An optimization scheduling method and system considering traction network resources, and an electronic device

Abstract

The application relates to the technical field of power system multi-source scheduling, and discloses an optimization scheduling method and system considering traction network resources, and an electronic device, which comprises the following steps: clustering analysis is performed on the data of each node of a power system to obtain extended power system data; a particle swarm algorithm is used to optimize a prediction model based on power market operation data to obtain a wind power prediction value; a Transformer model is used to extract features of photovoltaic power data to calculate a photovoltaic power prediction value; a spatial autocorrelation method is used to screen key station data to calculate a load power prediction value; a regulatable resource planning model is constructed based on the wind power prediction value, the photovoltaic power prediction value and the load power prediction value; and the regulatable resource planning model is iteratively optimized by taking multi-energy power generation and consumption time sequence information as a training set to obtain a final scheduling model. The application provides reliable technical support for the safe, stable and economic operation of a power system.

Description

Technical Field

[0001] This invention relates to the field of multi-source dispatching technology in power systems, and specifically to an optimized dispatching method, system, and electronic equipment that takes into account traction network resources. Background Technology

[0002] Power system dispatching is a crucial link in ensuring the safe, stable, and efficient operation of the power grid. It optimizes resource allocation and reduces system operating costs by rationally scheduling the start-up and shutdown of generating units, their output power, and power transmission routes. With the increasing proportion of renewable energy in the power system, the volatility and uncertainty of clean energy sources such as wind and solar power pose a significant challenge to the flexible supply and demand balance of the power system. This complexity is further exacerbated when special loads, such as those from railway traction networks, are integrated into the grid.

[0003] Currently, existing methods for assessing adjustable potential mainly focus on the user side, emphasizing the evaluation and analysis of user-adjustable loads. For wind and solar power generation, existing technologies primarily address their unpredictability and variability, but related assessment methods are often limited to the prediction models themselves, lacking a systematic assessment of the adjustable potential of wind and solar resources. Furthermore, due to the strong regional characteristics of wind and solar resources, prediction models from different regions are difficult to directly apply, which limits the applicability of assessment methods.

[0004] Existing technologies have significant shortcomings in the coordinated dispatch of traction grid resources and the power grid. Traditional dispatch methods often treat the traction grid as a rigid load, failing to fully consider its adjustment potential as a flexible resource. Particularly in multi-energy coordinated dispatch scenarios, there is a lack of effective mechanisms to organically combine traction grid resources with wind and solar energy consumption. This makes it difficult for the system to fully utilize the adjustable characteristics of the traction grid to balance the fluctuations in renewable energy, limiting the economic efficiency and flexibility of system operation.

[0005] Therefore, this application provides an optimized scheduling method that considers traction network resources to solve the above-mentioned technical problems. Summary of the Invention

[0006] The purpose of this invention is to provide an optimized scheduling method, system, and electronic equipment that takes into account traction network resources, in order to solve the technical problems in the prior art that the power system cannot achieve refined processing of missing data and that the flexible resources of the traction network are not effectively integrated to participate in the system's optimized scheduling.

[0007] To address the aforementioned technical problems, this invention provides an optimized scheduling method considering traction network resources, comprising:

[0008] Cluster analysis is performed on the data of each node in the power system to obtain multiple clusters. The power change rate of each cluster is calculated. Based on the power change rate, it is assessed whether the data point is an isolated value and the isolated value is set as a null value. Distance-based interpolation is performed on the null value to obtain the expanded power system data.

[0009] Based on electricity market operation data, input features for wind power prediction are selected, the redundancy between input features is calculated and redundant features are removed, and the prediction model is optimized using particle swarm optimization algorithm to obtain the predicted wind power value.

[0010] Based on electricity market operation data, the Transformer model is used to extract features from photovoltaic power data, attention weights are used to weight and sum the multi-head vectors, and the photovoltaic power prediction value is calculated by combining the induction conformal prediction method.

[0011] Based on electricity market operation data, the spatial autocorrelation method is used to screen key site data, the key site data is processed based on self-attention mechanism and spatiotemporal feature extraction process, and the load power prediction value is calculated by combining time series state information.

[0012] Based on the predicted wind power, photovoltaic power, and load power, an adjustable resource planning model is constructed, which includes an objective function and constraints. The objective function includes multiple revenue and cost terms, and the constraints include power constraints, energy consumption constraints, and time constraints.

[0013] Using the power generation and consumption time series information of multiple energy sources as a training set, the adjustable resource planning model is iteratively optimized to obtain the final scheduling model.

[0014] In some specific embodiments, cluster analysis is performed on the data of each node in the power system to obtain multiple clusters. The power change rate of each cluster is calculated. Based on the power change rate, it is assessed whether the data point is an isolated value and isolated values ​​are set as null values. Distance-based interpolation is performed on the null values ​​to obtain the expanded power system data, which further includes:

[0015] A clustering algorithm is used to divide the data into multiple clusters by calculating the distance between data points and cluster centers, and the cluster centers are updated to minimize the intra-cluster distance.

[0016] For each cluster, the forward rate of change and the backward rate of change are calculated as the power change rate, where the forward rate of change represents the ratio of the power difference between the current time and the next time, and the backward rate of change represents the ratio of the power difference between the current time and the previous time.

[0017] To assess whether a data point is an isolated value, the power change rate is compared with the confidence interval of the normal distribution. If the power change rate exceeds the confidence interval, the corresponding data point is marked as an isolated value and set to null.

[0018] For missing values, distance-based interpolation is performed. The interpolation weights are based on the distance between the element to be interpolated and each cluster, as well as the inverse distances of the nearest elements to the element to be interpolated. The missing values ​​are filled by weighted averaging.

[0019] In some specific embodiments, based on electricity market operation data, input features for wind power prediction are selected, the redundancy between input features is calculated and redundant features are removed, and the prediction model is optimized using a particle swarm optimization algorithm to obtain the wind power prediction value. This further includes:

[0020] Input features for wind power prediction include wind speed, humidity, temperature, and wind force, and historical data are collected as feature samples.

