New energy consumption assessment method and system based on multi-dimensional scene and source-load interaction characteristics

Abstract

The invention discloses a new energy consumption assessment method and system based on a multi-dimensional scene and source load interaction characteristic. The method comprises the following steps: collecting historical source load data, carrying out clustering analysis, and generating a source load scene; constructing a flexible resource scheduling model; basic data related to the new energy consumption index are collected and preprocessed; the method comprises the following steps: constructing an attention-enhanced time convolutional network model, training and optimizing, enhancing the attention on important time steps through an attention mechanism, generating a plurality of probability scenes containing N-1 fault conditions, simulating the running state of the system under the fault conditions, and obtaining the fault state of the system. Calculating and evaluating the new energy consumption capability and the minimum total load reduction quantity under different fault conditions; and under different probability scenes and fault conditions, calculating the new energy output of the system and the maximum new energy consumption capability of the system, and forming a comprehensive evaluation report. According to the method, the calculation efficiency is remarkably improved while the evaluation precision is improved, and a more accurate and practical solution is provided for new energy consumption evaluation.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system absorption capacity assessment, and specifically relates to a new energy absorption assessment method and system based on multi-dimensional scenarios and source-load interaction characteristics. Background Art

[0002] As the global demand for green energy grows, the installed capacity of intermittent renewable energy such as wind and solar energy continues to rise. However, the access of these renewable energy sources has brought challenges to the stable operation of the power system, especially in the case of high penetration. Therefore, accurately evaluating the renewable energy absorption capacity is crucial to ensure the safe and stable operation of the power system.

[0003] The existing technologies for evaluating the absorption capacity of renewable energy mainly include analytical method, time-series production simulation method and random production simulation method. The analytical method is a method based on mathematical models. It solves the optimal absorption plan of renewable energy by establishing power supply, load and network constraints. This method has a fast calculation speed, but the results are relatively conservative and fail to fully reflect the randomness and time-series characteristics of renewable energy output. The time-series production simulation method considers the time-series variation characteristics of renewable energy output and load by simulating the operating state of the power system under different time series. Although this method improves the evaluation accuracy, it still has limitations for long-term evaluation due to the large amount of calculation. The random production simulation method describes the uncertainty of renewable energy output and load in a probabilistic way and then converts it into deterministic operation. Although it improves the accuracy of the evaluation, the calculation efficiency is low when dealing with complex scenarios.

[0004] Therefore, there is an urgent need to develop a new energy absorption capacity assessment method and system to provide a more accurate and practical solution for the new energy absorption capacity assessment. Summary of the invention

[0005] Purpose of the invention: In order to solve the above technical problems, the present invention provides a new energy consumption evaluation method and system based on multi-dimensional scenarios and source-load interaction characteristics.

[0006] Technical solution: The present invention provides a new energy consumption evaluation method based on multi-dimensional scenarios and source-load interaction characteristics, comprising the following steps:

[0007] (1) Collect historical source-load data for cluster analysis and generate source-load scenarios;

[0008] (2) Construct a flexible resource scheduling model;

[0009] (3) Collect and pre-process basic data related to the new energy consumption index;

[0010] (4) Construct an attention-enhanced temporal convolutional network model, train and optimize it, enhance attention to important time steps through the attention mechanism, generate multiple probabilistic scenarios containing N-1 fault conditions, simulate the operating status of the system under fault conditions, calculate and evaluate the new energy absorption capacity under different fault conditions and minimize the total load reduction;

[0011] (5) Under different probability scenarios and fault conditions, calculate the system's new energy output and the system's maximum new energy absorption capacity to form a comprehensive evaluation report.

[0012] Furthermore, the cluster analysis in step (1) is to use K-means clustering algorithm to perform data clustering to generate K representative scenarios.

[0013] Cluster=KMeans(Data,K)

[0014] Where: Data is the matrix containing source and load data, and K is the number of clusters.

[0015] The generating source-load scenario is to use GAN to generate a set of source-load scenarios covering multiple possibilities.

[0016] GAN=G(Noise)→Generated Data

[0017] Where: G is the generator, input is noise, and output is the generated source load data.

[0018] By clustering historical data, we generate a variety of representative source-load interaction scenarios, including the characteristics of power supply and demand in different seasons and time periods, to ensure the comprehensiveness of the assessment. We use GAN to generate a set of source-load scenarios covering a variety of possibilities, accurately depict the uncertainty on both sides of the source and load, and improve the comprehensiveness of the assessment.

