the present application, in summary, the embodiments of the present application have at least the following technical effects: The present application loads a predetermined operation plan of a target rail train according to a train operation control terminal, and decomposes the predetermined operation plan by station to generate multiple node operation plans, wherein the target rail train includes multiple braking energy recovery devices; based on the multiple node operation plans, the braking energy recovery potential of the target rail train is predicted according to the ARIMA-LSTM combination model to obtain multiple node regenerative braking energies; based on the multiple node regenerative braking energies, energy recovery scheduling is performed on multiple braking energy recovery devices respectively, and multiple node braking energy recovery scheduling spaces are established; based on multiple braking energy recovery evaluation factors, a braking energy recovery evaluation component is constructed, wherein the multiple braking energy recovery evaluation factors include braking energy recovery efficiency, braking energy recovery safety and braking energy recovery stability; based on the braking energy recovery evaluation component, optimization analysis is performed on multiple node braking energy recovery scheduling spaces respectively according to the recovery scheduling optimization number constraint to obtain multiple node energy recovery optimization strategies; the multiple node energy recovery optimization strategies are encrypted and sent to the train operation control terminal, and the train operation control terminal performs braking energy recovery on the target rail train according to the multiple braking energy recovery devices and the multiple node energy recovery optimization strategies. The present invention solves the technical problem that kinetic energy is wasted in the energy recovery process of the prior art, resulting in low energy utilization. The operation plan is loaded by the train operation control end, and the model is used to predict the braking energy recovery potential. The energy recovery device of each node is scheduled and optimized, and the scheduling space is optimized based on the evaluation component. The optimization strategy is transmitted to the control end to realize efficient and stable energy recovery of rail trains and achieve the technical effect of improving energy utilization.

[0067] Embodiment 2, based on the same inventive concept as the method for optimizing the braking energy consumption of rail train operation in the above embodiment, Figure 2 As shown, the present application provides a braking energy consumption optimization system for rail train operation, and the system and method embodiments in the present application are based on the same inventive concept. The system includes: A node operation plan generation module 11, wherein the node operation plan generation module 11 loads the predetermined operation plan of the target rail train according to the train operation control terminal, and decomposes the predetermined operation plan by station to generate multiple node operation plans, wherein the target rail train includes multiple braking energy recovery devices; a recovery potential prediction module 12, wherein the recovery potential prediction module 12 predicts the braking energy recovery potential of the target rail train based on the multiple node operation plans according to the ARIMA-LSTM combination model to obtain multiple node regenerative braking energies; an energy recovery scheduling module 13, wherein the energy recovery scheduling module 13 performs energy recovery scheduling on the multiple braking energy recovery devices respectively based on the multiple node regenerative braking energies, and establishes multiple node braking energy recovery scheduling spaces; an evaluation component construction module 14, wherein the The evaluation component construction module 14 constructs a braking energy recovery evaluation component based on multiple braking energy recovery evaluation factors, wherein the multiple braking energy recovery evaluation factors include braking energy recovery efficiency, braking energy recovery safety and braking energy recovery stability; an optimization analysis module 15, wherein the optimization analysis module 15 is based on the braking energy recovery evaluation component, and according to the recovery scheduling optimization number constraint, respectively performs optimization analysis on the multiple node braking energy recovery scheduling spaces to obtain multiple node energy recovery optimization strategies; a braking energy recovery module 16, wherein the braking energy recovery module 16 encrypts the multiple node energy recovery optimization strategies and sends them to the train operation control terminal, and the train operation control terminal performs braking energy recovery on the target rail train according to the multiple braking energy recovery devices and the multiple node energy recovery optimization strategies.

[0068] Furthermore, the system is also used to implement the following functions: Based on the multiple node operation plans, load prediction is performed on the target rail train to obtain multiple node train predicted loads; braking energy recovery records are retrieved for the target rail train to obtain a sample node operation plan set, a sample node train load set and a sample node train recovery braking energy set; the ARIMA-LSTM combined model includes an ARIMA model and an LSTM model; based on the sample node operation plan set, the sample node train load set and the sample node train recovery braking energy set, the ARIMA model and the LSTM model are jointly learned to build a braking energy recovery potential prediction channel; the multiple node operation plans and the multiple node train predicted loads are input into the braking energy recovery potential prediction channel to obtain the regenerative braking energy of the multiple nodes.