[0021] Calculate the redundancy between input features, quantify the correlation between features using mutual information, and set a threshold to remove redundant features.

[0022] The particle swarm optimization algorithm is used to optimize the prediction model. The model parameters, including the decision ratio coefficient, optimal location, input variable weights and biases, are adjusted by iteratively searching for the optimal parameter combination.

[0023] Based on the optimized prediction model, the predicted wind power value is output.

[0024] In some specific embodiments, based on electricity market operation data, a Transformer model is used to extract features from photovoltaic power data, attention weights are used to perform weighted summation of multi-head vectors, and the photovoltaic power prediction value is calculated by combining an inductive conformal prediction method. Further steps include:

[0025] The Transformer model is used to extract features from photovoltaic power data, mapping the input sequence to different dimensions of a multi-head vector to capture temporal dependencies;

[0026] Attention weights are used to perform a weighted summation of the multi-head vectors, and weights are assigned according to the similarity between vectors. The features are then fused to obtain the final representation of the vector features.

[0027] The photovoltaic power prediction value is calculated by combining the inductive conformal prediction method, the prediction interval is constructed by using the conformal prediction quantile, and the output of the Transformer model is integrated.

[0028] The prediction results are optimized based on the significance level and the number of calibration set samples. The prediction uncertainty is controlled by adjusting the confidence level, and the photovoltaic power prediction value is obtained.

[0029] In some specific embodiments, based on electricity market operation data, key site data is screened using a spatial autocorrelation method, processed using a self-attention mechanism and spatiotemporal feature extraction process, and load power prediction values ​​are calculated by combining time-series state information, further including:

[0030] The spatial autocorrelation method was used to screen data from key stations, calculate the correlation between observations under spatial and temporal lags, and select representative stations based on the correlation coefficient.

[0031] The system processes key site data based on self-attention mechanism and spatiotemporal feature extraction process, extracts spatial distribution and time series features, and enhances feature representation capability.

[0032] The load power prediction value is calculated by combining the temporal state information, and the hidden state and cell state of the long short-term memory network are used to capture dynamic changes and fuse the state of the previous time step.

[0033] The predicted load power is calculated using the weights and bias parameters of the output layer.

[0034] In some specific embodiments, based on the predicted wind power, photovoltaic power, and load power, an adjustable resource planning model is constructed, which includes an objective function and constraints. The objective function includes multiple revenue and cost terms, and the constraints include power constraints, energy consumption constraints, and time constraints. Further, it includes:

[0035] Construct an objective function that includes revenue from external electricity sales, revenue from internal electricity sales, costs during the planning period, and revenue and costs of traction network participation in dispatching. Maximize the total revenue by weighted summation.

[0036] Set constraints, including instantaneous power upper and lower limit constraints, average power and total time correlation constraints, minimum traction energy consumption constraints, power constraints obtained by the traction network from the power grid, rated power constraints of the traction converter, upper and lower limit constraints of train running time, and maximum allowable average power constraints.

[0037] Establish an interactive regulation mechanism between the flexible resources of the traction network and the consumption of wind and solar energy, and achieve multi-energy synergy through coordinated power allocation;

[0038] An adjustable resource planning model is built based on the objective function and constraints, and the prediction results are integrated to generate a scheduling scheme.

[0039] In some specific embodiments, using multi-energy generation and consumption time-series information as a training set, the adjustable resource planning model is iteratively optimized to obtain the final scheduling model, which further includes:

[0040] Using multi-energy generation and consumption time-series information as the training set, the adjustable resource planning model is input, including historical power generation and consumption data sequences.

[0041] Set the number of iterations as a hyperparameter for training to control the termination condition of the optimization process;

[0042] Heuristic optimization algorithms are used to iteratively optimize the model parameters to adapt to the mixed discrete and continuous variables contained in the model;

[0043] The final scheduling model is obtained through convergence judgment. The final scheduling model is output when the parameter change is less than the threshold or the maximum number of iterations is reached.

[0044] Based on the same concept, the present invention also provides an optimized scheduling system considering traction network resources, comprising:

[0045] The data preprocessing module is configured to perform cluster analysis on the data of each node in the power system to obtain multiple clusters, calculate the power change rate of each cluster, evaluate whether the data point is an isolated value based on the power change rate and set the isolated value as a null value, perform distance-based interpolation calculation on the null value, and obtain the expanded power system data.

[0046] The wind power prediction module is configured to select input features for wind power prediction based on electricity market operation data, calculate the redundancy between input features and remove redundant features, and use particle swarm optimization algorithm to optimize the prediction model to obtain the wind power prediction value.

[0047] The photovoltaic power prediction module is configured to use the Transformer model to extract features from the photovoltaic power data based on electricity market operation data, use attention weights to weighted summation of multi-head vectors, and combine inductive conformal prediction method to calculate the photovoltaic power prediction value.

[0048] The load power prediction module is configured to use spatial autocorrelation method to screen key site data based on electricity market operation data, process key site data based on self-attention mechanism and spatiotemporal feature extraction process, and calculate load power prediction value by combining time series status information.

[0049] The planning model construction module is configured to construct an adjustable resource planning model containing an objective function and constraints based on the wind power forecast, photovoltaic power forecast, and load power forecast. The objective function includes multiple revenue and cost items, and the constraints include power constraints, energy consumption constraints, and time constraints.

[0050] The model optimization module is configured to use multi-energy generation and consumption time-series information as a training set to iteratively optimize the adjustable resource planning model to obtain the final scheduling model.

[0051] Based on the same concept, the present invention also provides an electronic device, including: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of an optimized scheduling method that takes into account traction network resources.

[0052] Based on the same concept, the present invention also provides a computer-readable storage medium storing a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of an optimized scheduling method taking into account traction network resources.

[0053] Compared with existing technologies, its advantages are as follows:

[0054] This invention discloses an optimized scheduling method, system, and electronic equipment that considers traction network resources, which can improve data quality: by using a data preprocessing method of clustering followed by interpolation, it achieves full-dimensional, minute-level information expansion of missing values ​​in the power system. This not only effectively identifies and processes abnormal data points but also maintains the temporal characteristics and spatial correlation of the data, providing a high-quality data foundation for subsequent analysis and effectively avoiding evaluation bias caused by missing data.

