Capacity optimization configuration method and device for traction power supply system and medium

Abstract

The invention discloses a capacity optimization configuration method and device for a tractive power supply system and a medium, and relates to the field of capacity optimization configuration, and the method comprises the steps: obtaining original data; according to original data, a source-storage-load power characteristic model is established, according to the source-storage-load power characteristic model, a double-layer optimization configuration model for source-storage collaborative configuration is established, and the double-layer optimization configuration model for source-storage collaborative configuration is solved based on a multi-target dung beetle optimization algorithm improved based on a Kriging agent model and a non-dominated rapid sorting method. According to the method and the device, refined operation management can be realized, and the optimal planning scheme can be quickly obtained with high quality.

Description

Technical Field

[0001] The present application relates to the field of capacity optimization configuration, and in particular, to a method, device, and medium for optimizing the capacity configuration of a traction power supply system. Background Art

[0002] The capacity optimization configuration of a traction power supply system (TPSS) is a key link to ensure the stability of train operation and the reliability of power supply. With the rapid development of new energy technologies, renewable energy sources such as wind energy and photovoltaic energy are widely used in rail transit power systems to reduce dependence on traditional power grids and improve energy utilization efficiency. However, due to the intermittency and volatility of renewable energy, its integration with the traction power supply system still faces many challenges, such as decreased system reliability, unbalanced power supply and demand matching, and insufficient economic optimization.

[0003] Currently, the research on the optimization configuration of traction power supply systems mainly focuses on traditional power supply and demand balance optimization methods, but most of the solutions fail to fully consider the uncertainties after the access of new energy and the impact of carbon trading mechanisms on system operation. In addition, existing research often only focuses on a single optimization goal, such as minimizing investment costs or minimizing operating costs, lacking a systematic optimization method that takes into account the balance of multiple goals. Therefore, how to achieve the collaborative optimization configuration of wind power, photovoltaic power, energy storage, and traction substations while taking into account reliability and economy has become a technical problem that needs to be solved urgently in current research. At the same time, the problems of low solution efficiency and poor quality of optimization algorithms also make the proposed models unable to obtain reliable solutions.

[0004] Therefore, based on the above problems, there is an urgent need to provide a new method for optimizing the capacity configuration of a traction power supply system to achieve refined operation management and ensure the rapid and high-quality acquisition of the optimal planning scheme. Summary of the Invention

[0005] The purpose of the present application is to provide a method, device, and medium for optimizing the capacity configuration of a traction power supply system, which can achieve refined operation management and ensure the rapid and high-quality acquisition of the optimal planning scheme.

[0006] To achieve the above purpose, the present application provides the following solutions: In the first aspect, the present application provides a method for optimizing the capacity configuration of a traction power supply system, and the method for optimizing the capacity configuration of a traction power supply system includes: Obtain the original data; the original data includes: historical output data of wind and light resources, technical parameters of the energy storage system, and load characteristic data of the traction locomotive; the historical output data of wind and light resources includes: historical output data of wind speed and historical output data of light; the technical parameters of the energy storage system include: rated capacity, charge-discharge efficiency, and state of charge; the load characteristic data of the traction locomotive includes: train operation diagram, locomotive operation curve, and locomotive model; Based on the original data, establish a source-storage-load power characteristic model; the source-storage-load power characteristic model includes: source-side power characteristic model, energy storage system charge-discharge characteristic model, and traction substation load model; the source-side power characteristic model is used to determine the typical time-series power curves of wind and light according to the historical output data of wind and light resources; the energy storage system charge-discharge characteristic model is used to determine the energy storage power characteristic equation according to the technical parameters of the energy storage system; the traction substation load model is used to determine the load demand of the return line traction substation according to the load characteristic data of the traction locomotive; Based on the source-storage-load power characteristic model, establish a two-layer optimal configuration model for source-storage collaboration; the two-layer optimal configuration model for source-storage collaboration includes: planning layer model and operation layer model; the planning layer model optimizes the configuration schemes of wind power, photovoltaic power, energy storage system, and traction substation with the goal of minimizing the average daily equivalent annual value investment cost; the operation layer model takes minimizing the daily operation cost as the goal and introduces a carbon emission reward and punishment ladder-type trading mechanism for operation management; the carbon emission reward and punishment ladder-type trading mechanism is used to calculate the carbon trading cost; Based on the multi-objective dung beetle optimization algorithm improved by the Kriging surrogate model and the non-dominated fast sorting method, solve the two-layer optimal configuration model for source-storage collaboration to determine the optimal configuration scheme.

[0007] Optionally, the establishment of the source-storage-load power characteristic model according to the original data specifically includes: According to the historical output data of wind and light resources, use the WGAN-GP method and the improved hierarchical K-means clustering algorithm to determine the source-side power characteristic model; According to the source-side power characteristic model and the technical parameters of the energy storage system, based on the state of charge, charge-discharge efficiency, life impact, and power constraint conditions of the energy storage system, establish an energy storage power characteristic equation; and according to the energy storage power characteristic equation, determine the energy storage system charge-discharge characteristic model; According to the load characteristic data of the traction locomotive, based on the load characteristics of the traction locomotive, determine the traction substation load model.

[0008] Optionally, the determination of the source-side power characteristic model by using the WGAN-GP method and the improved hierarchical K-means clustering algorithm according to the historical output data of wind and light resources specifically includes: Train the historical output data of wind and solar resources using the WGAN-GP method to determine the output scenarios of wind and solar resources; Use the improved hierarchical K-means clustering algorithm to reduce the scenarios of the output scenarios of wind and solar resources, and determine the probabilities and clustering centers of the output scenarios of wind and solar resources after reduction; Determine the source-side power characteristic model according to the probabilities of the output scenarios of wind and solar resources after reduction and the clustering centers.

[0009] Optionally, the training of the historical output data of wind and solar resources using the WGAN-GP method to determine the output scenarios of wind and solar resources specifically includes: Use the formula to determine the output scenarios of wind and solar resources; where is the loss function of GAN, is the generator, is the discriminator, is the expectation, is the historical data, is the historical data 's probability distribution, is the discriminator's score for the historical data , is the noise signal, is the noise signal 's probability distribution, is the discriminator's score for the sample G(z) generated by the generator, is the weight hyperparameter of the gradient penalty, is the discriminator gradient L2 norm, is the interpolation sample located between the historical data sample and the generated sample .

[0010] Optionally, the traction substation load model specifically includes: According to the traction locomotive load characteristic data, the line terrain elevation map and the time range, determine the number of locomotives within the power supply range of each traction substation and the locomotive parameters of each locomotive; the locomotive parameters include: running speed, acceleration, uphill and downhill conditions and the corresponding ramp parameters; According to the locomotive parameters of each locomotive, conduct a force analysis on each locomotive at a certain moment to determine the total power consumed by all locomotives within the power supply range of each traction substation at a certain moment.

