Power distribution network planning diagnosis method based on meta-learning reasoning

Abstract

For new business scenes such as a distributed power supply access power distribution network, the number of planned diagnosis samples is very small, power distribution network planning model creation and semantic knowledge processing are carried out by adopting a meta-learning method, and novel power distribution network diagnosis model knowledge reasoning is utilized to assist novel power distribution network target network frame planning construction. According to the power distribution network planning and diagnosing method based on meta-learning reasoning, on the basis of a meta-learning model, the generalization ability of the meta-learning model is combined with specific diagnosis indexes, and the small sample learning ability is processed by meta-learning, so that a novel power distribution network planning and diagnosing method is realized; the problems that a novel power distribution network planning and construction digital development foundation is not firm, the data sensing capacity is not high, the service fusion degree is insufficient, and distributed new energy network access and other new service samples are not complete and insufficient are solved.

Description

Technical Field

[0001] The present invention provides an intelligent recommendation method for power standard indicator clauses, which uses equipment operation and maintenance diagnosis meta-learning semantic retrieval service to retrieve distribution network structure normative standard data, power grid supply capacity and distribution network transfer capacity indicator data, establishes various evaluation frameworks for distribution network diagnosis indicators, uses the indicator diagnosis evaluation framework to perform grid evaluation, conducts post-effectiveness evaluation on the evaluation results, and continuously optimizes the distribution network diagnosis indicator evaluation capability. Background Art

[0002] Carry out standard digitalization principle and power digital standard meta-feature extraction, build standard digitalization technology system, propose distribution network planning diagnosis path method, study power digital standard features, use extension feature analysis, analyze from multiple dimensions such as business and elements, obtain distribution network planning diagnosis feature knowledge base, study distribution network planning model for new power system, and form distribution network planning semantic extraction model tool for new power system. Meta-feature analysis of distribution network planning diagnosis status, using qualitative and quantitative analysis methods, multiple fields in new power system, such as renewable energy, energy storage, smart grid, etc., all involve the integration and intersection of different technologies, and corresponding standards need to be formulated to regulate their development. At the same time, the original power system needs to be transformed and upgraded during the digital transformation process, which also requires the unification and coordination of standards. Decompose the enterprise strategic goals layer by layer and formulate a set of key performance indicators suitable for this business planning scenario. The indicator system specifically refers to what needs to be evaluated clearly, measurable means that it can be quantified or expressed with specific data, achievable means that the indicator can be achieved through efforts, relevant means that the indicator is related to other aspects of the work, and time limit refers to the time limit for completing the work. Construct an indicator-based meta-learner to assist planning and diagnosis; propose a multi-professional collaborative development path for the company's standard digitalization based on indicators in professional fields such as carbon monitoring, power supply, equipment operation and maintenance, and grid standard operation. The application is oriented to the four aspects of standard management, standard service, standard implementation, and standard decision-making. Through data intelligence drive, it forms standard collaborative management capabilities based on factor integration, standard digital service capabilities based on knowledge empowerment, standard intelligent implementation capabilities based on tool support, and standard scientific decision-making and planning capabilities based on data drive. Summary of the invention

[0003] The present invention is implemented by adopting the following technical solutions:

[0004] A. Make full use of the fact that the electricity consumption of an industry is directly related to its actual production and operation conditions to analyze the correlation between industries. On the one hand, this can prevent the traditional economic analysis from being affected by inflation, exchange rates, and tampering and fabrication factors. On the other hand, it can release the social and economic information contained in the electricity data and expand the analysis ideas of the industry economy.

[0005] B. Using the strength of the industry electricity correlation relationship instead of the industry electricity consumption volume to measure the importance of industry electricity can avoid overestimating high-energy-consuming and low-value-added industries and underestimating low-energy-consuming and high-value-added industries. It can relatively objectively reflect the economic status of the industry in the region and provide energy support for the government to formulate industrial development strategies. (For example, although some industries have low electricity consumption, they have a high degree of centrality, and relocation may have a greater impact on other industries; for example, some high-energy-consuming and high-pollution industries are analyzed as the local industry center, and relocation may have an overall impact on local industries.

[0006] C. By comparing the topological distribution in different periods, it is possible to reflect the changes in industry relationships and provide energy feedback for the implementation effect of the government's industrial policies. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 Meta-learning feature index reasoning distribution network planning diagnosis flow chart DETAILED DESCRIPTION

[0008] like Figure 1 Step 1: Plan the construction of the diagnostic capability level evaluation index system:

[0009] Standard refers to the content of the standard itself. Mapping to the machine-readable capability level dimension corresponds to the capability requirements of the standard label set.

