Extraction method of typical mode of new energy planning based on explainable artificial intelligence

Abstract

The present application relates to the technical field of new energy planning, and more particularly to a new energy planning typical mode extraction method and device based on explainable artificial intelligence, electronic equipment and computer readable storage medium, wherein the method comprises: setting source and load boundary conditions, generating candidate new energy configuration scenarios and running simulation to generate high-dimensional time series data sets; using an unsupervised clustering algorithm to extract patterns from the high-dimensional time series data sets to obtain multiple operating modes; identifying key factors based on Monte Carlo Shapley values and modeling explainability to obtain the relationship between the key factors and the multiple operating modes; determining whether the current scheme meets the planning target based on the relationship between the key factors and the multiple operating modes and the engineering evaluation index, and if not, reiterating the processing. The explainable artificial intelligence method is introduced into the new energy planning decision-making assistance, which strengthens the operation mechanism modeling capability, improves the planning scheme credibility, and supports scientific decision-making under the condition of high penetration of new energy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of new energy planning, and in particular to a new energy planning typical mode extraction method and device based on explainable artificial intelligence, an electronic device, and a computer readable storage medium. BACKGROUND

[0002] With the gradual increase of the penetration rate of new energy in the power system, new energy planning is facing the challenges of complex operation mode and increasing uncertainty. When extracting typical operation modes using traditional methods, it is often difficult to fully reveal the underlying operation mechanism and logical relationship, affecting the scientificity and adaptability of long-term new energy layout.

[0003] Currently, operation mode recognition mainly relies on unsupervised clustering, time series clustering, deep learning, and explainable enhancement methods. In addition, in other engineering fields, Shapley values have been used to improve the explainability of models, but their application in the power system is still in the exploratory stage. Although the above methods have made certain progress in pattern recognition, there are still limitations in extracting typical modes and analyzing mechanisms in new energy planning. First, the existing technology lacks a planning-oriented explainability framework, and the logical relationship between typical modes and new energy configuration structure has not been fully revealed, limiting its application value in long-term layout optimization. Second, the dominant factors of operation mode and their change rules have not been quantitatively identified, making it difficult to develop control strategies. Third, existing methods rely on artificial setting of experience factors or specific data distribution, which limits the universality and adaptability of the model. Finally, in view of the influence of new energy penetration rate changes on system complexity and operation uncertainty, existing research has not yet formed a systematic evaluation method, making it difficult to accurately quantify its effect on the structure of typical operation modes. SUMMARY

[0004] The present application aims to at least partially solve one of the technical problems in the related art.

[0005] To this end, the first object of the present application is to propose a new energy planning typical mode extraction method based on explainable artificial intelligence to solve the problem that existing technical means still have limitations in extracting typical modes and analyzing mechanisms in new energy planning.

[0006] The second object of the present application is to propose a device.

[0007] The third object of the present application is to propose an electronic device.

[0008] The fourth object of the present application is to propose a computer readable storage medium.

[0009] To achieve the above object, the first aspect of the present application proposes a new energy planning typical mode extraction method based on explainable artificial intelligence, comprising:

[0010] Setting source and load boundary conditions to generate candidate new energy configuration scenarios;

[0011] Running simulation based on the candidate new energy configuration scenarios to generate high-dimensional time series data sets;

[0012] Using unsupervised clustering algorithm to extract patterns from the high-dimensional time series data sets to obtain multiple operating modes;

[0013] Based on the key factor identification and explainable modeling of Monte Carlo, the relationship between the key factor and the multiple operating modes is obtained;

[0014] Based on the relationship between the key factor and the multiple operating modes and the engineering evaluation index, it is judged whether the current scheme meets the planning target. If not, re-iterate the process.

[0015] Preferably, the setting source and load boundary conditions to generate candidate new energy configuration scenarios comprises:

[0016] Determine the representative load curve, meteorological data year, resource potential, technical constraint, power grid transmission capacity and cost boundary, and generate candidate new energy configuration scenarios.

[0017] Preferably, the running simulation based on the candidate new energy configuration scenarios to generate high-dimensional time series data sets comprises:

[0018] Build a production simulation tool, and run simulation based on the candidate new energy configuration scenarios to obtain the output of all units including new energy units, line flow and load data, and generate high-dimensional time series data sets.