[0011] Optionally, the planning layer model specifically includes: Use the formula to determine the objective function of the planning layer model; Use the formula and formula Determine the constraint conditions of the planning layer model; wherein, is the average annual value investment cost per day, is the average annual investment cost of each electrical equipment, is the operating cost within a day, is the equal annuity factor of the equipment, is the purchase price per unit capacity of the equipment, superscript is the traction substation, superscript is the photovoltaic power generation equipment, superscript is the wind power generation equipment, superscript is the energy-type battery, superscript is the power-type battery, is the configured capacity of the traction transformer, is the configured capacity of the photovoltaic power generation, is the configured capacity of the wind power generation, is the configured capacity of the energy-type battery, is the configured capacity of the power-type battery, is the annual conversion factor of the life cycle cost, is the discount rate, is the working life of the equipment, is the optimization decision variable of the planning layer, is the lower limit of the installable area of the equipment, is the upper limit of the installable area of the equipment, is the area required for the equipment per unit capacity, is the lower limit of the installable capacity of the equipment, is the upper limit of the installable capacity of the equipment, is the configured capacity of the equipment.

[0012] Optionally, the operation layer model specifically includes: Use the formula to determine the objective function of the operation layer model; Use the formula to determine the power balance constraint condition; Use the formula to determine the grid interaction power output constraint condition; Use the formula and the formula to determine the wind and light output constraint conditions; Use the formula to determine the energy storage output constraint condition; where: is the operating cost within a day, is the electricity fee within a day, is the average daily operation and maintenance cost, is the intraday ladder carbon trading cost, is the intraday power grid power sales revenue, is the regional electricity price adjustment coefficient, is the 0-1 variable of power purchase and sales of local power supply companies, is the power purchase cost of CWPS-TPSS and local power supply companies, is the time number, is the total time, is the traction load at time is the time variation, is the demand charge of two-part tariff for CWPS-TPSS and local power supply companies, is the 0-1 variable of system power purchase and sales, is the power sales cost of CWPS-TPSS and local power supply companies, represents the set of devices, is the output cost per unit power of the device, is for each device at the output power at time is the optimization decision variable of the operation layer, is the photovoltaic power generation at time is the wind power generation at time is the output power of the energy storage device at time is the initial setting value of the state of charge of the energy storage device before operation, is the traction load power at time is the energy storage charging power at time is the energy storage discharging power at time is the configured capacity of the traction transformer, is the configured capacity of photovoltaic power generation, is the configured capacity of wind power generation, is the energy storage the discharging flag at time is the energy storage the charging flag at time is the energy storage the no-load flag at time is the configured capacity of the power-type battery, is the initial capacity of the energy storage device, is the end-of-day capacity of the energy storage device, is the energy storage energy at time is the energy storage device ​The capacity at a moment, is the charging efficiency, is the discharging efficiency, is the maximum coefficient of the allowable released energy for extending the energy storage life, is the minimum coefficient of the allowable released energy for extending the energy storage life, is the configured capacity of the energy-type battery, is the energy storage state of charge at a moment.

[0013] Optionally, the carbon emission reward and punishment ladder-type trading mechanism specifically includes: Using the formula to determine the carbon emission reward and punishment ladder-type trading mechanism; wherein, is the carbon trading cost, is the carbon trading price, is the growth rate of the carbon trading reward price, is the interval length of the carbon emissions, is the net carbon emissions in the carbon trading mechanism, is the growth rate of the carbon trading punishment price.

[0014] In a second aspect, the present application provides a computer device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the capacity optimization configuration method of the traction power supply system described in any one of the above.

[0015] In a third aspect, the present application provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the capacity optimization configuration method of the traction power supply system described in any one of the above are implemented.

[0016] According to the specific embodiments provided by the present application, the present application has the following technical effects: The present application provides a method, device, and medium for optimizing the capacity configuration of a traction power supply system, and constructs a two-layer optimization configuration model for source-storage collaborative configuration. Among them, the upper-layer planning layer model aims to minimize the average annual equivalent investment cost of wind power, photovoltaic power, traction substations, and energy storage devices, and optimizes the configuration schemes of wind power, photovoltaic power, traction substations, and energy storage devices; the lower-layer operation layer model takes into account both reliability and economy, uses the source-load volatility to measure the impact of the combined "wind-solar-storage" traction power supply system (Combined Wind-Solar Energy Storage Traction Power Supply System, CWPS-TPSS) on the reliability of the external power grid, and evaluates the support ability of the external power grid for CWPS-TPSS with the average load loss probability of off-grid locomotives. In addition, the optimization objectives of the operation layer model include minimizing the daily operation cost, and introducing a carbon emission reward and punishment ladder-type trading mechanism to achieve more refined operation management. To improve the solution efficiency, the present application uses a multi-objective dung beetle optimization algorithm improved by the Kriging surrogate model and the non-dominated quick sorting method to ensure that the optimal planning scheme can be obtained quickly and with high quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0018] Figure 1 Schematic diagram of the combined "wind-solar-storage" traction power supply system in an embodiment of the present application; Figure 2 Flow chart of a method for optimizing the capacity configuration of a traction power supply system in an embodiment of the present application; Figure 3 Step diagram of a method for optimizing the capacity configuration of a traction power supply system in an embodiment of the present application; Figure 4 Flow chart of the load simulation of the traction substation in an embodiment of the present application; Figure 5 Internal logical relationship diagram of the two-layer optimization configuration model for source-storage collaborative configuration in an embodiment of the present application; Figure 6 Step diagram of the improved multi-objective dung beetle optimization algorithm (Surrogate-Assisted Multi-Objective Dung Beetle Optimization Algorithm, SA-MODBO) in an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] 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.

[0020] 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.

[0021] This application aims to solve the prominent problems of large indirect carbon emissions and high energy costs of trunk railways under strong network lines. It combines photovoltaic, wind power and energy storage systems, and optimizes and integrates them with the existing open space resources of railways to build a Figure 1 CWPS-TPSS shown in the figure. In CWPS-TPSS, green electricity generated by renewable energy sources such as wind power and photovoltaic power can directly supply energy for locomotive traction through a single-phase DC-AC traction network, and can also be paid to feed back to the 110kV / 220kV external grid through an AC 27.5kV traction bus. In addition, green electricity can also be stored in an energy storage power station to achieve optimal energy allocation.

[0022] In an exemplary embodiment, Figure 2 and Figure 3 As shown, a method for optimizing the capacity configuration of a traction power supply system is provided, including the following S1-S4. Specifically, it includes: S1: Get the original data.

[0023] The original data include: historical output data of wind and solar resources, technical parameters of energy storage system and load characteristic data of traction locomotive; historical output data of wind and solar resources include: historical output data of wind speed and historical output data of light; technical parameters of energy storage system include: rated capacity, charging and discharging efficiency and state of charge (SOC); load characteristic data of traction locomotive include: train operation diagram, locomotive operation curve and locomotive model.

