Power system feature recognition-based power grid project reserve demand prediction method and system
Patent Information
- Application Number
- CN202411121837.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-15
- Publication Date
- 2024-11-12
AI Technical Summary
In the prior art, power grid project reserve management has problems such as insufficient investment focus, unbalanced investment in regional and professional fields, insufficient total reserves, and poor quality of reserves, resulting in inaccurate annual project reserve forecasts.
Using a method based on power system feature recognition, a new power system evolution feature characterization model is established by obtaining power grid project data, pre-processing, and using the LSTM algorithm, and a project reserve demand prediction model is established in combination with the decision tree algorithm to achieve accurate prediction of power grid project reserve demand .
It improves the accuracy of annual project reserve forecasts of the power grid, ensures reasonable investment focus, balanced investment in regions and professional fields, sufficient total reserves, and improved reserve quality, solving multiple problems in reserve management.
Smart Images

Figure CN118941245A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of power data processing, and in particular to a method and system for predicting reserve demand of power grid projects based on power system feature identification. Background Art
[0002] Projects are the cells and cornerstones of economic and social development and corporate development. As the starting stage of project work, project reserve is a key link in the company's innovative development, investment decision-making, and business promotion. Doing a good job in project reserve and promoting investment growth is an important means to implement the development concept, accelerate the construction of a development pattern, and strive to promote high-quality development.
[0003] Projects are also carriers of investment and platforms for development. Giving full play to the layout of project reserves in key and core areas can support economic and social development, lead industrial transformation and upgrading, and promote sustained, stable and healthy economic and social development. Grasping the planning and layout of project reserves is an important way to achieve high-quality economic development.
[0004] Currently, the company's project reserve management still has problems such as unclear investment priorities, unbalanced investment in regions and professional fields, insufficient total reserve volume, and low reserve quality. Summary of the invention
[0005] The purpose of this application is to provide a method and system for predicting power grid project reserve demand based on power system feature identification, which can improve the accuracy of the power grid in making annual project reserve forecasts and solve the problems existing in the prior art, such as lack of investment focus, unbalanced investment in regions and professional fields, insufficient total reserve, and low reserve quality.
[0006] To achieve the above objectives, this application provides the following solutions:
[0007] In a first aspect, the present application provides a method for predicting reserve demand of power grid projects based on power system feature identification, comprising:
[0008] Acquire power grid project data; the power grid project data includes: power system composition data, power grid project demand data and project reserve data;
[0009] Preprocessing the power grid project data to obtain preprocessed data;
[0010] A new power system evolution feature characterization model is established using the LSTM algorithm;
[0011] Inputting the preprocessed data into the novel power system evolution characteristic characterization model to obtain characteristic results;
[0012] Use decision tree algorithm to establish project reserve demand forecasting model;
[0013] The characteristic results are input into the project reserve demand forecasting model to obtain the power grid project reserve demand forecasting results.
[0014] Optionally, preprocessing the power grid project data to obtain preprocessed data specifically includes:
[0015] Using the DBSCAN algorithm to perform abnormal data detection on the power grid project data to obtain an abnormal data detection result, and removing abnormal data from the power grid project data based on the abnormal data detection result;
[0016] The format of the power grid project data from which abnormal data is removed is converted to obtain the preprocessed data.
[0017] Optionally, the DBSCAN algorithm is used to perform abnormal data detection on the power grid project data, specifically including:
[0018] Determine the Eps parameter and the MinPts parameter, and define the density threshold as the number of MinPts data points within a circle with the Eps parameter as the radius;
[0019] Using the spatial distribution characteristics of the power grid project data itself, a density threshold list is generated based on the K-means nearest neighbor algorithm and the mathematical expectation method;
[0020] The DBSCAN algorithm is used to perform abnormal data detection on the power grid project data based on the density threshold list.