[0069] Furthermore, the system is also used to implement the following functions: Taking the sample node operation scheme set and the sample node train load set as input information and the sample node train recovery braking energy set as output information, the ARIMA model is supervised trained to obtain a first recovery potential prediction model; the LSTM model is supervised trained according to the sample node operation scheme set, the sample node train load set and the sample node train recovery braking energy set to obtain a second recovery potential prediction model; the output data of the first recovery potential prediction model and the second recovery potential prediction model are collected to obtain a fusion training data set; training and testing are performed according to the fusion training data set and the sample node train recovery braking energy set to obtain a recovery potential prediction fusion model; the first recovery potential prediction model and the second recovery potential prediction model are used as the first-level nodes for braking energy recovery potential prediction, and the recovery potential prediction fusion model is used as the second-level node for braking energy recovery potential prediction; the first-level node for braking energy recovery potential prediction and the second-level node for braking energy recovery potential prediction are connected to generate the braking energy recovery potential prediction channel.

[0070] Furthermore, the system is also used to implement the following functions: According to the regenerative braking energy of the multiple nodes, extract the regenerative braking energy of the qth node, wherein q is a positive integer; collect the energy storage characteristic parameters of the multiple braking energy recovery devices to obtain an energy storage characteristic data set; take the regenerative braking energy of the qth node as the energy recovery scheduling target, make energy recovery decisions for the multiple braking energy recovery devices according to the energy storage characteristic data set, and obtain the braking energy recovery scheduling space of the qth node; collect the decision quantity parameters of the braking energy recovery scheduling space of the qth node to obtain the decision quantity of the qth space; judge whether the decision quantity of the qth space satisfies the decision quantity constraint; if the decision quantity of the qth space satisfies the decision quantity constraint, add the braking energy recovery scheduling space of the qth node to the braking energy recovery scheduling spaces of the multiple nodes; if the decision quantity of the qth space does not meet the decision quantity constraint, based on the decision quantity constraint, make a decision expansion on the braking energy recovery scheduling space of the qth node according to the energy recovery scheduling target and the energy storage characteristic data set.

[0071] Furthermore, the system is also used to implement the following functions: Importance evaluation is performed according to the multivariate braking energy recovery evaluation factors to obtain multiple key coefficients of the evaluation factors; weights are assigned to the multivariate braking energy recovery evaluation factors according to the multiple key coefficients of the evaluation factors to establish a braking energy recovery evaluation weighted network; based on the multivariate braking energy recovery evaluation factors, a braking energy recovery efficiency prediction network, a braking energy recovery safety prediction network and a braking energy recovery stability prediction network are trained; the braking energy recovery efficiency prediction network, the braking energy recovery safety prediction network and the braking energy recovery stability prediction network are used as the first-level nodes of braking energy recovery evaluation; the braking energy recovery evaluation weighted network is used as the second-level node of braking energy recovery evaluation; the first-level node of braking energy recovery evaluation and the second-level node of braking energy recovery evaluation are connected to generate the braking energy recovery evaluation component.

[0072] Furthermore, the system is also used to implement the following functions: According to the q-node braking energy recovery scheduling space, the m-th braking energy recovery scheduling decision is randomly extracted, wherein m is a positive integer; based on the braking energy recovery evaluation component, the m-th braking energy recovery scheduling decision is subjected to recovery prediction evaluation to obtain the m-th braking energy recovery evaluation coefficient; according to the q-node braking energy recovery scheduling space, the m+1-th braking energy recovery scheduling decision is extracted, and the m+1-th braking energy recovery evaluation coefficient is calculated in combination with the braking energy recovery evaluation component; based on the m-th braking energy recovery evaluation coefficient and the m+1-th braking energy recovery evaluation coefficient, the m-th braking energy recovery scheduling decision and the m+1-th braking energy recovery scheduling decision are identified as superior to obtain the current superior recovery scheduling decision; based on the braking energy recovery evaluation component, the current superior recovery scheduling decision is continuously iterated and optimized according to the q-node braking energy recovery scheduling space until the number of iterative optimizations meets the recovery scheduling optimization number constraint, and the q-th node energy recovery optimization strategy is generated; and the q-th node energy recovery optimization strategy is added to the multiple node energy recovery optimization strategies.

[0073] Furthermore, the system is also used to implement the following functions: According to the multiple node operation plans, extract the qth node operation plan; based on the target rail train, modeling is performed to obtain a target train model; based on the qth node operation plan, braking energy recovery simulation is performed on the target train model according to the mth braking energy recovery scheduling decision to obtain an mth decision energy recovery simulation data set; the mth decision energy recovery simulation data set is input into the first-level braking energy recovery evaluation node to obtain the mth recovery efficiency prediction coefficient, the mth recovery safety prediction coefficient and the mth recovery stability prediction coefficient; the mth recovery efficiency prediction coefficient, the mth recovery safety prediction coefficient and the mth recovery stability prediction coefficient are input into the second-level braking energy recovery evaluation node to output the mth braking energy recovery evaluation coefficient.