[0024] S2: Based on the original data, a source-storage-load power characteristic model is established.

[0025] The source-storage-load power characteristic model includes: source side power characteristic model, energy storage system charging and discharging characteristic model and traction substation load model.

[0026] S2 specifically includes: S21: Establish a source side power characteristic model.

[0027] When evaluating the natural scenery resources and modeling the power characteristics on the source side, the historical output data of the scenery resources are used as the input, and the power characteristic model on the source side is used as the output. To accurately characterize the source side output under different meteorological conditions, based on the evaluation of the natural scenery resources, the WGAN-GP scenario generation method is adopted to simulate and expand the historical output data of the scenery resources, and generate the scenery output scenarios under different meteorological conditions. Subsequently, the improved hierarchical K-means clustering algorithm is used for scenario reduction to construct the power characteristic model on the source side and extract the typical time series power curves of wind and light.

[0028] S21 specifically includes: S211: Use the WGAN-GP method to train the historical output data of the scenery resources to determine the scenery resource output scenarios; Scenery energy has intermittency and volatility. The Generative Adversarial Network (GAN) has powerful capabilities in generating high-dimensional data distributions and can be used for scenery scenario modeling. GAN consists of two deep neural network models, a generator G and a discriminator D. The generator processes the noise signal that follows a probability distribution by learning the latent distribution of the historical scenery output data to obtain the generated data that follows a probability distribution. Its goal is to make the probability distribution of the generated data approximate the probability distribution of the historical data. The discriminator judges the generated data of the generator and the historical data that follows a probability distribution and outputs the probability indicating that the generated data follows the true distribution. The two continuously optimize through mutual gaming to obtain a Nash equilibrium point and optimize themselves by maximizing the discrimination accuracy.

[0029] The loss function of GAN can be expressed as: .

[0030] Among them, is the generator, is the discriminator, is the expectation, is the historical data, is the historical data 's probability distribution, is the discriminator's score for the historical data , is the noise signal, is the noise signal 's probability distribution, usually the standard normal distribution is selected, is the discriminator's score for the sample G(z) generated by the generator.

[0031] However, traditional GAN training is prone to problems such as mode collapse and gradient disappearance, which affect the generation quality.

[0032] By introducing the Wasserstein distance to replace the JS divergence of the GAN, the training stability and convergence can be improved. The definition of the Wasserstein distance is as follows: .

[0033] Among them, is the probability distribution of historical data, is the probability distribution of generated data, sup is the supremum, that is, the least upper bound, is the weight parameter of the discriminator neural network.

[0034] Introducing gradient penalty to enhance Lipschitz continuity enables the model to converge faster, and its expression is: .

[0035] Among them, is the gradient penalty term, which is used to constrain the gradient norm of the discriminator to be close to 1 to satisfy the 1-Lipschitz condition, is the weight hyperparameter of the gradient penalty, which is used to control the influence degree of the gradient penalty term in the loss function, is the expected value of calculating the gradient penalty term on the batch data, is located at the historical data sample and the generated sample is the interpolation sample between them, is the discriminator for the input gradient, is the 2-norm, is the L2 norm (Euclidean norm) of the discriminator gradient, is the square of calculating the deviation between the gradient norm and 1, making the gradient norm closer to 1; among them, , is from sampled random numbers, used as interpolation weights, is the uniform distribution on [0,1].

[0036] At this time, the objective function of WGAN-GP is transformed into: .

[0037] Therefore, by training the WGAN-GP with the historical output data of wind and solar resources per unit power in a certain area, the resulting generator can generate wind and solar resource output scenarios that conform to the probability distribution of the real data in that area. This method can be applied to small-sample scenarios where the output data along the CWPS-TPSS is scarce, and can improve the representativeness and reliability of the data. By training the WGAN-GP with the historical output data of wind and solar resources per unit power in a certain area, the resulting generator can generate wind and solar resource output scenarios that conform to the probability distribution of the real data in that area. Provide data support for the subsequent two-layer optimal configuration model of source-storage collaborative configuration. Provide a typical wind-solar time series scenario set for the two-layer optimal configuration model of source-storage collaborative configuration, and improve the reliability of capacity configuration.

[0038] S212: Use the improved hierarchical K-means clustering algorithm to reduce the scenarios of the wind and solar resource output scenarios, and determine the probabilities and clustering centers of the reduced wind and solar resource output scenarios.

[0039] The number of wind and solar resource output scenarios generated by the WGAN-GP method in this application is huge and the calculation is complex. Therefore, it is necessary to reduce the operating scenarios and use a few typical scenarios to effectively describe the uncertainty characteristics. To avoid the defects of the traditional clustering algorithm's dependence on the number of target scenarios and subjectivity in determining the initial clustering values. This application proposes an improved hierarchical K-means clustering algorithm. First, use the maximum distance method to obtain the initial clustering values to avoid subjectivity brought by human experience; and use its advantage of not depending on the selection of the initial clustering center to autonomously determine the optimal number of clustering centers to reduce the source-load operating scenarios. Now, take the photovoltaic output as an example for illustration.

[0040] 1. Assume that a total of days of effective photovoltaic output data are obtained and normalized, that is, the original photovoltaic output data vector of the photovoltaic output of the th scenario is .

[0041] 2. Regard all the photovoltaic output data as an initial cluster.

[0042] 3. To prevent overfitting and underfitting, draw on the idea of the dichotomy method to dynamically divide the number of clusters. Use the K-means++ method to select the initial centroids. First, randomly select the first clustering center , secondly, for each data point in the dataset, calculate its shortest distance from the clustering center. Then select the next centroid according to the probability distribution , where each data point is selected as the centroid with a probability proportional to the probability distribution of its distance . The calculation formula is as follows: 。

[0043] Among them, is the th sample selected as the next clustering center, is the th distance between the sample and the clustering center. is the th distance between the sample and the clustering center, is the sample in the dataset except .

[0044] The calculation formula of .

[0045] Among them, is the selected centroid, is the set of centroids.

[0046] According to the calculated distance probability distribution, the second clustering center is selected in the way of weighted probability. Data points with a longer distance have a higher weight in the probability distribution, so as to ensure that the distribution of the initial clustering centers is more dispersed.

[0047] Since the number of clustering centroids K = 2 in the binary clustering and two centroids have been selected, the clustering process at this layer is completed.

[0048] 4. Since the Silhouette Coefficient (SC) has the advantage of simultaneously measuring the similarity (compactness) of a data point with other points in the same cluster and the similarity (separation) with the nearest neighboring cluster, SC is introduced as an index to measure the clustering effect. The closer the SC value is to 1, the better the clustering effect.

[0049] Let the number of iterations be , and calculate the th SC value . The calculation formula is as follows: .