[0010] Access refers to the access to data and information in the standard. Mapped to the machine-readable capability level dimension, it corresponds to the capability requirements of the standard information model.

[0011] Information refers to the data and information in the standard. Mapping to the machine-readable standard capability level dimension corresponds to the capability requirements of the standard information model.

[0012] Functionality refers to user-oriented functions and services. Mapping to the machine-readable standard capability level dimension corresponds to the capability requirements for semantic interoperability of the standard management framework.

[0013] Business refers to the organization and execution process of business. Mapping to the machine-readable standard capability level dimension corresponds to the capability requirements for digital application services.

[0014] Step 2: Planning indicator meta-method diagnostic model training:

[0015] First, construct the evaluation index hierarchy. Constructing the hierarchy is to stratify the entire evaluation system according to the degree of importance, which is an analytical process for the evaluation of digital capabilities. The digital capability evaluation is divided into three levels, of which the top level is the target level, which is also the value level of the standard. The second level is the criterion level. In the relevant standards of the power system, the core criteria for the evaluation of digital capabilities should be standard completeness, accessibility, editing capabilities, etc. The third level is the indicator level, which is the various factors related to the above indicators.

[0016] Next, we construct a comparison matrix. The judgment matrix is ​​the relative importance of the elements in the same layer after comparing them two by two. For the second-layer indicator, we construct the indicator layer judgment matrix as follows:

[0017]

[0018] When constructing the judgment matrix, the nine-point scale is used to quantitatively describe the importance, as shown in the following table:

[0019] TABLE 1. Definition of the Rule of Nine Scale

[0020]

[0021]

[0022] The indicators are divided into four categories. Where represents the value obtained by measuring the ith indicator, and represents the maximum and minimum threshold values ​​of the ith indicator respectively, and the value obtained after processing is expressed as.

[0023] Step 3: Use the diagnostic indicator model to calculate the indicator metamodel:

[0024] The resulting value is expressed as .

[0025] Positive correlation index: The larger the value, the better the evaluation result. The calculation formula is as follows:

[0026]

[0027] Anti-correlation index: The larger the value, the worse the evaluation result. The calculation formula is as follows:

[0028]

[0029] 0-1 type indicator: Its value is mandatory, that is, the indicator has only two mutually opposing situations. The calculation of this type of indicator is as follows:

[0030]

[0031] Fuzzy indicators: It is usually inconvenient to describe them with precise values, but they can be described with verbal variables as "very good", "average", "poor", etc., such as product quality, service quality, etc. Defuzzify them and convert them into clear values ​​between 0 and 100.

[0032] Step 4: Analysis of indicator diagnosis results:

[0033] Content analysis: Analyze the content of the manuscript to ensure that all necessary information and content has been covered. This includes checking the introduction, scope, application areas, definitions, terminology, requirements, test methods, etc. of the standard.

[0034] Figures and appendices: Evaluate whether the figures, images, appendices, etc. in the manuscript are complete and conform to standard formats and specifications.

[0035] Interactivity check: If your manuscript contains links, interactive elements, or participatory content (such as comments or feedback), you need to ensure that these elements work properly and provide complete information.

[0036] Internal consistency of the standard: Check whether there is consistency between the various parts of the manuscript. The standard manuscript should be a coherent whole and should not have internal contradictions.

[0037] External consistency: Check the consistency between the manuscript and other relevant standards or documents. This includes the unified definition of terms, consistent use of symbols, etc.

[0038] Step 5: Target grid diagnostic verification:

[0039] Retrieve the normative diagnosis and verification of the target grid of the distribution network, the diagnosis and verification of the grid supply capacity and the distribution network transfer capacity, use the meta-learning indicator diagnosis and evaluation framework to conduct grid evaluation, conduct post-evaluation of the evaluation results, continuously optimize the diagnostic indicator evaluation framework structure, and continuously improve the accuracy of the meta-learner distribution network diagnosis.

Claims

1. A distribution network planning diagnosis method using meta-learning reasoning, whose characteristics provide A meta-learning small sample semantic extraction method is used to extract distribution network diagnosis planning indicators using semantics to calculate distribution network planning.

2. The method according to claim 1, characterized in that A feature extraction method for distributed renewable energy access to the network is provided to extract feature knowledge of new distribution network services.

3. The method according to claim 1, characterized in that Handle model training for small samples and use the semantic knowledge structure of meta-learning to complete samples.

4. The method according to claim 1, characterized in that Convert planning problems into knowledge calculations of indicators, use the calculation results to make scientific inferences on the diagnosis, and assist in the planning of the target grid.

Images