[0021] Optionally, the process of determining the Eps parameter and the MinPts parameter includes:
[0022] Determining a distance distribution matrix of the power grid project data;
[0023] After sorting each row element in the distance distribution matrix in ascending order, the column elements are used to form a K-nearest neighbor distance vector of all data points in the power grid project data;
[0024] The elements in the K-nearest neighbor distance vector are averaged to obtain the K-average nearest neighbor distance of the K-nearest neighbor distance vector, and the K-average nearest neighbor distance is used as a candidate Eps parameter;
[0025] Determine the mathematical expectation value of the number of Eps neighborhood objects of all elements in the distance distribution matrix, and use the mathematical expectation value of the number of Eps neighborhood objects as the MinPts parameter of the power grid project data;
[0026] The candidate Eps parameters corresponding to different K values are selected in turn, and the MinPts parameter of the power grid project data corresponding to this candidate Eps parameter is combined to obtain the number of clusters generated under different K values, until the number of clusters remains unchanged for many times, and the maximum K value corresponding to the number of clusters at this time is taken as the optimal K value, and the candidate Eps parameter corresponding to the optimal K value is taken as the Eps parameter, and the MinPts parameter corresponding to the Eps parameter is taken as the optimal MinPts parameter.
[0027] Optionally, a new power system evolution feature characterization model is established using an LSTM algorithm, specifically including:
[0028] Constructing a first training data set and a first test data set; the first training data set and the first test data set are both data sets generated using historical data and include a plurality of labeled sample pairs; the historical data include historical composition data of the power system, historical demand data of power grid projects, and historical project reserve data;
[0029] A first initial model is constructed using an LSTM algorithm, and the first training data set and the first test data set are used to train and test the first initial model until set conditions are met, and the trained and tested first initial model is used as the new power system evolution feature characterization model.
[0030] Optionally, a decision tree algorithm is used to establish a project reserve demand forecasting model, specifically including:
[0031] Inputting the historical data into the novel power system evolution characteristic characterization model to obtain characteristic prediction results;
[0032] Constructing a second training data set and a second test data set based on the historical data and feature prediction results corresponding to the historical data;
[0033] A second initial model is constructed using a decision tree algorithm, and the second training data set and the second test data set are used to train and test the second initial model until the set conditions are met, and the trained and tested second initial model is used as the project reserve demand forecasting model.
[0034] Optionally, the first initial model includes one or more of an input gate, a forget gate and an output gate.
[0035] Optionally, the decision tree algorithm adopted is a CART algorithm.
[0036] In a second aspect, the present application provides a power grid project reserve demand forecasting system based on power system feature identification, which is used to implement the power grid project reserve demand forecasting method based on power system feature identification provided above; the system comprises:
[0037] A data acquisition module is used to acquire power grid project data; the power grid project data includes: power system composition data, power grid project demand data and project reserve data;
[0038] A preprocessing module, used for preprocessing the power grid project data to obtain preprocessed data;
[0039] The first model building module is used to establish a new power system evolution feature characterization model using the LSTM algorithm;
[0040] A feature extraction module, used for inputting the preprocessed data into the novel power system evolution feature characterization model to obtain feature results;
[0041] The second model building module is used to establish a project reserve demand forecasting model using a decision tree algorithm;
[0042] The project reserve demand forecasting module is used to input the characteristic results into the project reserve demand forecasting model to obtain the power grid project reserve demand forecasting results.
[0043] According to the specific embodiments provided in this application, this application discloses the following technical effects:
[0044] The method and system for predicting the reserve demand of power grid projects based on power system feature identification provided by the present application analyzes the factors that may affect the power system through a new power system evolution feature characterization model established by the LSTM algorithm, obtains feature results, and then realizes the prediction results of power grid project reserve demand through a project reserve demand prediction model established by a decision tree algorithm, so as to improve the accuracy of the power grid in annual project reserve prediction. In addition, the number of projects required and the specific impact of the feature results on project demand can be determined based on the obtained power grid project reserve demand prediction results, so as to solve the problems existing in the prior art, such as lack of investment focus, imbalance in regional and professional field investment, insufficient total reserve, and low reserve quality. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0046] Figure 1 This is an application environment diagram of a method for predicting power grid project reserve demand based on power system feature identification in one embodiment of the present application;
[0047] Figure 2A flowchart of a method for predicting power grid project reserve demand based on power system feature identification provided in one embodiment of the present application;
[0048] Figure 3 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION
[0049] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0050] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.