[0050] Among them, is the number of days of photovoltaic power output, is the th index of the sample, is the average distance from the sample to other samples in the same cluster, is the average distance from the sample to all samples in the nearest other cluster.

[0051] and are expressed as follows: 。

[0052] 。

[0053] In the formula: is the number of data points in the cluster ; is the cluster with a different cluster number from other clusters except the cluster where is located; represents the Euclidean distance between the data point and the data point .

[0054] 5. Perform the next-level clustering. And let , to obtain the -th value.

[0055] 6. Define . When > , return to step 5, where is the relative gain of the silhouette coefficient, is the given threshold, which can be set according to the change curve of the clustering SC value; otherwise, output the clustering result.

[0056] 7. Obtain the probability of the reduced PV output scenario and the cluster center.

[0057] Finally, the probabilities of each scenario in the reduced typical output scenario set are weighted as weights to obtain the wind-solar typical, which can effectively represent the overall level of wind-solar output in CWPS-TPSS and be used in the subsequent solution of the source-storage collaborative configuration model, providing effective data support for the subsequent solution model.

[0058] S22: Establish the charge-discharge characteristic model of the energy storage system.

[0059] When modeling the charge-discharge characteristics of the energy storage system, the power characteristic model on the source side and the technical parameters of the energy storage system are used as inputs, and the charge-discharge characteristic model of the energy storage system is used as the output; based on the state of charge, charge-discharge efficiency, life impact, and power constraint conditions of the energy storage system, a power characteristic equation suitable for the optimization of the energy storage system is constructed.

[0060] S23: Establish the traction substation load model.

[0061] When modeling the traction substation load, the load characteristic data of the traction locomotive are used as inputs, and the traction substation load model is used as the output. Based on the load characteristics of the traction locomotive, a traction substation load model is established to ensure the load demand of the traction substation on the return line.

[0062] For the simulation of the CWPS-TPSS traction load, this application proposes an effective method for simulating the traction load under the railway ramp in high-altitude mountainous areas. The traction load simulation calculation process is as follows Figure 4 as shown. The specific steps are as follows.

[0063] 1. According to the traction locomotive load characteristic data, the line terrain elevation map, and the time range along the line, considering the acceleration, uniform speed, and deceleration processes of the locomotive operation, calculate the running speed and acceleration of each locomotive at each moment. 2. Count the number of operating locomotives, the running speed, acceleration, uphill and downhill conditions, and the corresponding ramp parameters of each locomotive within the power supply range of a traction substation based on the train operation diagram and the line terrain elevation map. 3. Considering the force condition of the locomotive on the ramp, when the locomotive is going uphill / downhill, it is subjected to the traction force / braking force , the basic resistance and the ramp additional force (the component of gravity along the ramp) , that is, the action of three forces.

[0064] Among them, the basic resistance The calculation formula is: .

[0065] In the formula: is the mass of the locomotive, with the unit of kg; is the unit basic resistance, with the unit of N; is the acceleration due to gravity, taking .

[0066] The ramp additional force The calculation formula is: .

[0067] In the formula, is the unit additional resistance of the locomotive group along the ramp, with the unit of N.

[0068] 4. Conduct a force analysis on the locomotive at a certain moment to obtain: .

[0069] In the formula: is the locomotive speed, with the unit of m / s, is the locomotive acceleration, with the unit of m / s 2 .

[0070] Therefore, the traction power when going uphill is: .

[0071] The braking power when going downhill is: .

[0072] 5. Solve the total power consumed by all locomotives within the power supply range of a traction substation at a certain moment (i.e., moment ), The total power is the traction load power at moment , and the calculation formula is: .

[0073] Among them, is the number of all locomotives within the power supply arm range of a traction substation at a certain moment; represents the power consumed by the th locomotive group,

[0074] To accurately characterize the source-side output under different meteorological conditions, this application proposes a WGAN-GP scenario generation method and a scenario reduction technology of an improved hierarchical K-means clustering algorithm based on the assessment of wind and solar natural resources to construct a source-side power characteristic model. On this basis, considering the state of charge of the energy storage system, a charge and discharge characteristic model of the energy storage system is established, and a load model of the traction substation is further constructed. The source-storage-load power characteristic model provides basic support for subsequent simulation operation and optimization.

[0075] S3: According to the source-storage-load power characteristic model, establish a two-layer optimal configuration model for source-storage coordination.

[0076] S3 specifically includes: S31: Establish a planning layer model.

[0077] When modeling the planning layer of the two-layer optimal configuration model for source-storage coordination, the source-side power characteristic model, the charge and discharge characteristic model of the energy storage system, and the load model of the traction substation are used as inputs, and the proposed capacity of each output device in the planning layer of the two-layer optimal configuration model for source-storage coordination is used as the output; the planning layer model optimizes the configuration schemes of wind power, photovoltaic power, energy storage system, and traction substation with the goal of minimizing the average daily equivalent annual investment cost.

[0078] The planning layer aims to minimize the average daily equivalent annual investment cost. The initial investment cost of the capacity configuration of wind, light, and storage includes construction (civil engineering, transportation, installation, etc.). The daily operation cost is returned to the planning layer model after being solved by the operation layer model. The objective function of the operation layer model is as follows: .

[0079] Among them, is the average daily equal annual value investment cost, with the unit of CNY / d; is the average annual investment cost of each electrical equipment, with the unit of CNY / y; is the operating cost within a day, with the unit of CNY; is the equal annuity factor of the equipment; is the purchase price per unit capacity of the equipment; superscript is the traction substation; superscript is the photovoltaic power generation equipment; superscript is the wind power generation equipment; superscript is the energy-type battery; superscript is the power-type battery; is the set of output equipment, ; is the configured capacity of the traction transformer; is the configured capacity of the photovoltaic power generation; is the configured capacity of the wind power generation; is the configured capacity of the energy-type battery, with the unit of CNY / MW or CNY / MWh; is the configured capacity of the power-type battery, is the annual conversion factor of the life cycle cost, is the discount rate, is the working life of the equipment, is the optimization decision variable of the planning layer.

[0080] The configuration of the source and storage capacity is restricted by space resources during planning. It can be distributed in suitable areas near the traction network. At the same time, considering the economic cost pressure, the installed capacity of the equipment is restricted. The calculation formula of the constraint conditions of the planning layer model is as follows: .

[0081] .

[0082] Among them. is the lower limit of the installable area of the equipment, is the upper limit of the installable area of the equipment, is the area required for the equipment per unit capacity, is the lower limit of the installable capacity of the equipment, is the upper limit of the installable capacity of the equipment, is the configured capacity of the equipment, with the unit of CNY / MW or CNY / MWh.

[0083] S32: Establish the operation layer model.