[0019] Furthermore, the flexible resource scheduling model described in step (2) is a system scheduling model that includes wind power, photovoltaic power generation, and traditional power sources (such as hydropower, thermal power, etc.), and takes into account the start and stop, output adjustment, and load response characteristics of the power source. The power source with adjustment capability is optimized and scheduled to adapt to the randomness of new energy. The relevant formula is as follows:

[0020] The objective function is as follows:

[0021] minimizeC(x)

[0022] C(x) is the total cost function

[0023] Furthermore, the constraints of the flexible resource scheduling model described in step (2) include power balance constraints, upper and lower bound constraints on power generation and load reduction, conventional power generation constraints, and non-negative constraints on load reduction.

[0024] Furthermore, the basic data in step (3) includes historical wind power generation, photovoltaic power generation, load demand and related meteorological data. The preprocessing includes data cleaning and data standardization. Data cleaning ensures data quality by processing missing values ​​and outliers.

[0025] Data standardization improves the effect of model training by standardizing the data to make its mean 0 and variance 1.

[0026] Furthermore, the attention-enhanced temporal convolutional network model structure described in step (4) includes a convolutional layer, an attention mechanism layer, and a fully connected layer. The model is trained using historical data and the model parameters are optimized to minimize the prediction error. The local features in the time series are extracted through the convolutional layer, and the attention mechanism is used to enhance the focus on important time steps.

[0027] The fault conditions described in step (4) include generator fault, transformer fault, and transmission line fault. By defining the conditions under which each component in the system (such as generator, transformer, transmission line, etc.) may fail under a fault condition, multiple probability scenarios containing N-1 fault conditions are generated to simulate the operating state of the system under these fault conditions.

[0028] The constraints for calculating the minimized total load reduction in step (4) include power balance constraints, upper and lower bound constraints on power generation and load reduction, conventional power generation constraints, and non-negative load reduction constraints.

[0029] The present invention also provides a new energy consumption evaluation system based on multi-dimensional scenarios and source-load interaction characteristics, including a historical source-load data acquisition module, a data clustering analysis module, a resource scheduling model, a temporal convolutional network model, and a new energy consumption capacity calculation module. The temporal convolutional network model includes a convolutional layer, an attention mechanism layer, and a fully connected layer.

[0030] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following significant advantages: by using a generative adversarial network (GAN) to generate a set of source-load scenarios covering a variety of possibilities, accurately characterize the uncertainty on both sides of the source and load, and improve the comprehensiveness of the evaluation. Secondly, the correlation between the source and load is considered through the K-means clustering technology to generate representative scenarios to ensure the authenticity and effectiveness of the evaluation results. In addition, the attention-enhanced Time Convolutional Network (ATCN) based on the introduction of the attention-enhanced Time Convolutional Network (ATCN) makes full use of historical time series information to improve the accuracy and speed of the model's prediction of the new energy absorption capacity. Finally, combined with the N-1 safety check, the expected value of the minimum load reduction under different new energy penetration rates is optimized to ensure that the system can maintain stable operation even in the event of a fault. While improving the evaluation accuracy, the calculation efficiency is significantly improved, and all relevant factors can be considered more comprehensively, thereby providing a more accurate and practical solution for the evaluation of new energy absorption capacity. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a schematic diagram of the workflow of the new energy consumption evaluation method based on multi-dimensional scenarios and source-load interaction characteristics of the present invention. DETAILED DESCRIPTION

[0032] The technical solution of the present invention is further described in detail below through specific embodiments. However, it is necessary to point out that the following embodiments are only used to describe the content of the invention and do not constitute a limitation on the protection scope of the present invention.

[0033] A new energy consumption evaluation method based on multi-dimensional scenarios and source-load interaction characteristics includes the following steps:

[0034] Step S1: Scenario Generation and Analysis

[0035] By clustering historical data, we generate a variety of representative source-load interaction scenarios, including the characteristics of power supply and demand in different seasons and time periods, to ensure the comprehensiveness of the assessment. We use GAN to generate a set of source-load scenarios covering a variety of possibilities, accurately depict the uncertainty on both sides of the source and load, and improve the comprehensiveness of the assessment.

[0036] After collecting historical source-load data, S1-1 uses the K-means clustering algorithm to cluster the data and generate K representative scenarios.

[0037] Cluster=KMeans(Data,K)

[0038] Where: Data is the matrix containing source and load data, and K is the number of clusters.

[0039] S1-2 uses GAN to generate a set of source-load scenarios covering a variety of possibilities.