[0051] The power grid project reserve demand prediction method based on power system feature identification provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set up separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send the power grid project data to the server 104. After the server 104 receives the power grid project data, the server 104 pre-processes the power grid project data for the power grid project data to obtain pre-processed data; inputs the pre-processed data into the constructed new power system evolution feature characterization model to obtain feature results; inputs the feature results into the constructed project reserve demand prediction model to obtain the power grid project reserve demand prediction result. The server 104 can feed back the obtained power grid project reserve demand prediction result to the terminal 102. In addition, in some embodiments, the power grid project reserve demand prediction method based on power system feature identification can also be implemented separately by the server 104 or the terminal 102, such as the terminal 102 can directly predict the power grid project reserve demand for the power grid project data, or the server 104 can obtain the power grid project data from the data storage system and predict the power grid project reserve demand for the power grid project data.
[0052] The terminal 102 may be, but is not limited to, various desktop computers, laptop computers, smart phones, tablet computers, IoT devices, and portable wearable devices. The IoT devices may be smart speakers, smart TVs, smart air conditioners, smart vehicle-mounted devices, etc. The portable wearable devices may be smart watches, smart bracelets, head-mounted devices, etc. The server 104 may be implemented as an independent server or a server cluster consisting of multiple servers, or may be a cloud server.
[0053] In an exemplary embodiment, in view of the current problems of the company's project reserve management, such as lack of investment focus, imbalance in regional and professional field investment, insufficient total reserve, and low reserve quality, it is urgent to improve the internal mechanism and technical means of scientific decision-making and risk control, comprehensively consider the disturbance of internal and external factors and the differences in various needs, and build a project reserve advanced expansion layout mechanism that adapts to internal and external evolution, propose a multi-level reserve project selection and configuration method, develop reserve project status tracking and evaluation and active regulation technology, dynamically adjust the project reserve capacity, implement a batch, reserve a batch, and adjust a batch to support the company's project management level and investment efficiency. The development of long short-term memory networks (LSTM) and random forests can provide important technical support for feature identification and prediction in power systems, especially in the analysis of project reserve demand for new power systems. It shows unique value. Random forests can effectively process a large amount of nonlinear data and improve the accuracy and robustness of predictions by integrating multiple decision trees. This is particularly important when processing complex power system data, such as load prediction, fault detection, etc. Combining these two technologies, the stability and load demand of the power system can be more accurately evaluated, providing data support for the design and upgrading of the power system. By analyzing historical data and real-time data, LSTM and random forest can help decision makers better understand the changing trends and potential risks of electricity demand, so as to make more scientific and reasonable decisions on project reserves and resource allocation. This is crucial to ensure the reliability and efficiency of the power system, especially in emerging areas such as new energy access and distributed generation. Based on this, Figure 2 As shown, a method for predicting the reserve demand of power grid projects based on power system feature recognition is provided. The method is executed by a computer device, and can be executed by a computer device such as a terminal or a server alone, or by a terminal and a server together. In the embodiment of the present application, the method is applied to Figure 1 The server 104 in the example is used as an example to illustrate, including the following steps 200 to 205. Among them:
[0054] Step 200: Obtain power grid project data. The power grid project data includes: power system composition data, power grid project demand data and project reserve data.
[0055] For example, data such as the project reserves of the power system grid and the composition of the power system are obtained. Other influencing factors of the power grid project reserves include technology, geography, market and other data.
[0056] Extract key information such as direction, trend, scale, progress, etc. from various aspects that may affect project reserves, such as policies, technologies, markets, businesses, geography, and power system composition. Among them, the policy level can be to collect policy documents and regulations issued by relevant governments and regulatory agencies. The technical level can be to collect patent data, scientific research papers, technical reports, etc. The market level can be to obtain industry reports, market research data, sales data, etc. The business level can obtain information from corporate annual reports, financial statements, business profiles, etc. The geographical level can collect data on the geography, climate, population, economic development level, etc. of the relevant regions.