[0084] When modeling the operation layer of the two-layer optimal configuration model for source-storage collaborative configuration, the planning layer model is used as the input, and the output is the power output of each power generation device during the day under the capacity set in the planning layer model. Based on the planning layer model, an operation layer model is established, which combines the reliability and economy of both the supply and demand sides with multiple objectives. Two indicators, namely the source-load matching degree and the average loss probability of off-grid locomotive load, are introduced. The source-load volatility is used to measure the impact of CWPS-TPSS on the reliability of the external power grid, and the average load loss probability of off-grid locomotives is used to evaluate the support ability of the external power grid for CWPS-TPSS, so as to realize the integrated optimization of source-storage planning and operation and solve the optimal capacity configuration scheme. In addition, the operation optimization objectives include minimizing the daily operation cost, and a stepped trading mechanism for carbon emission rewards and penalties is introduced to achieve more refined operation management, so as to realize the dynamic optimization operation of CWPS-TPSS and ensure stable power supply under different load and weather conditions.

[0085] A stepped carbon trading cost calculation method is introduced in the operation layer model to further embed the carbon emission constraint into the operation optimization of CWPS-TPSS. Optimization is carried out not only in terms of economy and the reliability of both the supply and demand sides, but also taking into account the carbon emission constraint requirements of CWPS-TPSS operation, and a multi-objective operation optimization model as shown below is established.

[0086] The objective function of the operation layer model is as follows: 1. Minimize the daily operation cost: The daily operation cost consists of the daily electricity cost, the average daily operation and maintenance cost, the daily stepped carbon trading cost, and the daily power grid electricity sales revenue. The objective function of the operation layer model is as follows: .

[0087] Where: is the in-day operation cost, is the in-day electricity cost, is the average daily operation and maintenance cost, is the in-day stepped carbon trading cost, is the in-day power grid electricity sales revenue, is the regional electricity price adjustment coefficient, is a 0-1 variable for the local power supply company's power purchase and sale, is the power purchase cost between CWPS-TPSS and the local power supply company, that is, when purchasing power, the local power grid company transmits electric energy to the system through the traction substation, with the unit of CNY / MWd, is the time number, is the total time, is the traction load at time is the time variation, For the two-part tariff demand charge of CWPS-TPSS and the local power supply company, the unit is CNY / MWd, is the 0-1 variable of system power purchase and sale, that is, when selling electricity, the system transmits electric energy to the local power grid company through the traction substation, is the electricity selling cost of CWPS-TPSS and the local power supply company, the unit is CNY / MWd, is the output cost per unit power of the equipment, the unit is CNY / MWd, is the output power of each equipment at time t, is the optimization decision variable of the operation layer, is the photovoltaic power generation at time is the wind power generation at time is the output power of the energy storage equipment at time and is the initial set value of the state of charge SOC of the energy storage equipment before operation.

[0088] This application designs an incentive and penalty ladder-type trading mechanism for calculating the carbon emissions throughout the life cycle of low-carbon equipment , which is detailed as follows: .

[0089] Among them, is the carbon trading cost. When it is positive, it means that CWPS-TPSS needs to purchase carbon emission rights. When it is negative, it means that CWPS-TPSS can sell carbon emission rights to make a profit, is the carbon trading price, is the growth rate of the carbon trading reward price, is the interval length of carbon emissions, is the net carbon emissions in the carbon trading mechanism, is the growth rate of the carbon trading penalty price.

[0090] The net carbon emissions in the carbon trading mechanism are the carbon emissions of CWPS-TPSS minus the free carbon quota of CWPS-TPSS . The calculation formula for the free carbon quota of CWPS-TPSS is as follows: .

[0091] .

[0092] .

[0093] Among them, is the free carbon emission share per unit of electricity; 、 , and are the carbon emission factors for grid interaction, photovoltaic power generation, wind power generation, and energy storage operation, respectively.

[0094] The carbon emission factors per unit power generation of different power generation units are obtained by sorting relevant data, as shown in Table 1.

[0095] Table 1 Carbon emission factors per unit power generation of different power generation units

[0096] 2. The volatility of the source-load superposition curve is the smallest.

[0097] The minimum volatility index of the source-load superposition curve is established with reference to the characteristics of the average absolute value change rate of the source-load. It represents the volatility of the difference between the wind and light output curves and the traction load demand curve at adjacent moments. The larger the index calculation result, the greater the fluctuation between adjacent moments of the curve, the greater the impact on the external power supply network, and the greater the impact on the power supply reliability of the external power grid. The volatility calculation formula of the source-load superposition curve is as follows: .

[0098] Among them, is the source-load volatility, , and are the traction load, wind power generation, and photovoltaic power generation at time t + 1, respectively, with the unit of MW.

[0099] 3. The average load loss probability of off-grid locomotives is the smallest.

[0100] To avoid the occurrence of instability problems caused by sudden interruption of the transmission line and quantify the impact on the power supply reliability of the traction load. Based on this, this application proposes the average load loss probability of off-grid locomotives, and tries to reduce the power dependence of CWPS-TPSS on local power grid companies while ensuring economy. The calculation formula is as follows: .

[0101] The reliability index and are related as follows: .

[0102] The constraint conditions of the operation layer model are as follows: The constraint conditions of the operation layer model include power balance constraint, grid interaction power output constraint, wind and light output constraint, and energy storage output constraint.

[0103] 1. Power balance constraint.

[0104] The calculation formula of power balance constraint describes that the power supply must meet the power balance of demand. Any excess power will be sold to the power grid. On the other hand, in the case of power shortage, power needs to be purchased from the power grid; among them, the calculation formula of power balance constraint is as follows: 。

[0105] Among them: is the traction load power at time, when the value is positive, the locomotive is in traction mode, when the value is negative, the locomotive is in braking mode, and when the value is 0, the locomotive is in inert mode or there is no locomotive current taking in the power supply range, is the energy storage charging power at time, is the traction load at time, is the energy storage discharge power at time, is the photovoltaic power generation at time, is the wind power generation at time.

[0106] 2. Grid interaction power output constraint: The calculation formula of grid interaction power output constraint is as follows: 。

[0107] Among them, is the configured capacity of the traction transformer.

[0108] 3. Wind and solar power output constraint: The calculation formula of wind and solar power output constraint is as follows: 。

[0109] 。

[0110] Among them, is the configured capacity of photovoltaic power generation, is the configured capacity of wind power generation.

[0111] As can be seen from the above, the photovoltaic power output and the wind turbine power output are limited by the wind and light capacities at the planning level.

[0112] 4. Energy storage power output constraint: The calculation formula of energy storage power output constraint is as follows: 。

[0113] Among them, is the discharge flag of the energy storage at time, For energy storage is the charging flag at a certain moment, For energy storage is the no-load flag at a certain moment, is the configured capacity of the power-type battery, is the initial capacity of the energy storage device, is the end-of-day capacity of the energy storage device, is the energy storage energy at a certain moment, is the energy storage device the capacity at a certain moment, is the charging efficiency, is the discharging efficiency, is the maximum coefficient of the allowable released energy for extending the energy storage life, is the minimum coefficient of the allowable released energy for extending the energy storage life, is the configured capacity of the energy-type battery, For energy storage the state of charge at a certain moment.