[0055] Improved forecasting accuracy: Adaptive forecasting methods are employed to address the different characteristics of wind power, solar power, and load. Wind power forecasting combines particle swarm optimization with feature selection, solar power forecasting uses a combination of Transformer model and conformal forecasting, and load forecasting incorporates spatial autocorrelation and spatiotemporal feature extraction. These targeted measures enhance the accuracy and reliability of medium- and long-term forecasts, providing precise input data for assessing adjustable potential.

[0056] Multi-energy synergistic optimization: By constructing a planning model that considers the flexible resources of the traction grid, effective synergy between wind energy, solar energy, and traction grid resources is achieved. The model comprehensively considers economic objectives and operational constraints, establishes an interactive regulation mechanism for the traction grid's participation in new energy consumption, and improves the renewable energy consumption capacity and the economic efficiency of system operation.

[0057] Highly adaptable: Taking into full account the differences in resource characteristics in different regions, it can adapt to various operating scenarios through localized training and iterative optimization.

[0058] Improving operational efficiency: By refining the assessment of adjustable potential and optimizing dispatch strategies, it is possible to effectively reduce system operating costs, improve equipment utilization efficiency, and enhance the system's adaptability to renewable energy fluctuations, providing reliable technical support for the safe, stable, and economical operation of the power system. Attached Figure Description

[0059] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0060] Figure 1 This is a flowchart illustrating some specific embodiments of the optimized scheduling method for traction network resources according to the present invention.

[0061] Figure 2 This is one of the flowcharts of another embodiment of the optimized scheduling method for traction network resources according to the present invention;

[0062] Figure 3 This is a second flowchart illustrating another embodiment of the optimized scheduling method for traction network resources according to the present invention.

[0063] Figure 4 This is the third flowchart of another embodiment of the optimized scheduling method for traction network resources according to the present invention;

[0064] Figure 5 This is a schematic diagram of the structure of an optimized scheduling system for traction network resources according to some specific embodiments of the present invention;

[0065] Figure 6 This is a schematic diagram of the structure of an electronic device according to some specific embodiments of the present invention;

[0066] In the diagram, 710 is the processor; 720 is the memory; 730 is the input device; and 740 is the output device. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0068] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the application. The singular forms “a,” “said,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0069] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0070] It should be understood that although the terms first, second, third, etc., may be used in the embodiments of this application, these descriptions should not be limited to these terms. These terms are only used to distinguish the descriptions. For example, first may also be referred to as second without departing from the scope of the embodiments of this application, and similarly, second may also be referred to as first.

[0071] Depending on the context, the words “if” or “suppose” as used here can be interpreted as “when” or “in response to determination” or “in response to detection.” Similarly, depending on the context, the phrases “if determination” or “if detection (of the stated condition or event)” can be interpreted as “when determination” or “in response to determination” or “when detection (of the stated condition or event)” or “in response to detection (of the stated condition or event).”

[0072] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that an article or device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an article or device. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the article or device that includes said element.

[0073] It should be noted that any symbols and / or numbers present in the specification that are not marked in the accompanying drawings are not reference numerals.

[0074] Reference Figure 1 An optimized scheduling method considering traction network resources includes:

[0075] S101, perform cluster analysis on the data of each node in the power system to obtain multiple clusters, calculate the power change rate of each cluster, evaluate whether the data point is an isolated value based on the power change rate and set the isolated value as a null value, perform distance-based interpolation calculation on the null value to obtain the expanded power system data;

[0076] S102. Based on electricity market operation data, select input features for wind power prediction, calculate the redundancy between input features and remove redundant features, and use particle swarm optimization algorithm to optimize the prediction model to obtain the wind power prediction value.

[0077] S103, based on electricity market operation data, uses the Transformer model to extract features from photovoltaic power data, uses attention weights to weight and sum the multi-head vectors, and combines the induction conformal prediction method to calculate the photovoltaic power prediction value;

[0078] S104, based on electricity market operation data, uses spatial autocorrelation method to screen key site data, processes key site data based on self-attention mechanism and spatiotemporal feature extraction process, and calculates load power prediction value by combining time series state information;

[0079] S105. Based on the predicted wind power, photovoltaic power and load power, construct an adjustable resource planning model that includes an objective function and constraints. The objective function includes multiple revenue and cost items, and the constraints include power constraints, energy consumption constraints and time constraints.

[0080] S106, using the multi-energy power generation and consumption time sequence information as a training set, the adjustable resource planning model is iteratively optimized to obtain the final scheduling model.

[0081] Specifically, in this embodiment of the invention, cluster analysis is performed on the data of each node in the power system, dividing the data into multiple clusters. The power change rate of each cluster is calculated, including the forward change rate and the backward change rate. Based on these change rates, it is assessed whether the data points are isolated values, and the identified isolated values ​​are set as null values. Distance-based interpolation is performed on the null values, where the interpolation weight depends on the distance between the element to be interpolated and each cluster and the reciprocal of the distance to the nearest element, thereby obtaining the expanded power system data. Based on the power market operation data, input features for wind power prediction are selected, including wind speed, humidity, temperature, and wind force. The redundancy between these features is calculated, and redundant features are removed using the mutual information method. The particle swarm optimization algorithm is used to optimize the parameters of the prediction model, such as the decision ratio coefficient, optimal location, input variable weights, and bias, to obtain the predicted wind power value. For photovoltaic (PV) power forecasting, based on electricity market operation data, a Transformer model is used to extract features from the PV power data. A multi-head attention mechanism maps the input sequence to different dimensions of a multi-head vector. Attention weights are used to weight and sum the multi-head vectors to fuse features. This is then combined with an inductive conformal prediction method, using conformal prediction quantiles to construct the prediction interval, and integrating the Transformer model output to finally obtain the PV power forecast. For load power forecasting, based on electricity market operation data, a spatial autocorrelation method is used to screen key site data. The correlation between observations under spatial and temporal lags is calculated to select representative sites. Key site data is processed based on a self-attention mechanism and spatiotemporal feature extraction process to extract spatial distribution and time series features. Combined with temporal state information such as the hidden states and cell states of a long short-term memory network, the load power forecast is calculated. Based on predicted wind power, photovoltaic power, and load power, a controllable resource planning model is constructed. This model includes an objective function and constraints. The objective function includes revenue from external electricity sales, revenue from internal electricity sales, costs during the planning period, and revenue and costs of the traction network participating in dispatch. Total revenue is maximized through weighted summation. Constraints include upper and lower limits for instantaneous power, the correlation between average power and total time, minimum traction energy consumption, power constraints on the traction network's access to the grid, rated power constraints for traction converters, upper and lower limits for train running time, and maximum allowable average power constraints to ensure operational feasibility. Using multi-energy generation and consumption time-series information as a training set, the controllable resource planning model is iteratively optimized. Heuristic optimization algorithms are used to iteratively optimize the model parameters to adapt to the mixed discrete and continuous variables included in the model, gradually approaching the optimal solution to obtain the final dispatch model.