[0057] Step 201: pre-process the power grid project data to obtain pre-processed data.
[0058] In actual application, this step can be to clean the data set formed by the power grid project data to remove irrelevant information and convert it into a format that is easy to analyze. Among them, for numerical data, normalization can also be performed to facilitate comparison at the same scale.
[0059] Step 202: Use the LSTM algorithm to establish a new power system evolution feature representation model.
[0060] Step 203: input the preprocessed data into the new power system evolution characteristic characterization model to obtain characteristic results.
[0061] Step 204: Use a decision tree algorithm to establish a project reserve demand forecasting model.
[0062] Step 205: Input the characteristic results into the project reserve demand forecasting model to obtain the power grid project reserve demand forecasting results.
[0063] By implementing the above steps 200 to 205, the present application uses a new power system evolution feature characterization model established by the LSTM algorithm to analyze the factors that may affect the power system, obtain feature results, and then use the project reserve demand prediction model established by the decision tree algorithm to achieve the power grid project reserve demand prediction results, so as to improve the accuracy of the power grid in the annual project reserve prediction. In addition, based on the obtained power grid project reserve demand prediction results, the required number of projects and the specific impact of the feature results on project demand can be determined to solve the problems of the existing technology such as lack of investment focus, imbalance in regional and professional field investment, insufficient total reserve, and low reserve quality.
[0064] In addition, this application uses the LSTM algorithm to analyze the factors that may affect the new power system to obtain characteristic results, and then uses the decision tree algorithm to analyze the characteristic results to obtain project reserve requirements, determine the number of projects required, and the specific impact of the characteristic results on project demand, and then, with the goal of improving the company's project management level and investment efficiency, establish a set of new power systems to ensure that the project reserve layout matches development needs and accurately determines the project reserve layout under the construction background.
[0065] In another exemplary embodiment of the present application, in order to detect anomalies in power grid project data and remove abnormal data to reduce the interference caused by abnormal data, the preprocessing method used in the present application may be the DBSCAN algorithm. The DBSCAN algorithm needs to determine two parameters, Eps and MinPts, to indicate the degree of compactness of the distribution of data points, and divide areas with sufficient density into clusters to form the largest set of points connected by density. Based on this, the process of using the DBSCAN algorithm to detect abnormal data in power grid project data includes:
[0066] 1. Generate Eps parameter list.
[0067] Step 1. Calculate the distance distribution matrix of the data set D formed by the power grid project data, which is:
[0068] D n×n ={Dist(i,j)|1≤i≤n,1≤j≤n} (1)
[0069] Where D n×n is a real symmetric matrix of n×n, i.e., a distance distribution matrix. n is the number of objects (i.e., data points) contained in the dataset D. Dist(i, j) is the distance from the i-th object to the j-th object in the dataset D.
[0070] Step 2. Distance distribution matrix D n×n The elements of each row of are sorted in ascending order, then the distance vector D0 composed of the elements in the first column represents the distance from the object to itself, which is 0. The elements in the Kth column constitute the K-nearest neighbor distance vector D of all data points (i.e. objects). K .
[0071] Step 3. K-nearest neighbor distance vector D K The K-nearest neighbor distance vector D can be obtained by averaging the elements in K K-average nearest neighbor distance And K-average nearest neighbor distance As candidate Eps parameters. Calculate all K values to get the Eps parameter list D Eps , expressed as:
[0072]
[0073] 2. Generate MinPts parameter list.
[0074] The MinPts parameter list is generated using the mathematical expectation method. For a given Eps parameter list D Eps , find out the number of Eps neighborhood objects corresponding to each Eps parameter in turn, and calculate the mathematical expectation of the number of Eps neighborhood objects of all objects, and use this mathematical expectation as the neighborhood density threshold MinPts parameter of the data set D, which is:
[0075]
[0076] Where P i is the number of Eps neighborhood objects of the i-th object, and n is the total number of objects in the dataset D.