[0114] The logical structure of the two-layer optimal configuration model for source-storage collaborative configuration is as Figure 5 shown. The model adopts a framework that combines upper-layer capacity configuration optimization and lower-layer operation scheduling optimization, and solves for the optimal capacity configuration results through iterative loops and two-way feedback. The upper-layer planning layer model aims to minimize the average annual equivalent investment cost of wind-solar-storage and traction substations, while considering the spatial resource constraints during the planning process. The decision variables are the configured capacities of traction substations, wind turbines, photovoltaics, the configured capacity and power of energy storage devices, and the initial SOC value of the energy storage device. The lower-layer operation layer model starts from two perspectives of reliability and economy, selects the source-load volatility as a measure of the impact of CWPS-TPSS on the reliability of the external power grid, and evaluates the impact of the external power grid on the reliability of CWPS-TPSS using the average load loss probability of off-grid locomotives. In addition, the optimization objectives of the operation layer include minimizing the daily operation cost, minimizing the source-load volatility, and minimizing the off-grid locomotive load loss probability. The decision variables are the intra-day operating powers of wind, light, and energy storage devices, the interactive power between the traction substation and the external power grid, and the initial SOC value of the energy storage device. The operation layer model designs a carbon emission reward and punishment ladder-type trading mechanism to achieve more refined operation management. At the same time, the model also fully considers the impact of the initial SOC value of the energy storage system on the overall economy.

[0115] The capacity configuration results determined by the algorithm of the planning layer model are input into the operation layer model. Then, the operation layer model combines the reduced wind-solar resource output scenarios and the load demand of the returned line traction substation to perform multi-objective optimization and solution on the intra-day operation layer model.

[0116] Generally speaking, the upper-level planning focuses on the overall investment cost during the project life cycle, while the lower-level operation layer is responsible for the daily scheduling optimization of CWPS-TPSS from the perspectives of power supply reliability and economy for both supply and demand sides. In this two-layer optimization framework, by embedding the GUROBI solver and combining with the multi-objective DBO algorithm based on the surrogate model, interactive optimization solution is achieved to obtain the lowest daily comprehensive cost, the optimal capacity configuration plan of wind-solar-storage equipment, and the optimal daily scheduling strategy of CWPS-TPSS.

[0117] S4: Solve the two-layer optimization configuration model of source-storage collaborative configuration based on the multi-objective dung beetle optimization algorithm improved by the Kriging surrogate model and the non-dominated fast sorting method to determine the optimal configuration plan.

[0118] When solving the two-layer optimization configuration model of source-storage collaborative configuration, take the two-layer optimization configuration model of source-storage collaborative configuration as the input and the optimal configuration plan and its objective function value as the output; propose an improved multi-objective dung beetle optimization algorithm based on the surrogate model, combine the dung beetle algorithm (DBO), the Latin hypercube sampling technique, the non-dominated fast sorting method and the Kriging surrogate model to improve the quality and efficiency of the optimization solution.

[0119] There are significant conflicts and mutual constraints among multiple optimization objectives in the two-layer optimization configuration model of source-storage collaborative configuration proposed in this application, making it difficult to simultaneously satisfy the optimal solutions of each objective. In addition, with the increase in the number of optimization objectives and the expansion of the decision variable scale, the computational complexity of the model rises sharply, resulting in a significant decline in the solution quality.

[0120] To address this challenge, DBO divides the population into four sub-populations: rolling ball, foraging, stealing, and breeding, which is more diverse than the single-population optimization strategy. In addition, DBO adopts a dynamic boundary strategy to reduce the search space, thus having stronger global optimization ability and fast convergence characteristics. At the same time, the surrogate model-assisted evolutionary algorithm has been proven to be one of the effective methods for solving high-computation-cost optimization problems.

[0121] In view of the above problems, this application proposes an improved multi-objective dung beetle optimization algorithm combined with a surrogate model. This algorithm introduces a surrogate model in the multi-objective dung beetle optimization process, significantly improving the efficiency and solution quality of solving complex optimization problems, thus more accurately capturing and approaching the Pareto front of the proposed model. SA-MODBO provides an efficient solution tool for multi-objective optimization problems, helping to improve the scientificity and accuracy of optimization decisions.

[0122] 1. Improvement strategy of the single-objective DBO optimization algorithm.

[0123] The traditional single-objective DBO optimization algorithm divides the dung beetles in the population according to four behaviors: dung ball rolling, reproduction, foraging, and stealing. They are divided in a ratio of 6:6:7:11 in a population. 20% of the individuals in the distribution represent those for dung ball rolling, 20% of the individuals are for reproduction, 25% of the individuals are small dung beetles that perform foraging behavior, and the remaining 35% of the dung beetles are defined as thieves and are used to perform stealing behavior.

[0124] In order to better complete the search process for the global position, enhance the ability of the dung ball rolling beetles to find the optimal solution, and thus help the dung beetle algorithm to jump out of the problem of falling into the local optimal solution, a new Cauchy-Lorentz random walk strategy is proposed.

[0125] This new Cauchy-Lorentz random walk strategy adds Cauchy-Lorentz mutation to the basic random walk strategy, and will generate a series of sequences that mutate around the population best value.

[0126] The formula for the new Cauchy-Lorentz random walk strategy is as follows: 。

[0127] Among them, Generate a random number with as the position parameter of the distribution peak position and as the size parameter of half the width at half of the maximum value, is the (n + 1)-th individual in the g-th iteration, is the n-th individual in the g-th iteration, is the optimal individual in the g-th iteration, is the scale parameter, and both obey uniformly distributed random numbers between.

[0128] Among them, the calculation formula for the scale parameter is as follows: 。

[0129] It can be seen from the above formula that by guiding the current dung beetle rolling behavior, new trajectories are generated near the dung beetle dung ball rolling trajectory, which is of great significance for updating the dung ball rolling trajectory and plays a crucial role in rapid convergence; at the same time, the scale parameter is an adaptive value, creating more possibilities to obtain the optimal solution and enhancing the ability to jump out of the local optimum. In order to ensure that the dung ball rolling beetles in the dung beetle algorithm walk within a certain range, it is necessary to uniformly normalize the walking data, and its calculation formula is as follows: 。

[0130] Among them, is the The minimum value of the individual Cauchy-Lorentz walk variable, is the maximum value of the individual Cauchy-Lorentz walk variable, is the individual's minimum value of the Cauchy-Lorentz walk variable at the g th iteration, is the individual's maximum value of the Cauchy-Lorentz walk variable at the g th iteration.

[0131] 2. Multi-objective DBO optimization algorithm.