[0019] Preferably, the key factor identification and explainable modeling based on Monte Carlo Shapley value to obtain the relationship between the key factor and the multiple operating modes comprises:

[0020] Data preprocessing and clustering analysis are performed on the high-dimensional time series data to obtain preprocessed data;

[0021] Based on the preprocessed data, the Shapley value of the preprocessed data is obtained by using the Shapley value method of Monte Carlo sampling, and the Shapley value meeting the preset value is extracted as the key factor;

[0022] Based on the key factor, an explainable classification model is constructed, and a predicted clustering label is obtained;

[0023] The predicted clustering labels are analyzed by using an explainable classification model to obtain a relationship between key factors and the multiple operation modes.

[0024] Preferably, the Shapley value calculation formula is:

[0025]

[0026] wherein, is a complete feature set, is a feature is an arbitrary feature subset other than the feature represents a feature subset a profit function of a lower model, is a feature Shapley value of the feature.

[0027] Preferably, the analyzing the predicted clustering labels by using an explainable classification model to obtain a relationship between key factors and the multiple operation modes comprises: taking the key factors as input variables of the explainable classification model, obtaining classification results of precision, recall, accuracy and F1 value under different new energy penetration rate scenarios, and obtaining the relationship between the key factors and the multiple operation modes based on the classification results.

[0028] Preferably, the determining whether a current scheme meets a planning target based on the relationship between the key factors and the multiple operation modes and engineering evaluation indexes, and re-iterating if the current scheme does not meet the planning target comprises:

[0029] calculating model prediction accuracy, and evaluating operation cost, curtailment rate, regulation capacity and carbon emission level indexes;

[0030] if the expectation is not met, adjusting parameters in source-load boundary conditions and re-iterating.

[0031] To achieve the above purpose, a second aspect embodiment of the present application proposes a new energy planning typical mode extraction device based on explainable artificial intelligence, comprising:

[0032] a scenario generation module configured to set source-load boundary conditions and generate candidate new energy configuration scenarios;

[0033] a data acquisition module configured to run simulation based on the candidate new energy configuration scenarios and generate a high-dimensional time series data set;

[0034] a data extraction module configured to use an unsupervised clustering algorithm to extract patterns from the high-dimensional time series data set and obtain multiple operation modes;

[0035] an identification module configured to identify key factors based on Monte Carlo Shapley values and build an explainable model to obtain a relationship between the key factors and the multiple operation modes.

[0036] a judging module, judging whether the current scheme meets the planning target based on the relationship between the key factor and the plurality of operation modes and the engineering evaluation index, and if not up to the mark, re-iterating the processing.

[0037] To achieve the above purpose, a third aspect of the present application provides an electronic device, comprising: a processor, and a memory in communication connection with the processor;

[0038] The memory stores computer execution instructions.

[0039] The processor executes the computer execution instructions stored in the memory to realize the method of any one of the above.

[0040] To achieve the above purpose, a fourth aspect of the present application provides a computer readable storage medium, comprising computer execution instructions stored in the computer readable storage medium, the computer execution instructions are executed by the processor to realize the method of any one of the above.

[0041] The method provided by the present application extracts a typical mode of new energy planning based on explainable artificial intelligence. By constructing a clustering algorithm, the typical state of high-dimensional time series operation data is identified, and the representative operation mode is accurately extracted, thereby providing real and representative operation scene support for medium and long term new energy configuration scheme, and improving the adaptability and robustness of new energy layout scheme. The feature importance identification mechanism based on Monte Carlo-Shapley value is introduced to quantify the marginal contribution of each input factor to the formation of the operation mode, to break through the path from "black box clustering result" to "explainable factor driven", and to improve the traceability and credibility of the planning analysis. Combined with the explainable models such as logistic regression and decision tree, the mapping relationship between the key factor and the operation mode is established, the operation mechanism discrimination rule is extracted, the action path of new energy layout, load level, new energy resource endowment and other factors on system operation state evolution is revealed, and the planning personnel are assisted to carry out causal attribution analysis and multi-scheme comparison. It has the comprehensive advantages of strengthening the operation mechanism modeling ability, improving the credibility and transparency of the planning scheme, and supporting the scientific decision-making under the condition of high penetration of new energy, and can provide key support tools for operation analysis, scheme evaluation and strategy design in the planning stage of new type power system.