[0082] In some applications, cluster analysis is performed on the data of each node in the power system to obtain multiple clusters. The power change rate of each cluster is calculated, and based on the power change rate, data points are assessed to determine whether they are isolated values, which are then set as null values. Distance-based interpolation is performed on the null values ​​to obtain expanded power system data. This includes using clustering algorithms to divide the data into multiple clusters by calculating the distance between data points and cluster centers, and updating the cluster centers to minimize intra-cluster distances. For each cluster, the forward and backward change rates are calculated as the power change rate, where... The forward rate of change represents the ratio of the power difference between the current time and the next time, while the backward rate of change represents the ratio of the power difference between the current time and the previous time. Data points are assessed for isolation based on the power rate of change by comparing it with the confidence interval of a normal distribution. If the power rate of change exceeds the confidence interval, the corresponding data point is marked as an isolated value and set as a null value. Null values ​​are then filled using distance-based interpolation. The interpolation weights are based on the distance between the element to be interpolated and each cluster, as well as the reciprocal of the distances to the nearest elements to be interpolated. A weighted average is used to fill in the missing values.

[0083] Understandably, cluster analysis is performed on the data of each node in the power system. A clustering algorithm divides the data into multiple clusters by calculating the distance between data points and cluster centers, and iteratively updates the cluster centers to minimize intra-cluster distances. For each cluster, forward and backward rates of change are calculated as power change rates, where the forward rate of change represents the power difference ratio between the current and next time points, and the backward rate of change represents the power difference ratio between the current and previous time points. Data points are assessed for isolation based on their power change rates by comparing them with the confidence interval of a normal distribution. If the power change rate exceeds the confidence interval, the corresponding data point is marked as an isolated value and set as a null value. Distance-based interpolation is then performed on the null values. The interpolation weights are based on the distance between the element to be interpolated and each cluster, as well as the reciprocal of the distances to the nearest elements to be interpolated. A weighted average is used to fill in the missing values, ultimately obtaining the expanded power system data.

[0084] In some applications, based on electricity market operation data, input features for wind power prediction are selected, redundancy between input features is calculated and redundant features are removed, and the prediction model is optimized using a particle swarm optimization algorithm to obtain the predicted wind power value. This process includes selecting input features for wind power prediction, including wind speed, humidity, temperature, and wind force, and collecting historical data as feature samples; calculating the redundancy between input features, quantifying the correlation between features using mutual information methods, and setting thresholds to remove redundant features; optimizing the prediction model using a particle swarm optimization algorithm, adjusting model parameters including decision scaling factor, optimal location, input variable weights, and bias, and iteratively searching for the optimal parameter combination; and outputting the predicted wind power value based on the optimized prediction model.

[0085] Understandably, based on electricity market operation data, including electricity price factors, power generation information, and power consumption information, input features for wind power prediction are selected, including wind speed, humidity, temperature, and wind force, and historical data is collected as feature samples. The redundancy between input features is calculated, and the correlation between features is quantified using mutual information methods. A threshold is set to remove redundant features to simplify the model. Specific steps include: calculating the mutual information values ​​between all pairwise input features to construct a mutual information matrix; and setting a mutual information threshold. (For example =0.8), if the mutual information value between two features is greater than 0.8. If a feature pair is redundant, it is considered redundant. For redundant feature pairs, one is retained (usually the feature with higher mutual information to the target variable is retained), and the other is deleted. If the feature dimension is too high, principal component analysis or feature importance ranking can be used for dimensionality reduction. The particle swarm optimization algorithm is used to optimize the prediction model, adjusting the model parameters including the decision ratio coefficient, inertia weight, cognitive factor, and social factor. The optimal parameter combination is obtained through iterative search. Based on the optimized prediction model, the input features are processed, and the wind power prediction value is output.

[0086] In some applications, based on electricity market operation data, a Transformer model is used to extract features from photovoltaic power data. Attention weights are used to perform a weighted summation of multi-head vectors, and an inductive conformal prediction method is combined to calculate the predicted photovoltaic power value. This process includes: extracting features from photovoltaic power data using a Transformer model, mapping the input sequence to different dimensions of multi-head vectors to capture temporal dependencies; using attention weights to perform a weighted summation of multi-head vectors, assigning weights based on the similarity between vectors, and fusing features to obtain the final representation of vector features; calculating the predicted photovoltaic power value using an inductive conformal prediction method, constructing the prediction interval using conformal prediction quantiles, and integrating the Transformer model output; optimizing the prediction results based on the significance level and the number of calibration set samples, and controlling the prediction uncertainty by adjusting the confidence level to obtain the predicted photovoltaic power value.