[0132] To solve the proposed two-layer optimization configuration model for source-reservoir collaborative configuration, the non-dominated sorting method and crowding distance concept in NSGA-II are introduced, enabling the traditional single-objective DBO to utilize the Pareto dominance relationship to guide the population evolution.

[0133] Among them, the non-dominated quicksort method proposed in this application can reduce the time complexity from the traditional time complexity to , where is the number of objective functions, is the total population size. The fast non-dominated sorting process is as follows: (1) Calculate the dominance relationship and initialization: For all individuals pp in the population , the number of individuals that dominate is , and the set of individuals dominated by individual is . If , then individual is not dominated by any other individual. At this time, belongs to the first non-dominated layer, denoted as the set , and all individuals in this layer are defined as . The calculation formula is as follows: .

[0134] (2) Divide the non-dominated layers: Initialize the set to store the non-dominated solutions of the next layer; traverse each individual in . For each individual dominated by , update the number of individuals dominated by as . If the updated , it means that has become a new non-dominated solution, and add it to the set V; and define the set V as the non-dominated solution set of the next layer . Repeat the above process until all individuals in the population are assigned to the corresponding non-dominated layer. The calculation formula is as follows: .

[0135] (3) Repeated sorting: Repeat the second step until all individuals in the population are divided into the corresponding non-dominated layer .

[0136] The surrogate model is a simplified mathematical model that can approximately replace complex simulation models, used to reduce computational complexity and improve optimization efficiency. The Kriging model is a widely used surrogate model that predicts the response value of unknown observation points through the weighted average of existing sample points, has good non-linear approximation ability and unique error estimation function, and can meet the needs of multi-objective functions and large-scale optimization calculations. Based on this, the Kriging model is used to quickly seek the optimal equipment configuration capacity and the best operation strategy, significantly improving the efficiency of the optimization process.

[0137] To further improve the efficiency of the optimization process, this application generates uniformly distributed sample points based on Latin hypercube sampling, combines the main equipment configuration parameters with the optimization operation results, constructs a Kriging model, and analyzes the total economic cost of CWPS-TPSS, the volatility of the source-load superposition curve, and the probability index of the average load loss of off-grid locomotives within a certain equipment capacity configuration range. The Kriging model is selected as a tool to implicitly map the relationship between the capacities of wind, light, energy storage devices, and traction substations and the three objective functions due to its high prediction accuracy, strong robustness, and wide model applicability. Through the Kriging model, it is possible to more efficiently maintain the calculation accuracy while significantly shortening the calculation time and achieve a fast and accurate estimation of the objective function value. At the same time, in order to better balance the exploration and development process, the uncertainty of the Kriging prediction model is considered.

[0138] This application constructs a Kriging model using the DACE toolbox based on Matlab, providing strong support for complex system optimization and effectively simplifying the solution process of the three-objective optimization problem.

[0139] Aiming at the significant conflict and mutual restriction of each optimization objective in the two-layer optimization configuration model of source-storage collaborative configuration, this application proposes an improved multi-objective dung beetle optimization algorithm based on an auxiliary surrogate model. This algorithm is based on the dung beetle algorithm and combines Latin hypercube oversampling, non-dominated fast sorting method, and the Kriging surrogate model considering uncertainty to improve the solution quality and efficiency. The optimization algorithm proposed in this application is used to solve the established two-layer optimization configuration model of source-storage collaborative configuration to achieve the efficient optimization configuration of CWSS-TPSS.

[0140] In summary, as Figure 6 shown, the specific steps of the SA-MODBO optimization algorithm constructed in this application are as follows: 1. Based on the Latin hypercube sampling method, select the initial capacity planning combinations in CWSS-TPSS and establish an initial sample library.

[0141] 2. Call the selected initial capacity planning combinations to the operation layer model to optimize the operation process of CWSS-TPSS and obtain the decision results.

[0142] 3. Construct a Kriging surrogate model according to the data of the sample particles obtained by calculation and calculate the operation results of the planning layer.

[0143] 4. Use the SA-MODBO algorithm to update the capacity planning combinations, and further call the operation layer to optimize the operation process to obtain the optimization results of the corresponding operation layer decision variables and objective functions.

[0144] 5. Call the operation layer to optimize the operation process for the updated capacity planning combinations as the points to be selected for the capacity planning combinations; specifically: regard the new capacity planning combinations as the points to be selected, compare them with the objective function values in the initial sample library, and add them to the capacity planning combinations in the initial sample library if there is improvement.

[0145] 6. Repeat steps 3-5 until no new planning points are output. Search for the optimal planning scheme of the current Kriging model.

[0146] 7. Terminate the program, output the optimal planning scheme, that is, the Pareto front. At the same time, output the optimal scheduling strategy and multi-objective function values of the operation layer of CWSS-TPSS under this configuration.

[0147] In an exemplary embodiment, a computer device is provided. The computer device can be a server or a terminal. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, 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. Among them, 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 the capacity optimization configuration data of the traction power supply system. The input / output interface of the computer device is used to exchange information between the processor and external devices. 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, it implements a method for optimizing the capacity configuration of a traction power supply system.

[0148] In an exemplary embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0149] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by the processor, the steps in the above method embodiments are implemented.

[0150] 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 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 need to comply with relevant regulations.

[0151] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. 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 methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. 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), magnetoresistive 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 can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0152] The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0153] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, 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, it should be considered as the scope described in this specification.

[0154] Specific examples are used in this article to elaborate on the principles and implementation manners of the present application. The descriptions of the above embodiments are only used to help understand the method and its core idea of the present application; at the same time, for those of ordinary skill in the art, according to the idea of the present application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present application.

Claims

1. A capacity optimization configuration method for a traction power supply system, characterized in that: The capacity optimization configuration method of the traction power supply system comprises: Obtaining original data; original data includes: historical output data of wind and solar resources, technical parameters of energy storage system and load characteristic data of traction locomotive; historical output data of wind and solar resources includes: historical output data of wind speed and historical output data of light; technical parameters of energy storage system include: rated capacity, charging and discharging efficiency and state of charge; load characteristic data of traction locomotive includes: train operation diagram, locomotive operation curve and locomotive model; According to the original data, a source-storage-load power characteristic model is established; the source-storage-load power characteristic model includes: source-side power characteristic model, energy storage system charging and discharging characteristic model and traction substation load model; the source-side power characteristic model is used to determine the typical wind and light time series power curve according to the historical output data of wind and solar resources; the energy storage system charging and discharging characteristic model is used to determine the energy storage power characteristic equation according to the technical parameters of the energy storage system; the traction substation load model is used to determine the load demand of the return line traction substation according to the traction locomotive load characteristic data; According to the source-storage-load power characteristic model, a two-layer optimization configuration model for source-storage coordinated configuration is established; the two-layer optimization configuration model for source-storage coordinated configuration includes: a planning layer model and an operation layer model; the planning layer model optimizes the configuration scheme of wind power, photovoltaic, energy storage system and traction substation with the goal of minimizing the daily average annual investment cost; the operation layer model takes minimizing the daily operation cost as the goal, and introduces a carbon emission reward and punishment ladder trading mechanism for operation management; the carbon emission reward and punishment ladder trading mechanism is used to calculate the carbon trading cost; Based on the Kriging proxy model and the improved multi-objective dung beetle optimization algorithm of the non-dominated quick sorting method, the double-layer optimization configuration model of source-reservoir collaborative configuration is solved to determine the optimal configuration scheme.