[0042] Additional aspects and advantages of the application will be set forth in part in the description which follows, and in part will become apparent to those skilled in the art upon examination of the following and upon having the benefit of the description presented in the following description and appended claims. BRIEF DESCRIPTION OF DRAWINGS

[0043] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description, taken in conjunction with the accompanying drawings, in which:

[0044] Figure 1 A flowchart of a first specific embodiment of a new energy planning typical mode extraction method based on explainable artificial intelligence provided by the present application;

[0045] Figure 2 A new energy planning framework based on explainable artificial intelligence;

[0046] Figure 3 A key factor identification and explainability analysis framework based on Monte Carlo-Shapley;

[0047] Figure 4 A structure block diagram of a new energy planning typical mode extraction device based on explainable artificial intelligence provided by the embodiment of the present application. DETAILED DESCRIPTION

[0048] The core of the present application is to provide a new energy planning typical mode extraction method, device, electronic equipment and computer readable storage medium based on explainable artificial intelligence, which introduces an explainable artificial intelligence method into new energy planning auxiliary decision-making, strengthens the operation mechanism modeling capability, improves the planning scheme credibility and transparency, and supports scientific decision-making under the condition of high penetration of new energy, etc. Comprehensive advantages.

[0049] In order to enable personnel in the art to better understand the present application scheme, the present application will be further described in detail below in combination with the drawings and specific embodiments. Obviously, the described embodiments are only part of the embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of the present application.

[0050] Please refer to Figure 1 , Figure 1 A flowchart of a first specific embodiment of a planning typical mode extraction method based on explainable artificial intelligence provided by the present application; the specific operation steps are as follows:

[0051] Step S101: Set source load boundary conditions, and generate candidate new energy configuration scenarios;

[0052] Step S102: Run simulation based on the candidate new energy configuration scenarios, and generate a high-dimensional time series data set;

[0053] Step S103: Use an unsupervised clustering algorithm to extract patterns from the high-dimensional time series data set, and obtain multiple operating modes;

[0054] Step S104: Key factor identification and explainability modeling based on Monte Carlo Shapley value, obtain the relationship between the key factor and the multiple operating modes;

[0055] Step S105: Determine whether the current scheme meets the planning target based on the relationship between the key factors and the multiple operation modes and the engineering evaluation index. If not, re-iterate the process.

[0056] Based on the above embodiment, the step S101 is described in detail in this embodiment:

[0057] In one embodiment, representative load curves, meteorological data years, resource potential, technical constraints, grid transmission capacity and cost boundaries are determined, and candidate new energy configuration scenarios are generated.

[0058] Specifically, according to the actual demand of regional new energy planning, the representative load curve, meteorological data year (such as typical meteorological year or extreme year), resource potential, technical constraint, grid transmission capacity, cost boundary and other key parameters are set to construct a series of candidate new energy configuration scenarios. This boundary condition set provides input basis for subsequent operation simulation and scheme optimization, and can also be iteratively adjusted according to the analysis results.

[0059] Based on the above embodiment, the step S102 is described in detail in this embodiment:

[0060] In one embodiment, a production simulation tool is constructed, and based on the candidate new energy configuration scenarios, simulation is run to obtain all unit outputs including new energy units, line flow and load data, and generate a high-dimensional time series data set.

[0061] Specifically, based on the given source and load boundary conditions, an 8760-hour annual scale operation simulation model is constructed, and system operation simulation under multiple scenarios is carried out, and a high-dimensional time series data set of operation indexes including all unit outputs including new energy units, line flow and load data is output. This step can use a refined production simulation tool to ensure the true restoration of the operation state under high new energy proportion conditions.

[0062] Based on the above embodiment, the step S103 is described in detail in this embodiment:

[0063] In one embodiment, the high-dimensional time series data is preprocessed and cluster analyzed to obtain preprocessed data; based on the preprocessed data, the Shapley value of the preprocessed data is obtained by using the Shapley value method of Monte Carlo sampling, and the Shapley value meeting the preset value is extracted as the key factor; based on the key factor, an explainable classification model is constructed, and a predicted clustering label is obtained; the explainable classification model is used to analyze the predicted clustering label to obtain the relationship between the key factor and the multiple operation modes.