[0074] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0075] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0076] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A method for optimizing braking energy consumption of a rail vehicle, characterized in that: The method comprises: According to the train operation control terminal, a predetermined operation plan of a target rail train is loaded, and the predetermined operation plan is decomposed into stations to generate multiple node operation plans, wherein the target rail train includes multiple braking energy recovery devices; Based on the multiple node operation schemes, predict the braking energy recovery potential of the target rail train according to the ARIMA-LSTM combined model to obtain the regenerative braking energy of multiple nodes; Based on the regenerative braking energy of the multiple nodes, respectively perform energy recovery scheduling on the multiple braking energy recovery devices to establish multiple node braking energy recovery scheduling spaces; Based on the multivariate braking energy recovery evaluation factors, a braking energy recovery evaluation component is constructed, wherein the multivariate braking energy recovery evaluation factors include braking energy recovery efficiency, braking energy recovery safety and braking energy recovery stability; Based on the braking energy recovery evaluation component, the plurality of node braking energy recovery scheduling spaces are respectively optimized and analyzed according to the recovery scheduling optimization number constraint to obtain the plurality of node energy recovery optimization strategies; The multiple node energy recovery optimization strategies are encrypted and sent to the train operation control terminal, and the train operation control terminal performs braking energy recovery on the target rail train according to the multiple braking energy recovery devices and the multiple node energy recovery optimization strategies.

2. The method according to claim 1, characterized in that Based on the multiple node operation schemes, the braking energy recovery potential of the target rail train is predicted according to the ARIMA-LSTM combined model to obtain the regenerative braking energy of multiple nodes, including: Perform load prediction on the target rail train based on the multiple node operation plans to obtain multiple node train predicted loads; Retrieving the braking energy recovery record of the target rail train to obtain a sample node operation plan set, a sample node train load set, and a sample node train recovery braking energy set; The ARIMA-LSTM combined model includes an ARIMA model and an LSTM model; Based on the sample node operation scheme set, the sample node train load set and the sample node train recovery braking energy set, the ARIMA model and the LSTM model are jointly learned to build a braking energy recovery potential prediction channel; The multiple node operation plans and the multiple node train predicted loads are input into the braking energy recovery potential prediction channel to obtain the multiple node regenerative braking energies.

3. The method according to claim 2, characterized in that Based on the sample node operation scheme set, the sample node train load set and the sample node train recovery braking energy set, the ARIMA model and the LSTM model are jointly learned to build a braking energy recovery potential prediction channel, including: Taking the sample node operation scheme set and the sample node train load set as input information and the sample node train recovery braking energy set as output information, the ARIMA model is supervised and trained to obtain a first recovery potential prediction model; Performing supervised training on the LSTM model according to the sample node operation scheme set, the sample node train load set, and the sample node train recovery braking energy set to obtain a second recovery potential prediction model; Collecting output data of the first recycling potential prediction model and the second recycling potential prediction model to obtain a fusion training data set; Training and testing are performed according to the fused training data set and the sample node train recovery braking energy set to obtain a recovery potential prediction fusion model; The first recovery potential prediction model and the second recovery potential prediction model are used as a first-level node for braking energy recovery potential prediction, and the recovery potential prediction fusion model is used as a second-level node for braking energy recovery potential prediction; The braking energy recovery potential prediction primary node and the braking energy recovery potential prediction secondary node are connected to generate the braking energy recovery potential prediction channel.

4. The method according to claim 1, characterized in that Based on the regenerative braking energy of the multiple nodes, respectively performing energy recovery scheduling on the multiple braking energy recovery devices, and establishing multiple node braking energy recovery scheduling spaces, including: Extracting the qth node regenerative braking energy according to the plurality of node regenerative braking energies, wherein q is a positive integer; Collecting energy storage characteristic parameters of the plurality of braking energy recovery devices to obtain an energy storage characteristic data set; Taking the regenerative braking energy of the qth node as the energy recovery scheduling target, making energy recovery decisions for the multiple braking energy recovery devices according to the energy storage characteristic data set, and obtaining the braking energy recovery scheduling space of the qth node; Collecting the decision quantity parameter of the qth node braking energy recovery scheduling space to obtain the qth space decision quantity; Determine whether the qth spatial decision quantity satisfies the decision quantity constraint; If the qth spatial decision quantity satisfies the decision quantity constraint, adding the qth node braking energy recovery scheduling space to the multiple node braking energy recovery scheduling spaces; If the qth spatial decision quantity does not satisfy the decision quantity constraint, based on the decision quantity constraint, the qth node braking energy recovery scheduling space is expanded according to the energy recovery scheduling target and the energy storage characteristic data set.