2. The capacity optimization configuration method of the traction power supply system according to claim 1, characterized in that: The source-storage-load power characteristic model is established based on the original data, specifically including: According to the historical output data of wind and solar resources, the source-side power characteristic model is determined using the WGAN-GP method and the improved hierarchical K-means clustering algorithm; According to the source-side power characteristic model and the technical parameters of the energy storage system, an energy storage power characteristic equation is established based on the state of charge, charging and discharging efficiency, life impact and power constraints of the energy storage system; and according to the energy storage power characteristic equation, a charging and discharging characteristic model of the energy storage system is determined; According to the traction locomotive load characteristic data and based on the load characteristics of the traction locomotive, a traction substation load model is determined.

3. The capacity optimization configuration method of the traction power supply system according to claim 2, characterized in that: According to the historical output data of wind and solar resources, the source side power characteristic model is determined by using the WGAN-GP method and the improved hierarchical K-means clustering algorithm, which specifically includes: The WGAN-GP method is used to train the historical output data of wind and solar resources to determine the output scenarios of wind and solar resources; Using an improved hierarchical K-means clustering algorithm to reduce the wind and solar resource output scenario, and determine the probability and cluster center of the wind and solar resource output scenario after reduction; A source-side power characteristic model is determined according to the probability of the wind and solar resource output scenario after the reduction and the cluster center.

4. The capacity optimization configuration method of the traction power supply system according to claim 3 is characterized in that: The WGAN-GP method is used to train the historical output data of wind and solar resources to determine the output scenario of wind and solar resources, specifically including: Using the formula Determine the wind and solar resource output scenario; in, is the loss function of GAN, For the generator, is the discriminator, For expectations, For historical data, For historical data The probability distribution of For the discriminator to historical data The rating, is the noise signal, The noise signal The probability distribution of The score of the sample G(z) generated by the generator by the discriminator, is the weight hyperparameter of the gradient penalty, is the L2 norm of the discriminator gradient, For historical data samples and generate samples Interpolate samples between .

5. The capacity optimization configuration method of the traction power supply system according to claim 1, characterized in that: The traction substation load model specifically includes: According to the traction locomotive load characteristic data, the line topography and elevation map, and the time range, determine the number of locomotives within the power supply range of each traction substation and the locomotive parameters of each locomotive; the locomotive parameters include: running speed, acceleration, uphill and downhill conditions, and corresponding ramp parameters; According to the locomotive parameters of each locomotive, a stress analysis is performed on each locomotive at a certain moment to determine the total power consumed by all locomotives within the power supply range of each traction substation at a certain moment.

6. The capacity optimization configuration method of the traction power supply system according to claim 1, characterized in that: The planning layer model specifically includes: Using the formula Determine the objective function of the planning layer model; Using the formula and formula Determine the constraints of the planning level model; in, is the daily average annual investment cost, is the average annual investment cost of each electrical equipment, is the daily operating cost, is the equipment equal annuity coefficient, The unit capacity purchase price of the equipment, For traction substation, Photovoltaic power generation equipment, For wind power generation equipment, For energy type battery, For power type battery, Assign capacity to traction transformers, Allocate capacity for photovoltaic power generation, Allocate capacity for wind power generation, Configure capacity for energy type batteries, Configure capacity for power type batteries. is the annual conversion factor of the life cycle cost, is the discount rate, The working life of the equipment. Optimize decision variables for the planning layer, is the lower limit of the equipment installation area, is the upper limit of the equipment installation area. is the area required for equipment per unit capacity, is the lower limit of the equipment's installable capacity, The upper limit of the device's installable capacity. Configure capacity for the device.

7. The capacity optimization configuration method of the traction power supply system according to claim 1, characterized in that: The operation layer model specifically includes: Using the formula Determine the objective function of the operational layer model; Using the formula Determine power balance constraints; Using the formula Determine the grid interaction power output constraint conditions; Using the formula and formula Determine the constraints on wind and solar power output; Using the formula Determine energy storage output constraints; in: is the daily operating cost, For daily electricity charges, is the average daily operation and maintenance cost, is the intraday tiered carbon trading cost, is the revenue from electricity sales of the power grid within the day, is the regional electricity price adjustment coefficient, It is a 0-1 variable for purchasing and selling electricity for local power supply companies. The electricity purchase cost between CWPS-TPSS and local power supply companies, is the time number, is the total time, for The traction load at the moment, is the time variation, It is the demand electricity fee of the two-part electricity price system of CWPS-TPSS and local power supply companies. It is the 0-1 variable for purchasing and selling electricity in the system. is the electricity sales cost of CWPS-TPSS and local power supply companies, Represents a collection of devices. is the unit power equipment output cost, For each device Output power at the moment, Optimize decision variables for the operational layer, for The photovoltaic power generation at the time, for The wind power generation at the time, For energy storage equipment at all times The output power, It is the initial setting value of the state of charge before the energy storage device is put into operation. for The traction load power at the moment, for Energy storage charging power at all times, for The energy storage discharge power at the moment, Assign capacity to traction transformers, Allocate capacity for photovoltaic power generation, Allocate capacity for wind power generation, For energy storage The discharge mark of the moment, For energy storage Charging sign at the moment, For energy storage The empty sign at the moment, Configure capacity for power type batteries. is the initial capacity of the energy storage device, is the end-of-day capacity of the energy storage device, for Energy storage at all times, For energy storage devices The capacity of the moment, For charging efficiency, is the discharge efficiency, In order to extend the energy storage life, the maximum coefficient of energy release allowed is: In order to extend the energy storage life, the minimum coefficient of energy release allowed is: Configure capacity for energy type batteries, For energy storage State of charge at the moment.

8. The capacity optimization configuration method of the traction power supply system according to claim 1, characterized in that: The carbon emission reward and punishment ladder trading mechanism specifically includes: Using the formula Determine a tiered trading mechanism for carbon emissions rewards and penalties; in, is the carbon trading cost, is the carbon trading price, The increase in the price of carbon trading incentives, is the interval length of carbon emissions, is the net carbon emissions in the carbon trading mechanism, The increase in penalty prices for carbon trading.

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 capacity optimization configuration method for the traction power supply system according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the capacity optimization configuration method of the traction power supply system described in any one of claims 1 to 8 is implemented.

Images