[0077] Furthermore, in the above data table selection process, the optimal Eps and MinPts parameters are adaptively determined, and the determination process includes the following steps:
[0078] Step 1). Based on the above description, calculate the Eps parameter list D of the data set D according to the K-average nearest neighbor method Eps Then, the K-average nearest neighbor distances corresponding to different K values (K = 1, 2, ..., n) are selected in turn, that is, the Eps parameter list D is selected in turn. Eps The elements in are used as candidate Eps parameters, and the MinPts parameters are obtained by formula (3). The obtained candidate Eps parameters and MinPts parameters are input into the DBSCAN algorithm to perform clustering analysis on the data set, and the number of clusters generated under different K values is obtained. When the number of generated clusters is the same for three consecutive times, it is considered that the clustering result is stable, and the number of clusters N is recorded as the optimal number of clusters.
[0079] Step 2). Continue to execute step 1) until the number of generated clusters is no longer N, and select the maximum K value corresponding to the number of clusters N as the optimal K value. K-average nearest neighbor distance corresponding to the optimal K value As the optimal Eps parameter, the MinPts parameter corresponding to the optimal Eps parameter is the optimal MinPts parameter.
[0080] Step 3). After obtaining the optimal Eps and MinPts parameters, perform abnormal data removal and combine the corresponding influencing factors with the power grid project data to form a data set containing various influencing factors and reserve projects.
[0081] In another exemplary embodiment of the present application, the development of LSTM (Long Short-Term Memory Network) and random forest provides important technical support for feature identification and prediction in power systems, especially in the analysis of project reserve requirements of new power systems. Random forest can effectively process a large amount of nonlinear data and improve the accuracy and robustness of prediction by integrating multiple decision trees. Based on this, the implementation process of the above step 202 is:
[0082] Step 2021. Construct a first training data set and a first test data set. The first training data set and the first test data set are both data sets generated using historical data and include multiple labeled sample pairs. The historical data includes historical composition data of the power system, historical demand data of power grid projects, and historical project reserve data.
[0083] Step 2022. Use the LSTM algorithm to build a first initial model, and use the first training data set and the first test data set to train and test the first initial model until the set conditions are met, and use the trained and tested first initial model as the new power system evolution feature characterization model.
[0084] This embodiment uses the LSTM algorithm to establish a new power system evolution feature characterization model, and obtains independent characteristic factors that affect the reserves of different types of projects. For example, keywords in policies. A certain technical route in technology. The geographical advantages of certain technologies in terms of geographical location, etc. However, these features are independent of each other at this time. Then, by bringing these features into the random forest model (i.e., the project reserve demand prediction model) for cluster analysis, the specific impact of these different features on the demand of different projects and the difference in the degree of impact can be obtained.
[0085] In another exemplary embodiment of the present application, the process of constructing the first training data set and the first test data set also requires preprocessing of historical data, and the preprocessing method used is the DBSCAN algorithm provided above. The process of preprocessing historical data by the DBSCAN algorithm can be referred to the above description. After the optimal Eps and MinPts, abnormal data processing is performed, and the corresponding influencing factors are combined with the historical data to form a sample data set containing various influencing factors and reserve items. It will not be repeated here. The input parameters and output parameters in the training data set are denoted as X train and Y train , the input parameters and output parameters in the test data set are denoted as X test and Y test .
[0086] Since the sample data set contains data of different dimensions such as power grid composition, technological development factors and other related influencing factors, if the parameters of some factors are not processed, the influence will be ignored, which will also affect the training efficiency of the prediction model. In addition, adding data normalization can also improve the convergence speed of the algorithm and the training efficiency of the model. Therefore, the historical data is normalized, and the formula is as follows:
[0087]
[0088] In the formula, y′ test_i is the normalized data. test_i is the training data. test_max ,y test_min are the maximum and minimum values in the data set.
[0089] In another exemplary embodiment of the present application, there are three types of gating in the LSTM unit, namely: input gate, forget gate and output gate. The gating can be regarded as a fully connected layer, and the storage and update of information by LSTM is realized by these gating.