[0087] Understandably, based on electricity market operation data, a Transformer model is used to extract features from photovoltaic power data, mapping the input sequence to different dimensions of multi-head vectors to capture temporal dependencies. Attention weights are used to weight and sum the multi-head vectors, assigning weights based on the similarity between vectors, and fusing features to obtain the final representation of vector features. The photovoltaic power prediction value is then calculated using an inductive conformal prediction method. Conformal prediction is performed after the Transformer model output, and the specific steps include: using the Transformer model to obtain a preliminary photovoltaic power prediction sequence; and calculating the inconsistency score based on the calibration set. ;in, For the first The actual photovoltaic power value of each sample For the Transformer prediction model, set the significance level. (Usually taken as 0.05 or 0.1), calculate the conformal prediction quantile. Constructing prediction intervals The prediction results are optimized based on the significance level and the number of calibration set samples. The prediction uncertainty is controlled by adjusting the confidence level, and the final photovoltaic power prediction value is obtained.

[0088] In some applications, based on electricity market operation data, a spatial autocorrelation method is used to screen key site data. This key site data is then processed using a self-attention mechanism and spatiotemporal feature extraction process. Finally, load power predictions are calculated by combining time-series state information. This process includes using a spatial autocorrelation method to screen key site data, calculating the correlation between observations under spatial and temporal lags, and selecting representative sites based on the correlation coefficient; processing key site data using a self-attention mechanism and spatiotemporal feature extraction process to extract spatial distribution and time-series features, enhancing feature representation capabilities; calculating load power predictions by combining time-series state information, using the hidden states and cell states of a long short-term memory network to capture dynamic changes, and fusing the state from the previous time step; and calculating load power predictions through the weights and bias parameters of the output layer.

[0089] Understandably, based on electricity market operation data, a spatial autocorrelation method is used to screen key site data. By calculating the correlation index of observed values ​​under spatial distance lag and time interval lag conditions, representative key sites are selected based on the magnitude of the correlation coefficient. The selected key site data is processed based on a self-attention mechanism and a spatiotemporal feature extraction process. Spatial distribution features and temporal evolution features are extracted through spatial dimension feature interaction and time series pattern learning to enhance feature representation capabilities. Load power prediction values ​​are calculated by combining time-series state information. The hidden state vector and cell state vector of the long short-term memory network are used to capture the dynamic change characteristics of the system, and the state information of the previous time step is fused. The features are mapped to load power prediction values ​​through the weight matrix and bias parameters of the output layer.

[0090] In some applications, based on the predicted wind power, photovoltaic power, and load power, an adjustable resource planning model is constructed, which includes an objective function and constraints. The objective function includes various revenue and cost terms, and the constraints include power constraints, energy consumption constraints, and time constraints. This involves constructing the objective function, which includes revenue from external electricity sales, revenue from internal electricity sales, costs during the planning period, and revenue and costs of the traction network participating in dispatch, maximizing total revenue through weighted summation; setting constraints, including upper and lower limits for instantaneous power, constraints on the correlation between average power and total time, minimum traction energy consumption constraints, power constraints obtained by the traction network from the grid, rated power constraints for traction converters, upper and lower limits for train running time, and maximum allowable average power constraints; establishing an interactive adjustment mechanism between the traction network's flexible resources and wind and solar energy consumption, achieving multi-energy synergy through coordinated power allocation; and building an adjustable resource planning model based on the objective function and constraints, integrating the prediction results to generate a dispatch scheme.

[0091] Understandably, based on wind power forecasts, photovoltaic power forecasts, and load power forecasts, an adjustable resource planning model is constructed, including an objective function and constraints. The objective function includes revenue from external electricity sales, revenue from internal electricity sales, costs during the planning period, and revenue and costs of the traction network participating in dispatch, maximizing total revenue through a weighted summation method. Constraints include upper and lower limits for instantaneous power, constraints on the correlation between average power and total time, minimum traction energy consumption constraints, power constraints on the traction network's acquisition from the grid, rated power constraints for traction converters, upper and lower limits for train running time, and maximum allowable average power constraints. An interactive adjustment mechanism is established between the traction network's flexible resources and wind and solar energy consumption, achieving multi-energy coordinated operation by coordinating power allocation among various energy sources. Based on the above objective function and constraints, an adjustable resource planning model is built, integrating wind power forecasts, photovoltaic power forecasts, and load power forecasts to generate the optimal dispatch scheme.

[0092] In some applications, the adjustable resource planning model is iteratively optimized using multi-energy generation and consumption time-series information as a training set to obtain the final scheduling model. This includes inputting multi-energy generation and consumption time-series information, including historical generation and consumption data sequences, into the adjustable resource planning model; setting the number of iterations as a hyperparameter for training to control the termination condition of the optimization process; using a heuristic optimization algorithm to iteratively optimize the model parameters to adapt to the mixed discrete and continuous variables contained in the model; and obtaining the final scheduling model through convergence judgment, outputting the final scheduling model when the parameter change is less than a threshold or the maximum number of iterations is reached.

[0093] Understandably, the training set is composed of time series information on power generation and consumption from multiple energy sources, which is input into the adjustable resource planning model. This includes time series data consisting of historical power generation data and power load data. The number of iterations is set as a hyperparameter for model training to control the termination condition of the optimization process. Heuristic optimization algorithms (such as improved particle swarm optimization or genetic algorithms) are used to iteratively optimize the model parameters to adapt to the mixed discrete and continuous variables contained in the model. The final scheduling model is obtained through convergence judgment. When the update amount of the model parameters is less than the set threshold or the preset maximum number of iterations is reached, the training process is terminated and the final scheduling model is output.

[0094] The following is combined with Figures 2-4 This invention describes another embodiment of an optimized scheduling method for traction network resources:

[0095] like Figure 2 As shown, this embodiment includes:

[0096] Based on data information from various nodes of the power system, clustering is performed first and then interpolation is performed to achieve full-dimensional, minute-level information expansion of missing values ​​in the power system.

[0097] Taking into account electricity price factors, power generation information, and power consumption information, an adaptation method is used to make medium- and long-term forecasts for sources and loads respectively, thereby improving the accuracy of adjustable resource potential assessment.

[0098] An adjustable resource planning model was built, and an interactive regulation method for the participation of flexible resources of the traction network in wind and solar energy consumption was proposed. Using the time series information of multi-energy generation and consumption as the training set, the overall model was continuously optimized in the iteration to obtain the optimal scheduling model.