[0064] Specifically, the clustering algorithm in unsupervised learning (such as K-means, DBSCAN, spectral clustering, etc.) is used to perform clustering analysis on the operation data samples, to extract several typical operation modes from the large-scale operation state, and to perform visual presentation, to reveal the typical response characteristics, regulation state and fluctuation level of the system under different configurations. This process realizes the structured abstraction from "data set" to "operation mode", and the Shapley value calculation formula is:

[0065]

[0066] wherein, is the entire feature set, is a feature outside the arbitrary feature subset, represents the feature subset the benefit function of the lower model, is the Shapley value of the feature .

[0067] Based on the above embodiment, the step S104 is described in detail in this embodiment:

[0068] In one embodiment, the key factors are taken as input variables of the explanatory classification model, and the classification results of precision, recall, accuracy and F1 value under different new energy penetration rate scenarios are obtained. Based on the classification results, the relationship between the key factors and the plurality of operation modes is obtained.

[0069] Specifically, based on the clustering analysis, a Shapley value calculation method based on Monte Carlo sampling is constructed to evaluate the marginal contribution of each input factor (such as new energy output, load structure, key area line flow, etc.) to the clustering result, and to identify the Top key factor that plays a decisive role in the operation mode division. At the same time, an interpretable artificial intelligence model (such as decision tree, logistic regression, etc.) is introduced to establish the causal mapping between the factors and the operation modes, to extract the decision path and the discrimination rule, and to enhance the understandability of the clustering result and the system structure recognition ability.

[0070] Based on the above embodiment, the step S105 is described in detail in this embodiment:

[0071] In one embodiment, the model prediction accuracy is calculated to evaluate the operation cost, the curtailment rate, the regulation capacity and the carbon emission level indicators; if the expected value is not reached, the parameters in the source and load boundary conditions are adjusted and re-iterated.

[0072] Specifically, based on the classification model verification results and engineering evaluation indicators, it is judged whether the current new energy configuration scheme achieves the expected target (such as operation cost, curtailment rate, regulation capacity, carbon emission level, etc.). If the model prediction accuracy is high and the system operation indicators meet the planning target, the current configuration is output as a feasible scheme; if it does not meet the expectation, the source and load boundaries are adjusted based on the key factor identification results, and the parameters are reset to return to the first step for the next round of planning simulation and analysis.

[0073] Through the organic integration of the above five steps, the method realizes the closed-loop new energy planning technical path from "configuration hypothesis → operation simulation → mode identification → mechanism analysis → scheme evaluation → parameter adjustment", which is especially suitable for regional new energy structure optimization and uncertainty control under the background of continuous improvement of new energy penetration rate, new energy-load-network structure complexity. Compared with traditional methods, the scheme has stronger data-driven ability, higher operation state description accuracy and stronger causal mechanism explainability, and has good engineering practicability and method universality.

[0074] Based on the above embodiment, the embodiment describes a typical method for extracting a new energy planning based on explainable artificial intelligence, as shown in Figure 2 , specifically as follows:

[0075] Step 1: Set the source and load boundary conditions;

[0076] Step 2: Fine operation simulation of high-proportion new energy power system;

[0077] Step 3: Data-driven typical operation mode extraction and visualization;

[0078] Step 4: Key factor identification and explainable modeling based on Monte Carlo-Shapley value;

[0079] Step 5: Scheme evaluation and source and load boundary adjustment;

[0080] Among them, the core goal of step 4 is to effectively extract the typical operation mode under the multi-source configuration scheme in the planning stage, and realize the modeling of the traceable and explainable causal mechanism, as shown in Figure 3 , specifically as follows:

[0081] 1: Data preprocessing and cluster analysis

[0082] First, the unit output, line flow, load and other information are extracted from the multi-scenario operation data generated by simulation, and systematic data preprocessing is performed, including feature normalization, etc. Then, unsupervised clustering algorithms suitable for high-dimensional operation data structure (such as K-means, DBSCAN, etc.) are used for typical mode identification, and the 8760-hour operation sample is divided into several representative operation categories as the basis for subsequent explanatory analysis.

[0083] 2: Key factor identification

[0084] Based on the obtained clustering labels, the Shapley value method based on Monte Carlo sampling is used to evaluate the marginal contribution of each input feature to the clustering structure. This method quantifies the contribution of each feature in the classification category by randomly arranging and iteratively sampling feature combinations. According to the Shapley value ranking, the top 10% key factors are extracted as the core variables for model construction and causal mechanism mining.

[0085] Specifically, Shapley value is used to identify the key feature factors that have the greatest impact on clustering results. The original calculation method of Shapley is as follows:

[0086]

[0087] where is the complete feature set, is the feature except for the feature subset , represents the feature subset , the profit function (evaluation index) of the lower model, is the Shapley value of feature .