[0090] The general form of gating can be expressed as:
[0091] f t =σ(W fx ·x t + W fh h t-1 +b f ) (5)
[0092] i t =σ(W ix· x t + W ih· h t-1 +b i ) (6)
[0093] g t =tanh(W gx· x t +W gh· h t-1 +b g ) (7)
[0094] o t =σ(W ox· x t + W oh· h t-1 +b o ) (8)
[0095] S t =g t ⊙i t +S t-1⊙f t-1 (9)
[0096] h t =tanh(S t )⊙o t (10)
[0097] In the formula, the input and output vectors of the hidden layer are x and t and h t , time step is t, f t 、i t , g t , o t 、h t and t are the states of the forget gate, input gate, input node, output gate, intermediate output, and state unit respectively. fx , W fh , W ix , W ih , W gx , W gh , W ox and W oh are the corresponding gates and input x t and intermediate output h t-1 The matrix weights to be multiplied; b f 、b i 、b g and b o are the bias terms of the corresponding gates, σ represents the change of the sigmoid function, tanh is the hyperbolic tangent activation function, and ⊙ represents the bitwise multiplication of the elements in the vector.
[0098] In another exemplary embodiment of the present application, random forest can assist decision makers to better understand the changing trend and potential risks of electricity demand, so as to make more scientific and reasonable decisions on project reserves and resource allocation. This is crucial to ensure the reliability and efficiency of the power system, especially in emerging fields such as access to new energy and distributed generation. Based on this, the implementation process of the above step 204 can be:
[0099] Step 2041. Input historical data into the new power system evolution feature characterization model to obtain feature prediction results.
[0100] Step 2042: Construct a second training data set and a second test data set based on the historical data and the feature prediction results corresponding to the historical data.
[0101] Step 2043. Use the decision tree algorithm to build a second initial model, and use the second training data set and the second test data set to train and test the second initial model until the set conditions are met, and use the trained and tested second initial model as the project reserve demand forecasting model.
[0102] Furthermore, random forest (RF) is a combination classifier that uses the bootstrap resampling method to extract multiple samples from the original sample, builds a decision tree model for each bootstrap sample, and then combines these decision trees together to obtain the final classification or prediction result through voting. When studying random forests, decision trees must be involved, so the decision tree algorithm can be determined first. For example, if the decision tree algorithm used is the CART algorithm, then the process of using the decision tree algorithm to build the second initial model in step 2043 can be described as:
[0103] 1. When a node splits, the CART algorithm uses the Gini indicator to measure data partitioning. The calculation process is as follows:
[0104] (1) Calculate the Gini coefficient of the sample:
[0105]
[0106] Among them, T i Represents the probability that the category of the i-th object appears in the sample set S.
[0107] (2) Calculate the Gini coefficient of each partition. If the sample set S is divided into two subsets S1 and S2, the Gini coefficient of this partition is:
[0108]
[0109] Where Gini(S1) represents the Gini coefficient of subset S1, and Gini(S2) represents the Gini coefficient of subset S2.
[0110] (3) Suppose a series of decision trees h1(x), h2(x), ..., h k (x) and two random vectors X and Y. For each decision tree, in the process of random forest construction, there is an initial data set and an unsampled data set, and the unsampled data set is denoted as O. k (x). k (x,y j ) is the random vector x in O k The classification category of the vote in (x) is y j The ratio is:
[0111]
[0112] The numerator is the sum of the number of correct classifications of each decision tree and its corresponding unextracted data set, and the denominator is the sum of the number of samples of all unextracted data sets. j) as an estimate of the probability of correct classification by random forest, we can examine the two variables of random forest strength and correlation.
[0113] In another exemplary embodiment of the present application, the error accuracy of the model is evaluated using a test data set, and the prediction ability of the model is tested. Specifically, the normalization parameters of the training set are used to evaluate the error accuracy of the test set data X test Normalize, input the normalized data into the model for calculation, and perform denormalization to obtain the prediction result Y test_pre . The predicted result Y test_pre With the original data Y test The comparison is done by absolute error MAE, root mean square error RMSE and correlation coefficient R 2 Evaluation is performed, corresponding to formula (14) to formula (16).