[0099] like Figure 3 As shown, based on data information from each node of the power system, clustering is performed followed by interpolation to achieve full-dimensional, minute-level information augmentation of missing values ​​in the power system. First, the K-means method is used to cluster the data, with the following formula:

[0100] ;

[0101] in, Let t represent the t-th cluster center. This represents the number of objects in the t-th cluster. This represents the i-th object in the t-th cluster.

[0102] Based on each different clustering result, the cluster power rate of change (CPCR) method is used to obtain the forward rate of change of the algorithm's computation time t relative to adjacent time points. and backward rate of change Based on relevant theories related to the normal distribution, the state of this point as an isolated value is evaluated:

[0103] ;

[0104] in, It represents the forward rate of change at time t. Represents the backward rate of change at the same moment. This represents the power value at time t.

[0105] After obtaining the rate of change, the mean and standard deviation within the normal distribution range are then calculated according to the equation:

[0106] ;

[0107] in, It is defined as the average of the long-term rate of change. The standard deviation of the long-term rate of change It is defined as the average of the callback change rate. The standard deviation of the rate of change of the pullback is represented by N, and N represents the total number of data points. This leads to the judgment... and Check if the data exceeds the confidence interval; if so, set the data to empty. Finally, fill in the interpolation values. The interpolation weight calculation formula is:

[0108] ;

[0109] Among them, among them, Indicates the first Interpolation weights for each cluster, This represents the Euclidean distance between the point to be interpolated and the center of the cluster. Indicates the first The reciprocal of the distance between each cluster, This represents the number of nearest neighbor clusters.

[0110] like Figure 4 As shown, considering electricity price factors, power generation information, and power consumption information, an adaptation method is used to predict source and load separately, improving the accuracy of adjustable resource assessment. Adjustable resources are assessed by category. For wind power, candidate input prediction features are... for:

[0111] ;

[0112] in, , , , Let represent the wind force, wind speed, humidity, and temperature at time t, respectively. , , and These represent the upper bounds of the previous values ​​of the input variables (i.e., wind speed, humidity, temperature, and wind force), respectively.

[0113] After removing unnecessary features, the wind power prediction formula based on the remaining features is as follows:

[0114] ;

[0115] in, It is the decision ratio coefficient of the particle swarm optimization algorithm. This is the optimal position for the particle swarm optimization algorithm. These are the weights of the input variables. is the mapping function of the input features, and b is the bias.

[0116] Due to the large fluctuations in photoelectric power, feature extraction is difficult. Therefore, a Transformer-based feature extraction method is adopted, and the calculation formula is as follows:

[0117] ;

[0118] in, , These are the two different dimensions of the multi-head vectors derived from the decomposition. Scaling factor For the first Key vectors The transpose operation.

[0119] The vector is weighted and summed using attention weights to obtain the final representation of the current vector's features:

[0120] ;

[0121] in, The attention weights obtained during training, This is the third dimension of the multi-head vector derived from the decomposition.

[0122] The overall formula for the ICP (Inductive Conformal Prediction) method based on the Transformer is shown below:

[0123] ;

[0124] in, This is the predicted photovoltaic power at time h in the future. Let i be the actual photovoltaic power value of the i-th sample. For Transformer prediction models, Given the input sequence, Here, is the conformal prediction quantile, and nn is the number of samples in the calibration set. The significance level is indicated by .

[0125] To address the dispersed nature of the load data, the SA (spatial autocorrelation) method is first used to extract the different values ​​of features at different locations. The calculation formula is as follows:

[0126] ;

[0127] in, and In position and and time and The observed value, h is the spatial lag, representing the distance between the two locations. Time lag indicates the time difference between two observations. It consists of spatial distance h and time interval The number of separated observation pairs.

[0128] The formula for calculating the adjustable resource assessment method for load is as follows:

[0129] ;

[0130] in, Let be the predicted load power at time t. For key site data selected based on spatial autocorrelation, For self-attention mechanism, This is the core module for spatiotemporal feature extraction. and This refers to the hidden state and cell state from the previous moment. and These are the weights and bias parameters for the output layer.

[0131] A flexible resource planning model is constructed, and an interactive regulation method for the participation of flexible resources in the traction network in wind and solar energy consumption is proposed. Using multi-energy generation and consumption time-series information as the training set, the overall model is continuously optimized through iteration to obtain the optimal scheduling model. The overall objective function and constraints are as follows:

[0132] ;

[0133] in, Revenue from the sale of electricity to external parties, For the revenue from selling electricity domestically, Costs over the entire planning period and These represent the revenue and costs of the traction network participating in power system dispatching, respectively.

[0134] The power balance constraint equations for the power grid are shown below:

[0135] ;

[0136] In the formula, , , and It refers to the output of gas turbines, new energy units, demand response loads, and batteries during time period t. This represents the load power of the power grid at time t. The last two formulas represent the upper and lower limits of instantaneous power and the correlation constraint between average power and total time, respectively. The minimum traction energy required for a train to complete a specific journey. Let t be the power that the traction network draws from the power grid at time t. The rated power of the traction converter, These are the upper and lower limits of train travel time. This represents the maximum permissible average power.

[0137] Set the number of iterations, calculate the optimal solution, and obtain the optimal scheduling model.

[0138] For the purpose of simplicity, the method steps disclosed in the above embodiments are described as a series of actions. However, those skilled in the art should understand that the embodiments of the present invention are not limited to the described order of actions, because according to the embodiments of the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions involved are not necessarily essential to the embodiments of the present invention.

[0139] like Figure 5 As shown, the present invention also provides an optimized scheduling system considering traction network resources, comprising:

[0140] The data preprocessing module 201 is configured to perform cluster analysis on the data of each node of the power system to obtain multiple clusters, calculate the power change rate of each cluster, evaluate whether the data point is an isolated value based on the power change rate and set the isolated value as a null value, perform distance-based interpolation calculation on the null value, and obtain the expanded power system data.