[0088] Due to the complexity of Shapley value calculation, this method proposes a Shapley value approximation calculation method based on Monte Carlo simulation. By randomly arranging the feature order multiple times, the process of gradually adding features is simulated, and the amount of improvement in clustering results before and after the introduction of each feature is calculated as its marginal contribution. Finally, the average importance score of each feature is obtained. This method uses the Adjusted Rand Index (ARI) as the profit function to measure the consistency between the new and old clustering labels under the feature combination. As shown in Algorithm 1:

[0089] Algorithm 1: Key factor identification based on Monte Carlo-SHAP

[0090] Input: feature data X, clustering model , original clustering label

[0091] Output: Feature importance score

[0092] Initialization: For all features Let

[0093] ForN Monte Carlo sampling

[0094]

[0095] End For

[0096] Average over all

[0097] Return

[0098] 3: Explainable model construction and verification

[0099] To verify the representativeness of the extracted key factors, an explainable classification model (such as logistic regression, decision tree, etc.) based on these factors is constructed to predict the clustering label. This step evaluates the restoration ability of key factors to clustering results, and verifies the reliability and consistency of Shapley analysis results through classification accuracy, model stability, etc. The model complexity is controlled within a reasonable range to ensure that the analysis results have good interpretability and engineering applicability.

[0100] Specifically, considering the interpretability and stability of the model, the present application selects two typical explainable models for comparative analysis: Multinomial Logistic Regression and Decision Tree. The former is an extension of generalized linear models, suitable for multi-class discrimination tasks, can output the conditional probability of each class, and reflect the marginal effect of each feature on class discrimination through regression coefficients; the latter generates tree structure by recursively dividing feature space, the model structure is intuitive, and it is convenient to extract decision path and division rule, so as to reveal the association mechanism between feature combination and clustering label.

[0101] ​In the model training and validation process, the Top 10% Shapley value key factors identified above are used as input variables to evaluate their fitting ability to the clustering structure under different classification models. The performance indicators such as precision, recall, accuracy, and F1-score are used to measure the classification performance of the model under different new energy penetration scenarios. If the model can still achieve high prediction accuracy under the premise of relying only on a small number of key factors, it indicates that these factors have high information compression ability and can effectively describe the discrimination pattern of system operation characteristics between different clustering categories. Different evaluation indicators have the following meanings:

[0102] Precision represents the proportion of samples classified as class, which reflects the accuracy of the model:

[0103]

[0104] where represents the number of samples classified as the class and actually the class, represents the number of samples classified as the class but not actually the class.

[0105] Recall represents the proportion of samples actually in class that are correctly classified as class, reflecting the coverage ability of the model:

[0106]

[0107] where represents the number of samples actually in the class but classified as other categories.

[0108] F1-score is the harmonic mean of precision and recall, which is a comprehensive trade-off between the two:

[0109]

[0110] Accuracy represents the proportion of correctly classified samples to the total number of samples:

[0111]

[0112] where is the total number of classified samples.

[0113] ​If the model can still obtain high classification accuracy under the condition of relying only on a small number of key factors, it indicates that these factors have good information compression and discrimination ability, and can effectively reflect the evolution law and classification boundary of system operation mode. Further, under the premise that the model prediction accuracy meets the requirements, classification explanation rules can be extracted based on different model structures:

[0114] For the multi-classification logistic regression model, the positive and negative effects of each key factor on the discrimination probability of a certain class can be judged by analyzing the signs and sizes of the regression coefficients under different classes. For example, if the regression coefficient of a factor is positive, it means that the larger the value of the factor, the more likely the sample will be classified into that class, and vice versa.

[0115] For the decision tree model, the discrimination path and decision conditions of each class can be extracted based on the generated tree structure, and the threshold value of the key variable and its combination logic can be determined, so as to form a traceable and structured operation mode explanation chain.

[0116] This analysis process not only verifies the representativeness of key factors, but also realizes the transparent modeling process from "variable input" to "class output", providing methodological support for the attribution analysis and strategy deduction of system operation mode in the planning stage.