[0114]
[0115] In the formula, m represents, Represents the test set results, Represents the prediction test set results.
[0116] Based on the test description of this embodiment, if the above data size meets the standard, it means that the model itself has little error. The model established in this application can obtain historical data of the past period of time that is not used to establish the model for simulation calculation. On this basis, combined with the project requirements of that period of time, the specific impact of the characteristic factors is compared to see if it is close to the model speculation. In this way, it can be judged whether the model is accurate.
[0117] In summary, this application has established a new method that can ensure that the project reserve layout matches the development needs and accurately determine the project reserve layout in the context of the accelerated construction of new power systems. This method mainly uses a model that combines LSTM and random forests. Based on historical data, it identifies keywords, semantic structures, and context information analysis through training and pattern matching of massive text data, thereby determining the impact of the construction process of the new power system on reserve demand and the changes brought about by the impact from all factors provided by the power grid. It can also obtain a variety of new features that affect project reserves. These new features can then be used to improve the accuracy and efficiency of the company's preparation of reserve projects, so that it can better improve the company's project management level and investment efficiency.
[0118] Based on the same inventive concept, the embodiment of the present application also provides a power grid project reserve demand forecasting system for implementing the power grid project reserve demand forecasting method based on power system feature identification involved above. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more embodiments of the power grid project reserve demand forecasting system based on power system feature identification provided below can refer to the limitations of the power grid project reserve demand forecasting method based on power system feature identification above, and will not be repeated here.
[0119] In an exemplary embodiment, a power grid project reserve demand forecasting system based on power system feature recognition is provided, comprising:
[0120] The data acquisition module is used to acquire power grid project data, including power system composition data, power grid project demand data and project reserve data.
[0121] The preprocessing module is used to preprocess the power grid project data to obtain preprocessed data.
[0122] The first model building module is used to establish a new power system evolution feature characterization model using the LSTM algorithm.
[0123] The feature extraction module is used to input the preprocessed data into the new power system evolution feature characterization model to obtain feature results.
[0124] The second model building module is used to establish a project reserve demand forecasting model using a decision tree algorithm.
[0125] The project reserve demand forecasting module is used to input the characteristic results into the project reserve demand forecasting model to obtain the power grid project reserve demand forecasting results.
[0126] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Figure 3As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store power grid project reserve demand forecast data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a power grid project reserve demand forecasting method based on power system feature identification is implemented.
[0127] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.
[0128] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.
[0129] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.
[0130] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.
[0131] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).
[0132] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.
[0133] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0134] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.
Claims
1. A method for predicting power grid project reserve demand based on power system feature identification, characterized in that: The power grid project reserve demand prediction method based on power system feature identification includes: Acquire power grid project data; the power grid project data includes: power system composition data, power grid project demand data and project reserve data; Preprocessing the power grid project data to obtain preprocessed data; A new power system evolution feature characterization model is established using the LSTM algorithm; Inputting the preprocessed data into the novel power system evolution characteristic characterization model to obtain characteristic results; Use decision tree algorithm to establish project reserve demand forecasting model; The characteristic results are input into the project reserve demand forecasting model to obtain the power grid project reserve demand forecasting results.
2. The method for predicting power grid project reserve demand based on power system feature identification according to claim 1 is characterized in that: Preprocessing the power grid project data to obtain preprocessed data specifically includes: Using the DBSCAN algorithm to perform abnormal data detection on the power grid project data to obtain an abnormal data detection result, and removing abnormal data from the power grid project data based on the abnormal data detection result; The format of the power grid project data from which abnormal data is removed is converted to obtain the preprocessed data.