5. The method according to claim 1, characterized in that Based on the multivariate braking energy recovery evaluation factors, a braking energy recovery evaluation component is constructed, including: Performing importance evaluation according to the multivariate braking energy recovery evaluation factors to obtain multiple evaluation factor key coefficients; Allocating weights to the multivariate braking energy recovery evaluation factors according to the multiple evaluation factor key coefficients to establish a braking energy recovery evaluation weighted network; Based on the multivariate braking energy recovery evaluation factor, training a braking energy recovery efficiency prediction network, a braking energy recovery safety prediction network, and a braking energy recovery stability prediction network; The braking energy recovery efficiency prediction network, the braking energy recovery safety prediction network and the braking energy recovery stability prediction network are used as the first-level nodes for braking energy recovery evaluation; The braking energy recovery evaluation weighted network is used as a braking energy recovery evaluation secondary node; The braking energy recovery evaluation primary node and the braking energy recovery evaluation secondary node are connected to generate the braking energy recovery evaluation component.

6. The method according to claim 1, characterized in that Based on the braking energy recovery evaluation component, the plurality of node braking energy recovery scheduling spaces are respectively optimized and analyzed according to the recovery scheduling optimization number constraint, and a plurality of node energy recovery optimization strategies are obtained, including: According to the qth node braking energy recovery scheduling space, randomly extract the mth braking energy recovery scheduling decision, where m is a positive integer; Based on the braking energy recovery evaluation component, a recovery prediction evaluation is performed on the m-th braking energy recovery scheduling decision to obtain an m-th braking energy recovery evaluation coefficient; Extracting the m+1th braking energy recovery scheduling decision according to the qth node braking energy recovery scheduling space, and calculating the m+1th braking energy recovery evaluation coefficient in combination with the braking energy recovery evaluation component; Based on the mth braking energy recovery evaluation coefficient and the m+1th braking energy recovery evaluation coefficient, the mth braking energy recovery scheduling decision and the m+1th braking energy recovery scheduling decision are identified as superior to obtain a current superior recovery scheduling decision; Based on the braking energy recovery evaluation component, the current winning recovery scheduling decision is continuously iterated and optimized according to the q-th node braking energy recovery scheduling space until the number of iterative optimizations meets the recovery scheduling optimization number constraint, and the q-th node energy recovery optimization strategy is generated; The q-th node energy recovery optimization strategy is added to the multiple node energy recovery optimization strategies.

7. The method according to claim 6, characterized in that Based on the braking energy recovery evaluation component, a recovery prediction evaluation is performed on the m-th braking energy recovery scheduling decision to obtain the m-th braking energy recovery evaluation coefficient, including: Extracting the qth node operation plan according to the multiple node operation plans; Modeling is performed based on the target track train to obtain a target train model; Based on the qth node operation plan, performing a braking energy recovery simulation on the target train model according to the mth braking energy recovery scheduling decision to obtain an mth decision energy recovery simulation data set; Inputting the mth decision energy recovery simulation data set into the first-level node of braking energy recovery evaluation to obtain the mth recovery efficiency prediction coefficient, the mth recovery safety prediction coefficient and the mth recovery stability prediction coefficient; The mth recovery efficiency prediction coefficient, the mth recovery safety prediction coefficient and the mth recovery stability prediction coefficient are input into a braking energy recovery evaluation secondary node, and the mth braking energy recovery evaluation coefficient is output.

8. A braking energy consumption optimization system for rail train operation, characterized in that: The system comprises: A node operation plan generation module, wherein the node operation plan generation module loads a predetermined operation plan of a target rail train according to a train operation control terminal, and performs station decomposition on the predetermined operation plan to generate a plurality of node operation plans, wherein the target rail train includes a plurality of braking energy recovery devices; A recovery potential prediction module, wherein the recovery potential prediction module predicts the braking energy recovery potential of the target rail train based on the multiple node operation schemes and according to an ARIMA-LSTM combined model to obtain multiple node regenerative braking energies; An energy recovery scheduling module, wherein the energy recovery scheduling module performs energy recovery scheduling on the multiple braking energy recovery devices based on the regenerative braking energy of the multiple nodes, and establishes multiple node braking energy recovery scheduling spaces; An evaluation component construction module, wherein the evaluation component construction module constructs a braking energy recovery evaluation component based on a multivariate braking energy recovery evaluation factor, wherein the multivariate braking energy recovery evaluation factor includes braking energy recovery efficiency, braking energy recovery safety, and braking energy recovery stability; An optimization analysis module, wherein the optimization analysis module performs optimization analysis on the multiple node braking energy recovery scheduling spaces based on the braking energy recovery evaluation component and according to the recovery scheduling optimization number constraint to obtain multiple node energy recovery optimization strategies; A braking energy recovery module, wherein the braking energy recovery module encrypts the multiple node energy recovery optimization strategies and sends them to the train operation control terminal, and the train operation control terminal performs braking energy recovery on the target rail train according to the multiple braking energy recovery devices and the multiple node energy recovery optimization strategies.

Images