[0117] 4: Operation mechanism explanation and decision rule extraction

[0118] Based on the explanatory classification model, the discrimination rule path of the typical mode is extracted, and the discrimination mechanism of the key factor in different operation categories is analyzed by combining the numerical interval and combination relationship of the key factor. Through the decision logic inside the model, the core structural conditions driving the evolution of system operation mode under different new energy planning schemes are identified, forming an interpretable causal link from "input configuration" to "operation response", which provides mechanism support for new energy structure optimization, operation adaptability evaluation and control strategy formulation.

[0119] The embodiment provides a new energy planning typical mode extraction method based on explainable artificial intelligence, realizes typical state identification of high-dimensional time sequence operation data through construction of a clustering algorithm, accurately extracts a representative operation mode, provides a real and representative operation scene support for a medium and long term new energy configuration scheme, and improves adaptability and robustness of the new energy layout scheme. A feature importance identification mechanism based on Monte Carlo-Shapley value is introduced, the marginal contribution of each input factor to the formation of the operation mode is quantified, the path from the "black box clustering result" to the "explainable factor driven" is opened, and the traceability and credibility of the planning analysis are improved. In combination with an explainable model such as a logistic regression and a decision tree, a mapping relationship between key factors and operation modes is established, operation mechanism discrimination rules are extracted, the action path of factors such as new energy layout, load level and renewable power output on system operation state evolution is revealed, and planning personnel are helped to carry out causal attribution analysis and multi-scheme comparison. The method has comprehensive advantages such as strengthening operation mechanism modeling capability, improving planning scheme credibility and transparency, and supporting scientific decision-making under the condition of high penetration of new energy, and can provide a key support tool for operation analysis, scheme evaluation and strategy design in the planning stage of a new power system.

[0120] Please refer to Figure 4 , Figure 4 A structure block diagram of a new energy planning typical mode extraction device based on explainable artificial intelligence is provided for the embodiment of the application. The specific device can include:

[0121] A scene generation module 100 sets source load boundary conditions and generates a candidate new energy configuration scene;

[0122] A data acquisition module 200 generates a high-dimensional time sequence data set based on operation simulation of the candidate new energy configuration scene;

[0123] A data extraction module 300 uses an unsupervised clustering algorithm to extract a mode from the high-dimensional time sequence data set, and obtains a plurality of operation modes;

[0124] An identification module 400 identifies key factors based on Monte Carlo Shapley value and builds an explainable model, and obtains a relationship between the key factors and the plurality of operation modes;

[0125] A judgment module 500 judges whether a current scheme meets a planning target based on the relationship between the key factors and the plurality of operation modes and an engineering evaluation index. If the current scheme does not meet the planning target, the current scheme is iteratively processed.

[0126] The device for extracting a typical mode of new energy planning based on explainable artificial intelligence of the embodiment is used for realizing the method for extracting a typical mode of new energy planning based on explainable artificial intelligence, and the specific embodiments of the device for extracting a typical mode of new energy planning based on explainable artificial intelligence can be seen from the foregoing embodiment part of the method for extracting a typical mode of new energy planning based on explainable artificial intelligence, for example, the scene generation module 100, the data acquisition module 200, the data extraction module 300, the identification module 400 and the judgment module 500 are respectively used for realizing steps S101, S102, S103, S104 and S105 in the foregoing method for extracting a typical mode of new energy planning based on explainable artificial intelligence, so the specific embodiments can be referred to the description of the respective embodiment part, and details are not described herein again.

[0127] In order to realize the above-mentioned embodiments, the present application further provides an electronic device, comprising: a processor, and a memory connected with the processor in communication; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to realize the method provided by the foregoing embodiments.

[0128] In order to realize the above-mentioned embodiments, the present application further provides a computer readable storage medium, the computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed by the processor to realize the method provided by the foregoing embodiments.

[0129] In order to realize the above-mentioned embodiments, the present application further provides a computer program product, comprising a computer program, the computer program is executed by the processor to realize the method provided by the foregoing embodiments.

[0130] The collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in the present application comply with relevant laws and regulations and do not violate public order and good customs.

[0131] It should be noted that the personal information from the user should be collected for legal and reasonable purposes, and should not be shared or sold outside these legal uses. In addition, such collection / sharing should be carried out after obtaining the informed consent of the user, including but not limited to informing the user to read the user agreement / user notice before the user uses the function, and signing the agreement / authorization including authorization of relevant user information. In addition, any necessary steps should be taken to protect and ensure access to such personal information data, and to ensure that other people with access to personal information data comply with their privacy policy and processes.