3. The method for predicting power grid project reserve demand based on power system feature identification according to claim 2 is characterized in that: The DBSCAN algorithm is used to detect abnormal data on the power grid project data, specifically including: Determine the Eps parameter and the MinPts parameter, and define the density threshold as the number of MinPts data points within a circle with the Eps parameter as the radius; Using the spatial distribution characteristics of the power grid project data itself, a density threshold list is generated based on the K-means nearest neighbor algorithm and the mathematical expectation method; The DBSCAN algorithm is used to perform abnormal data detection on the power grid project data based on the density threshold list.
4. The method for predicting power grid project reserve demand based on power system feature identification according to claim 3 is characterized in that: The process of determining the Eps parameter and the MinPts parameter includes: Determining a distance distribution matrix of the power grid project data; After sorting each row element in the distance distribution matrix in ascending order, the column elements are used to form a K-nearest neighbor distance vector of all data points in the power grid project data; The elements in the K-nearest neighbor distance vector are averaged to obtain the K-average nearest neighbor distance of the K-nearest neighbor distance vector, and the K-average nearest neighbor distance is used as a candidate Eps parameter; Determine the mathematical expectation value of the number of Eps neighborhood objects of all elements in the distance distribution matrix, and use the mathematical expectation value of the number of Eps neighborhood objects as the MinPts parameter of the power grid project data; The candidate Eps parameters corresponding to different K values are selected in turn, and the MinPts parameter of the power grid project data corresponding to this candidate Eps parameter is combined to obtain the number of clusters generated under different K values, until the number of clusters remains unchanged for many times, and the maximum K value corresponding to the number of clusters at this time is taken as the optimal K value, and the candidate Eps parameter corresponding to the optimal K value is taken as the Eps parameter, and the MinPts parameter corresponding to the Eps parameter is taken as the optimal MinPts parameter.
5. The method for predicting power grid project reserve demand based on power system feature identification according to claim 1 is characterized in that: The LSTM algorithm is used to establish a new power system evolution feature representation model, which includes: Constructing a first training data set and a first test data set; both the first training data set and the first test data set are data sets generated using historical data and include a plurality of labeled sample pairs; the historical data include historical composition data of the power system, historical demand data of power grid projects, and historical project reserve data; A first initial model is constructed using an LSTM algorithm, and the first training data set and the first test data set are used to train and test the first initial model until set conditions are met, and the trained and tested first initial model is used as the new power system evolution feature characterization model.
6. The method for predicting power grid project reserve demand based on power system feature identification according to claim 5 is characterized in that: The decision tree algorithm is used to establish a project reserve demand forecasting model, which includes: Inputting the historical data into the novel power system evolution characteristic characterization model to obtain characteristic prediction results; Constructing a second training data set and a second test data set based on the historical data and feature prediction results corresponding to the historical data; A second initial model is constructed using a decision tree algorithm, and the second training data set and the second test data set are used to train and test the second initial model until the set conditions are met, and the trained and tested second initial model is used as the project reserve demand forecasting model.
7. The method for predicting power grid project reserve demand based on power system feature identification according to claim 5 is characterized in that: The first initial model includes one or more of an input gate, a forget gate, and an output gate.
8. The method for predicting power grid project reserve demand based on power system feature identification according to claim 6 is characterized in that: The decision tree algorithm used is the CART algorithm.
9. A power grid project reserve demand forecasting system based on power system feature identification, characterized in that: The power grid project reserve demand forecasting system based on power system feature identification is used to implement the power grid project reserve demand forecasting method based on power system feature identification as described in any one of claims 1 to 8; The system comprises: A data acquisition module is used to acquire power grid project data; the power grid project data includes: power system composition data, power grid project demand data and project reserve data; A preprocessing module, used for preprocessing the power grid project data to obtain preprocessed data; The first model building module is used to establish a new power system evolution feature characterization model using the LSTM algorithm; A feature extraction module, used for inputting the preprocessed data into the novel power system evolution feature characterization model to obtain feature results; The second model building module is used to establish a project reserve demand forecasting model using a decision tree algorithm; The project reserve demand forecasting module is used to input the characteristic results into the project reserve demand forecasting model to obtain the power grid project reserve demand forecasting results.