[0040] GAN=G(Noise)→Generated Data

[0041] Where: G is the generator, input is noise, and output is the generated source load data.

[0042] Step S2: Building a flexible resource scheduling model

[0043] Furthermore, a system dispatch model including wind power, photovoltaic power generation and traditional power sources (such as hydropower, thermal power, etc.) is constructed. The model takes into account the start and stop of power sources, output adjustment and load response characteristics. Power sources with regulation capabilities are optimized for dispatch to adapt to the randomness of new energy sources. The relevant formula is as follows:

[0044] The objective function is as follows:

[0045] minimize C(x)

[0046] C(x) is the total cost function

[0047] Constraints:

[0048] 1. The power balance constraints are as follows:

[0049]

[0050] Where: is the power generation of the i-th wind turbine at time t, is the power generation of the jth photovoltaic cell at time t, is the power generation of the kth energy storage device at time t.

[0051] 2. The upper and lower bounds of power generation and load reduction are as follows:

[0052]

[0053] 3. The power generation capacity of conventional power sources is constrained as follows:

[0054]

[0055] 4. The non-negative constraints on load reduction are as follows:

[0056]

[0057] Step S3: Application of Stochastic Production Simulation Method and Attention-Enhanced Temporal Convolutional Network (ATCN)

[0058] (1) Data collection and preprocessing

[0059] Data collection: Obtain historical wind power, photovoltaic power generation, load demand and related meteorological data.

[0060] Data cleaning: Process missing values ​​and outliers to ensure data quality.

[0061] Data standardization: Standardize the data so that its mean is 0 and its variance is 1 to improve the effect of model training.

[0062] (2) Training ATCN model

[0063] Model construction: Construct an attention-enhanced temporal convolutional network (ATCN), whose structure includes convolutional layers, attention mechanism layers, and fully connected layers.

[0064] Model training: Use historical data to train the model and optimize the model parameters θ to minimize the prediction error.

[0065] Feature extraction: extract local features in the time series through the convolutional layer, and then enhance the focus on important time steps through the attention mechanism.

[0066] (3) Generate a probabilistic scenario including N-1 safety checks

[0067] ATCN-based prediction: Use the trained ATCN model to predict future renewable energy output:

[0068]

[0069] Probability distribution description: Probability distribution is used to describe the uncertainty of renewable energy output and system load, and multiple probability scenarios are generated, which reflect different operating conditions and fault scenarios.

[0070] N-1 fault conditions: Define the conditions under which each component in the system (such as generators, transformers, transmission lines, etc.) may fail in a fault state. Generate multiple probabilistic scenarios containing N-1 fault conditions to simulate the operating state of the system under these fault conditions.

[0071] (4) Calculation of new energy consumption capacity

[0072] In each generated probability scenario, the system's new energy absorption capacity is calculated and the absorption capacity under different fault conditions is evaluated.

[0073] (5) Solving the optimization problem

[0074] The goal is to minimize the total load reduction. The objective function is as follows:

[0075]

[0076] The constraints are as follows:

[0077] (1) Power balance constraints

[0078]

[0079] The purpose is to Under fault conditions f, the total generation at time t plus the load reduction should be equal to the load demand.

[0080] (2) Upper and lower bounds of power generation and load reduction

[0081]

[0082] The purpose is to The power generation at time t and fault state f should be between 0 and the maximum value.

[0083]

[0084] The purpose is to use photovoltaic cells in the scene The power generation at time t and fault state f should be between 0 and the maximum value.

[0085]

[0086] The purpose is to store energy in the scene The power generation at time t and fault state f should be between 0 and the maximum value.

[0087] (3) Constraints on power generation from conventional power sources

[0088]

[0089] Its purpose is to use conventional power supply in the scene The power generation at time t and fault state f should be between 0 and the maximum value.

[0090] (5) Load reduction non-negative constraint

[0091]

[0092] The purpose is to reduce the load in the scenario It should be a non-negative value at time t and fault state t.

[0093] Step S4: Analyze the evaluation results

[0094] (1) Comprehensive analysis of the results of the previous steps

[0095] All probability scenarios, fault simulation results and new energy output forecast data generated in step S3 are integrated.