[0132] The present application contemplates an implementation that provides users with the ability to selectively opt in or opt out of permitting the collection and / or use of their personal information data. That is, the present disclosure contemplates providing users with the ability to prevent or limit the collection and / or use of their personal information data. For example, the present disclosure contemplates providing users with the ability to prevent or limit the collection and / or use of their personal information data by, for example, blocking or deleting cookies. In addition, the present disclosure contemplates providing users with the ability to determine whether and how to interact with the present disclosure by, for example, blocking web beacons. Further, the present disclosure contemplates providing users with the ability to access and / or edit their personal information data when such data is collected by the present disclosure. In addition, the present disclosure contemplates that the collection and / or use of personal information data can be limited to only those users who expressly consent or give permission to the collection and / or use of their personal information data.

[0133] In the foregoing detailed description, reference is made to the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" etc. which describe a particular feature, structure, material, or characteristic included in at least one embodiment or example of the application. The illustrative examples described in this specification are not necessarily to be construed as being limitations on the scope or functionality of the application, because the example embodiments can be otherwise modified or implemented, or other embodiments can be implemented that are not explicitly described herein. In addition, it is to be understood that the same or equivalent words or phrases can represent the same or equivalent features or functions in different embodiments or examples. Furthermore, it is to be understood that the features or functions of the various examples or embodiments can be combined or separated into other examples or embodiments.

[0134] In addition, the terms "first", "second", etc. are used herein only to describe various features, and do not imply relative importance or a number of the features. Thus, a feature defined with "first", "second", etc. can include at least one of the feature, and can be explicitly or implicitly included in the description of the application. In the description of the application, the meaning of "a plurality" is at least two, such as two, three, etc., unless otherwise specifically defined.

[0135] Any processes or methods described in the flow charts or elsewhere in this specification can be understood as representing code modules, segments, or portions of code that include one or more executable instructions for implementing specific logic functions (or steps in a method). The various embodiments of the application can include additional or fewer steps or processes, and the order of the steps or processes can be altered, as will be apparent to those of skill in the art. The various embodiments of the application can also include additional or alternative features, as will be apparent to those of skill in the art.

[0136] The logic and / or steps represented in flow diagrams or otherwise described herein, for example, can be considered as a sequence of executable instructions, and can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a product of the manufacturing and / or processing. The computer-readable medium can include, but is not limited to, the following: an electronic connection (an electronic device having one or more wires), a portable computer diskette (a magnetic device), a RAM (random access memory), a ROM (read-only memory), an EPROM (erasable programmable ROM) or Flash memory, an optical fiber device, and a portable CD ROM. Additionally, the computer-readable medium can be paper or another suitable medium upon which the program is printed, as the program can be electronically captured, via, for example, an optical scanner, then compiled, interpreted, or otherwise processed, and stored in a computer memory in order to be executed.

[0137] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, various steps or methods can be implemented in software or firmware that is stored in memory and executed by a suitable instruction execution system. As such, in some embodiments, specifically configured hardware can be used to implement at least some of the functionality described herein. For example, if implemented in hardware, the hardware can include any or a combination of the following: a discrete logic circuit having logic gates for implementing logic functions upon data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0138] Those of skill in the art would understand that information and signals can be represented using any of a variety of technologies and techniques. For example, data, instructions, commands, information, signals, bits, symbols, and chips that can be referenced throughout the above description can be represented by voltages, currents, electromagnetic waves, magnetic fields or particles, optical fields or particles, or any combination thereof.