[0141] The wind power prediction module 202 is configured to select input features for wind power prediction based on electricity market operation data, calculate the redundancy between input features and remove redundant features, and use particle swarm optimization algorithm to optimize the prediction model to obtain the wind power prediction value.

[0142] The photovoltaic power prediction module 203 is configured to extract features from photovoltaic power data using the Transformer model based on electricity market operation data, use attention weights to weight and sum the multi-head vectors, and combine the inductive conformal prediction method to calculate the photovoltaic power prediction value.

[0143] The load power prediction module 204 is configured to use spatial autocorrelation method to screen key site data based on electricity market operation data, process key site data based on self-attention mechanism and spatiotemporal feature extraction process, and calculate load power prediction value by combining time series status information.

[0144] The planning model construction module 205 is configured to construct an adjustable resource planning model containing an objective function and constraints based on the wind power forecast, photovoltaic power forecast and load power forecast. The objective function includes multiple revenue and cost items, and the constraints include power constraints, energy consumption constraints and time constraints.

[0145] The model optimization module 206 is configured to use multi-energy power generation and consumption time series information as a training set to iteratively optimize the adjustable resource planning model to obtain the final scheduling model.

[0146] It is worth noting that although only some basic functional modules are disclosed in the embodiments of this invention, it does not mean that the composition of this system is limited to the above-mentioned basic functional modules. On the contrary, what this embodiment intends to express is that, based on the above-mentioned basic functional modules, those skilled in the art can arbitrarily add one or more functional modules in combination with existing technology to form an infinite number of embodiments or technical solutions. That is to say, this system is open rather than closed. The fact that this embodiment only discloses a few basic functional modules should not be considered as the scope of protection of the claims of this invention being limited to the disclosed basic functional modules. At the same time, for the convenience of description, the above device is described separately according to its functions as various units and modules. Of course, in implementing this invention, the functions of each unit and module can be implemented in one or more software and / or hardware.

[0147] like Figure 6 As shown, the present invention also provides an electronic device, including: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of an optimized scheduling method considering traction network resources.

[0148] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. For example... Figure 6 The structure shown in this embodiment of the invention includes an electronic device comprising one or more processors 710 and a memory 720; the processors 710 in this electronic device may be one or more. Figure 6Taking a processor 710 as an example; a memory 720 is used to store one or more programs; the one or more programs are executed by the one or more processors 710, so that the one or more processors 710 implement an optimized scheduling method considering traction network resources as described in any one of the embodiments of the present invention.

[0149] The electronic device may also include an input device 730 and an output device 740.

[0150] The processor 710, memory 720, input device 730, and output device 740 in this electronic device can be connected via a bus or other means. Figure 6 Taking the example of a connection between China and Israel via a bus.

[0151] The memory 720 in this electronic device serves as a computer-readable storage medium, capable of storing one or more programs. These programs can be software programs, computer-executable programs, or modules, such as the program instructions / modules corresponding to the optimized scheduling method considering traction network resources provided in this embodiment of the invention. The processor 710 executes various functional applications and data processing of the electronic device by running the software programs, instructions, and modules stored in the memory 720, thereby implementing the optimized scheduling method considering traction network resources described in the above embodiment.

[0152] The memory 720 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the electronic device. Furthermore, the memory 720 may include high-speed random access memory and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 720 may further include memory remotely located relative to the processor 710, which can be connected to the device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0153] Input device 730 can be used to receive input digital or character information, and to generate key signal inputs related to user settings and function control of the electronic device. Output device 740 may include display devices such as a display screen.

[0154] The present invention also provides a computer-readable storage medium storing a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of an optimized scheduling method taking into account traction network resources.

[0155] Specifically, the computer storage medium in this embodiment of the invention can be any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. For example, a computer-readable storage medium can be—but is not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0156] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An optimized scheduling method considering traction network resources, characterized in that, include: Cluster analysis is performed on the data of each node in the power system to obtain multiple clusters. The power change rate of each cluster is calculated. Based on the power change rate, it is assessed whether the data point is an isolated value and the isolated value is set as a null value. Distance-based interpolation is performed on the null value to obtain the expanded power system data. Based on electricity market operation data, input features for wind power prediction are selected, the redundancy between input features is calculated and redundant features are removed, and the prediction model is optimized using particle swarm optimization algorithm to obtain the predicted wind power value. Based on electricity market operation data, the Transformer model is used to extract features from photovoltaic power data, attention weights are used to weight and sum the multi-head vectors, and the photovoltaic power prediction value is calculated by combining the induction conformal prediction method. Based on electricity market operation data, the spatial autocorrelation method is used to screen key site data, the key site data is processed based on self-attention mechanism and spatiotemporal feature extraction process, and the load power prediction value is calculated by combining time series state information. Based on the predicted wind power, photovoltaic power, and load power, an adjustable resource planning model is constructed, which includes an objective function and constraints. The objective function includes multiple revenue and cost terms, and the constraints include power constraints, energy consumption constraints, and time constraints. Using the power generation and consumption time series information of multiple energy sources as a training set, the adjustable resource planning model is iteratively optimized to obtain the final scheduling model.

2. The optimized scheduling method considering traction network resources according to claim 1, characterized in that, Cluster analysis is performed on the data from each node of the power system to obtain multiple clusters. The power change rate of each cluster is calculated. Based on the power change rate, it is assessed whether the data point is an isolated value and isolated values ​​are set as null values. Distance-based interpolation is performed on the null values ​​to obtain the expanded power system data, which further includes: A clustering algorithm is used to divide the data into multiple clusters by calculating the distance between data points and cluster centers, and the cluster centers are updated to minimize the intra-cluster distance. For each cluster, the forward rate of change and the backward rate of change are calculated as the power change rate, where the forward rate of change represents the ratio of the power difference between the current time and the next time, and the backward rate of change represents the ratio of the power difference between the current time and the previous time. To assess whether a data point is an isolated value, the power change rate is compared with the confidence interval of the normal distribution. If the power change rate exceeds the confidence interval, the corresponding data point is marked as an isolated value and set to null. For missing values, distance-based interpolation is performed. The interpolation weights are based on the distance between the element to be interpolated and each cluster, as well as the inverse distances of the nearest elements to the element to be interpolated. The missing values ​​are filled by weighted averaging.