[0096] Analyze the impact of different new energy penetration rates on system safety and reliability. The formula is as follows:

[0097]

[0098] (2) Calculate the system’s new energy consumption capacity

[0099] Compute the system's renewable energy output under different scenarios and fault conditions:

[0100]

[0101] Calculate the system's maximum renewable energy consumption capacity:

[0102]

[0103] (3) Formation of a comprehensive assessment report

[0104] Combined with the fault simulation results in step S3, the safety and reliability of the system under different fault conditions are analyzed:

[0105]

[0106] That is, the system's new energy acceptance limit can be evaluated under different operating conditions and fault scenarios. Based on the evaluation results, suggestions are put forward to optimize the new energy absorption capacity, including grid dispatching strategies and new energy development planning suggestions.

[0107] In summary, the method and system for evaluating the consumption of new energy based on multi-dimensional scenarios and source-load interaction characteristics proposed in the present invention use K-means clustering and GAN to generate source-load scenarios covering a variety of possibilities, accurately depicting the uncertainty on both sides of the source and load. And introduce a time convolutional network based on attention enhancement to make full use of historical time series information to improve the accuracy and speed of new energy output prediction. By combining N-1 safety verification and probabilistic scenario generation, the expected value of the minimum load reduction under different new energy penetration rates is optimized, and the system's new energy consumption capacity under different operating scenarios is comprehensively evaluated, providing a scientific basis for new energy development planning and power grid dispatching.

Claims

1. A new energy consumption evaluation method based on multi-dimensional scenarios and source-load interaction characteristics, characterized in that: The following steps are involved: (1) Collect historical source-load data for cluster analysis and generate source-load scenarios; (2) Construct a flexible resource scheduling model; (3) Collect and pre-process basic data related to the new energy consumption index; (4) Construct an attention-enhanced temporal convolutional network model, train and optimize it, enhance attention to important time steps through the attention mechanism, generate multiple probabilistic scenarios containing N-1 fault conditions, simulate the operating status of the system under fault conditions, calculate and evaluate the new energy absorption capacity under different fault conditions and minimize the total load reduction; (5) Under different probability scenarios and fault conditions, calculate the system's new energy output and the system's maximum new energy absorption capacity to form a comprehensive evaluation report.

2. The new energy consumption assessment method based on multi-dimensional scenarios and source-load interaction characteristics according to claim 1 is characterized in that: The cluster analysis in step (1) is to use the K-means clustering algorithm to cluster data and generate K representative scenarios. Cluster=KMeans(Data,K) Where: Data is the matrix containing source and load data, and K is the number of clusters.

3. The new energy consumption assessment method based on multi-dimensional scenarios and source-load interaction characteristics according to claim 1 is characterized in that: The flexible resource scheduling model described in step (2) is a system scheduling model that includes wind power, photovoltaic power generation and traditional power sources, and takes into account the start and stop, output adjustment and load response characteristics of the power source.

4. The new energy consumption assessment method based on multi-dimensional scenarios and source-load interaction characteristics according to claim 1 is characterized in that: The constraints of the flexible resource scheduling model described in step (2) include power balance constraints, upper and lower bound constraints on power generation and load reduction, conventional power generation constraints, and non-negative constraints on load reduction.

5. The new energy consumption assessment method based on multi-dimensional scenarios and source-load interaction characteristics according to claim 1 is characterized in that: The basic data described in step (3) include historical wind power generation, photovoltaic power generation, load demand and related meteorological data.

6. The new energy consumption assessment method based on multi-dimensional scenarios and source-load interaction characteristics according to claim 1 is characterized in that: The attention-enhanced temporal convolutional network model structure described in step (4) includes a convolutional layer, an attention mechanism layer, and a fully connected layer.

7. The new energy consumption assessment method based on multi-dimensional scenarios and source-load interaction characteristics according to claim 1 is characterized in that: The fault conditions described in step (4) include generator fault, transformer fault, and transmission line fault.

8. The new energy consumption assessment method based on multi-dimensional scenarios and source-load interaction characteristics according to claim 1 is characterized in that: The constraints for calculating the minimized total load reduction in step (4) include power balance constraints, upper and lower bound constraints on power generation and load reduction, conventional power generation constraints, and non-negative load reduction constraints.

9. A new energy consumption assessment system based on multi-dimensional scenarios and source-load interaction characteristics, characterized in that: It includes historical source and load data acquisition module, data clustering analysis module, resource scheduling model, time convolution network model, and new energy absorption capacity calculation module.

10. The new energy consumption assessment system based on multi-dimensional scenarios and source-load interaction characteristics according to claim 9 is characterized in that: The temporal convolutional network model includes a convolutional layer, an attention mechanism layer and a fully connected layer.

Images