[0139] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing module, or each of the units can be physically present separately, or two or more units can be integrated in one module. The integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0140] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A method for extracting typical modes of new energy planning based on explainable artificial intelligence, characterized in that, The method comprises the following steps: Setting source and load boundary conditions to generate candidate new energy configuration scenarios; Running simulation based on the candidate new energy configuration scenarios to generate high-dimensional time series data sets; Using unsupervised clustering algorithm to extract patterns from the high-dimensional time series data sets to obtain multiple operating modes; Based on the relationship between the key factors and the multiple operating modes, the key factors are identified and the interpretability modeling is performed, including data preprocessing and clustering analysis of the high-dimensional time series data to obtain preprocessed data, obtaining the Shapley value of the preprocessed data by using the Shapley value method of Monte Carlo sampling based on the preprocessed data, and extracting the Shapley value meeting the preset value as the key factor, constructing an explainable classification model based on the key factor, and obtaining a predicted clustering label, analyzing the predicted clustering label by using the explainable classification model to obtain the relationship between the key factor and the multiple operating modes; the relationship between the key factor and the multiple operating modes is obtained by using the explainable classification model to analyze the predicted clustering label, including: taking the key factor as the input variable of the explainable classification model, obtaining the classification results of precision, recall, accuracy and F1 value under different new energy penetration rate scenarios, and obtaining the relationship between the key factor and the multiple operating modes based on the classification results; wherein, in order to verify the representativeness of the extracted key factor, an explainable classification model based on the key factor is constructed to predict the clustering label, the restoration ability of the key factor to the clustering result is evaluated, and the reliability and consistency of the Shapley analysis result are verified through the classification accuracy and model stability index; Based on the relationship between the key factor and the multiple operating modes and the engineering evaluation index, it is judged whether the current scheme meets the planning target, and if not, the iteration is reprocessed.

2. The new energy planning typical mode extraction method based on explainable artificial intelligence according to claim 1, characterized in that, The setting of source and load boundary conditions to generate candidate new energy configuration scenarios comprises: Determine the representative load curve, meteorological data year, resource potential, technical constraint, power grid transmission capacity and cost boundary, and generate candidate new energy configuration scenarios. 3.The method of claim 1, wherein, The running simulation based on the candidate new energy configuration scenarios to generate high-dimensional time series data sets comprises: Build a production simulation tool, and run simulation based on the candidate new energy configuration scenarios to obtain the output of all units including new energy units, line flow and load data, and generate high-dimensional time series data sets. 4.The method of claim 1, wherein, The Shapley value calculation formula is: where, is the full set of features, is a feature any subset of features other than denotes a subset of features the profit function of the model, is a feature Shapley value of 5.The method of claim 1, wherein the method further comprises: The judgment of whether the current scheme meets the planning target based on the relationship between the key factor and the multiple operating modes and the engineering evaluation index, and the reiteration if not comprises: Calculate the model prediction accuracy, and evaluate the operation cost, the curtailment rate, the regulation capacity and the carbon emission level index; If not as expected, adjust the parameters in the source and load boundary conditions and reiterate.

6. An interpretable artificial intelligence-based new energy planning typical mode extraction device, characterized in that, The method comprises the following steps: A scenario generation module sets source and load boundary conditions to generate candidate new energy configuration scenarios; A data acquisition module runs simulation based on the candidate new energy configuration scenarios to generate high-dimensional time series data sets; a data extraction module, which extracts patterns from the high-dimensional time series data set by using an unsupervised clustering algorithm to obtain a plurality of operation modes; an identification module, which identifies key factors based on Monte Carlo Shapley values and builds an explainable model to obtain the relationship between the key factors and the plurality of operation modes, including performing data preprocessing and clustering analysis on the high-dimensional time series data to obtain preprocessed data, obtaining Shapley values of the preprocessed data by using a Monte Carlo sampling Shapley value method based on the preprocessed data, extracting Shapley values meeting preset values as key factors, building an explainable classification model based on the key factors, and obtaining a predicted clustering label, analyzing the predicted clustering label by using the explainable classification model, and obtaining the relationship between the key factors and the plurality of operation modes; the analyzing the predicted clustering label by using the explainable classification model and obtaining the relationship between the key factors and the plurality of operation modes includes taking the key factors as input variables of the explainable classification model, obtaining classification results of precision, recall, accuracy, and F1 value under different new energy penetration rate scenarios, and obtaining the relationship between the key factors and the plurality of operation modes based on the classification results; wherein, to verify the representativeness of the extracted key factors, an explainable classification model based on the key factors is built to predict the clustering label, the reduction ability of the key factors to the clustering result is evaluated, and the reliability and consistency of the Shapley analysis result are verified through classification accuracy and model stability indicators; a judgment module, which judges whether a current scheme meets a planning target based on the relationship between the key factors and the plurality of operation modes and engineering evaluation indicators, and if not, reiterates the processing.

7. An electronic device, comprising: comprise: a processor, and a memory connected to the processor in communication; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to implement the steps of the method for extracting typical operation modes of new energy planning based on explainable artificial intelligence according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed by the processor to implement the steps of the method for extracting typical operation modes of new energy planning based on explainable artificial intelligence according to any one of claims 1-5.

Images