3. The optimized scheduling method considering traction network resources according to claim 1, characterized in that, Based on electricity market operation data, input features for wind power prediction are selected, redundancy among input features is calculated and redundant features are removed, and the prediction model is optimized using a particle swarm optimization algorithm to obtain the predicted wind power value. Further steps include: Input features for wind power prediction include wind speed, humidity, temperature, and wind force, and historical data are collected as feature samples. Calculate the redundancy between input features, quantify the correlation between features using mutual information, and set a threshold to remove redundant features. The particle swarm optimization algorithm is used to optimize the prediction model. The model parameters, including the decision ratio coefficient, optimal location, input variable weights and biases, are adjusted by iteratively searching for the optimal parameter combination. Based on the optimized prediction model, the predicted wind power value is output.

4. The optimized scheduling method considering traction network resources according to claim 1, characterized in that, Based on electricity market operation data, a Transformer model is used to extract features from photovoltaic power data. Attention weights are used to weight and sum the multi-head vectors. Combined with an inductive conformal prediction method, the predicted photovoltaic power value is calculated. Further steps include: The Transformer model is used to extract features from photovoltaic power data, mapping the input sequence to different dimensions of a multi-head vector to capture temporal dependencies; Attention weights are used to perform a weighted summation of the multi-head vectors, and weights are assigned according to the similarity between vectors. The features are then fused to obtain the final representation of the vector features. The photovoltaic power prediction value is calculated by combining the inductive conformal prediction method, the prediction interval is constructed by using the conformal prediction quantile, and the output of the Transformer model is integrated. The prediction results are optimized based on the significance level and the number of calibration set samples. The prediction uncertainty is controlled by adjusting the confidence level, and the photovoltaic power prediction value is obtained.

5. The optimized scheduling method considering traction network resources according to claim 1, characterized in that, Based on electricity market operation data, a spatial autocorrelation method is used to screen key site data. This key site data is then processed using a self-attention mechanism and spatiotemporal feature extraction process. Finally, load power forecasts are calculated by combining time-series state information. Further steps include: The spatial autocorrelation method was used to screen data from key stations, calculate the correlation between observations under spatial and temporal lags, and select representative stations based on the correlation coefficient. The system processes key site data based on self-attention mechanism and spatiotemporal feature extraction process, extracts spatial distribution and time series features, and enhances feature representation capability. The load power prediction value is calculated by combining the temporal state information, and the hidden state and cell state of the long short-term memory network are used to capture dynamic changes and fuse the state of the previous time step. The predicted load power is calculated using the weights and bias parameters of the output layer.

6. The optimized scheduling method considering traction network resources according to claim 1, characterized in that, Based on the predicted wind power, photovoltaic power, and load power, an adjustable resource planning model is constructed, which includes an objective function and constraints. The objective function includes multiple revenue and cost terms, and the constraints include power constraints, energy consumption constraints, and time constraints. Further, it includes: Construct an objective function that includes revenue from external electricity sales, revenue from internal electricity sales, costs during the planning period, and revenue and costs of traction network participation in dispatching. Maximize the total revenue by weighted summation. Set constraints, including instantaneous power upper and lower limit constraints, average power and total time correlation constraints, minimum traction energy consumption constraints, power constraints obtained by the traction network from the power grid, rated power constraints of the traction converter, upper and lower limit constraints of train running time, and maximum allowable average power constraints. Establish an interactive regulation mechanism between the flexible resources of the traction network and the consumption of wind and solar energy, and achieve multi-energy synergy through coordinated power allocation; An adjustable resource planning model is built based on the objective function and constraints, and the prediction results are integrated to generate a scheduling scheme.

7. The optimized scheduling method considering traction network resources according to claim 1, characterized in that, Using multi-energy power generation and consumption time-series information as a training set, the adjustable resource planning model is iteratively optimized to obtain the final scheduling model, which further includes: Using multi-energy generation and consumption time-series information as the training set, the adjustable resource planning model is input, including historical power generation and consumption data sequences. Set the number of iterations as a hyperparameter for training to control the termination condition of the optimization process; Heuristic optimization algorithms are used to iteratively optimize the model parameters to adapt to the mixed discrete and continuous variables contained in the model; The final scheduling model is obtained through convergence judgment. The final scheduling model is output when the parameter change is less than the threshold or the maximum number of iterations is reached.

8. An optimized scheduling system considering traction network resources, characterized in that, include: The data preprocessing module is configured to perform cluster analysis on the data of each node in the power system to obtain multiple clusters, calculate the power change rate of each cluster, evaluate whether the data point is an isolated value based on the power change rate and set the isolated value as a null value, perform distance-based interpolation calculation on the null value, and obtain the expanded power system data. The wind power prediction module is configured to select input features for wind power prediction based on electricity market operation data, calculate the redundancy between input features and remove redundant features, and use particle swarm optimization algorithm to optimize the prediction model to obtain the wind power prediction value. The photovoltaic power prediction module is configured to use the Transformer model to extract features from the photovoltaic power data based on electricity market operation data, use attention weights to weighted summation of multi-head vectors, and combine inductive conformal prediction method to calculate the photovoltaic power prediction value. The load power prediction module is configured to use spatial autocorrelation method to screen key site data based on electricity market operation data, process key site data based on self-attention mechanism and spatiotemporal feature extraction process, and calculate load power prediction value by combining time series status information. The planning model construction module is configured to construct an adjustable resource planning model containing an objective function and constraints based on the wind power forecast, photovoltaic power forecast, and load power forecast. The objective function includes multiple revenue and cost items, and the constraints include power constraints, energy consumption constraints, and time constraints. The model optimization module is configured to use multi-energy generation and consumption time-series information as a training set to iteratively optimize the adjustable resource planning model to obtain the final scheduling model.

9. An electronic device, characterized in that, include: The system includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus; the memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of the method according to any one of claims 1 to 7.

Images