Low-voltage distribution network planning method and device based on big data analysis

Abstract

The invention is suitable for the technical field of power distribution networks, and provides a low-voltage power distribution network planning method based on big data analysis, and the method comprises the steps: classifying the power consumption behaviors of different types of users in a power distribution network region, and obtaining a plurality of power consumption behavior similar sub-regions; for each power consumption behavior similar sub-region, generating a load characteristic mode; constructing a multi-dimensional load change model, predicting load modes of different power consumption behavior similar sub-regions through the multi-dimensional load change model based on the load characteristic mode, generating a comprehensive load mode prediction result, calculating to obtain a power grid capacity demand of the power distribution network region, and combining the power grid capacity data of the power distribution network region to obtain a comprehensive load mode prediction result. Generating a power distribution network planning scheme; based on the technical indexes, the economic indexes and the environment indexes, the power distribution network planning scheme is evaluated, the planning level of the low-voltage power distribution network can be remarkably improved, optimal configuration of resources is promoted, and development of an intelligent power grid is promoted.

Description

Technical Field

[0001] This application belongs to the technical field of distribution networks, and particularly relates to a low-voltage distribution network planning method and device based on big data analysis. Background Art

[0002] With the continuous improvement of the social science and technology level, the development of big data technology has become increasingly mature and is widely used in various industries. On the other hand, with the large-scale promotion and use of new energy, the number of power grid users is increasing day by day, the power grid structure is becoming more complex, and the planning and optimization of low-voltage distribution networks are facing more and more challenges.

[0003] Traditional distribution network planning methods often rely on simple load forecasting and fixed planning standards, ignoring factors such as the differences in electricity consumption behaviors of different regions and user types and seasonal changes, resulting in the lack of pertinence and flexibility of the planning scheme. Summary of the Invention

[0004] The embodiments of this application provide a low-voltage distribution network planning method and device based on big data analysis, which can solve one of the above-mentioned existing technical problems.

[0005] In a first aspect, the embodiments of this application provide a low-voltage distribution network planning method based on big data analysis, including:

[0006] Classify the electricity consumption behaviors of different types of users in the distribution network area, partition the distribution network area according to the area where each user type is located, and obtain multiple sub-areas with similar electricity consumption behaviors;

[0007] For each sub-area with similar electricity consumption behaviors, combine the seasonal change characteristics and holiday change characteristics of electricity consumption to generate a load characteristic pattern;

[0008] Construct a multi-dimensional load change model, and based on the load characteristic pattern, predict the load patterns of different sub-areas with similar electricity consumption behaviors through the multi-dimensional load change model to generate a comprehensive load pattern prediction result;

[0009] Based on the comprehensive load pattern prediction result, calculate the grid capacity demand of the distribution network area, and combine the grid capacity data of the distribution network area to generate a distribution network planning scheme;

[0010] Evaluate the distribution network planning scheme based on technical indicators, economic indicators, and environmental indicators.

[0011] Further, the classification of the electricity consumption behaviors of different types of users in the distribution network area, partitioning the distribution network area according to the area where each user type is located, and obtaining multiple sub-areas with similar electricity consumption behaviors includes:

[0012] Based on the power consumption behavior feature vectors of each user type, each user within the distribution network area is classified into the corresponding user type;

[0013] For the classified users, the DBSCAN algorithm is used to generate physical power supply blocks;

[0014] For each physical power supply block, based on the geographical coordinate information of each user type, it is divided into multiple sub-areas with similar power consumption behaviors through spectral clustering, forming a double-layer partition structure of power supply grid - power consumption unit.

[0015] Furthermore, for each sub-area with similar power consumption behaviors, by combining the seasonal change characteristics and holiday change characteristics of power consumption, load characteristic patterns are generated, including:

[0016] For the user load data within each sub-area with similar power consumption behaviors, the time series decomposition algorithm is used for analysis to obtain the seasonal component, and based on the seasonal component, the seasonal change characteristics are obtained;

[0017] Identify the holiday periods in the user load data within each sub-area with similar power consumption behaviors to obtain holiday load data, and extract the holiday change characteristics from the holiday load data;

[0018] Based on the seasonal change characteristics and the holiday change characteristics, the user load data within each sub-area with similar power consumption behaviors is clustered through the K-shape clustering algorithm to generate multiple load characteristic patterns, and the load characteristic patterns are used to describe the law of the user load change over time within each sub-area with similar power consumption behaviors.

[0019] Furthermore, the construction of the multi-dimensional load change model includes:

[0020] Obtain the historical load data of the time dimension, space dimension, and load type dimension, preprocess the historical load data, and decompose the preprocessed data;

[0021] For the decomposition results, extract the characteristics of the time dimension, space dimension, and load type dimension respectively to construct a multi-dimensional feature matrix;

[0022] Analyze the multi-dimensional feature matrix through the PCA algorithm to obtain multiple first load feature components;

[0023] If there is a non-linear relationship between the first load feature components extracted by the PCA algorithm, then use the KPCA algorithm to perform secondary feature extraction on the multiple first load feature components to generate second load feature components;

[0024] Build an ARIMA model, train the ARIMA model based on the second load characteristic component to generate a multi-dimensional load change model, and output the predicted result of the comprehensive load pattern for a preset time period through the multi-dimensional load change model.

[0025] Further, based on the predicted result of the comprehensive load pattern, calculate the grid capacity demand of the distribution network area, and combine the grid capacity data of the distribution network area to generate a distribution network planning scheme, including:

[0026] Obtain the predicted load quantity and the load quantity change trend of each sub-region with similar electricity consumption behaviors in the distribution network area from the predicted result of the comprehensive load pattern;

[0027] Based on the load quantity change trend, use the rolling time window algorithm to calculate the peak capacity demand of the distribution network area;

[0028] Based on the predicted load quantity, obtain the annual average load rate of the distribution network area;

[0029] Based on the peak capacity demand, the annual average load rate and the grid capacity data, calculate the capacity gap index of the distribution network area;

[0030] Based on the capacity gap index, use the NSGA-II multi-objective optimization algorithm to generate a distribution network planning scheme.

[0031] Further, the specific calculation formula for calculating the peak capacity demand of the distribution network area is as follows:

[0032] C peak =max(L pred )×1.2

[0033] Wherein, represents the comprehensive load quantity change trend of the distribution network area at the unit time granularity within the preset time, k represents the number of sub-regions with similar electricity consumption behaviors, and w i is the weight of the i-th sub-region with similar electricity consumption behaviors;

[0034] The calculation formula for the annual average load rate of the distribution network area is as follows:

[0035]

[0036] Wherein, k represents the number of sub-regions with similar electricity consumption behaviors, L q_i represents the predicted load quantity of the i-th sub-region with similar electricity consumption behaviors, and C tated represents the actual available capacity of the grid equipment in the distribution network area;

[0037] The specific calculation formula for calculating the capacity gap index of the distribution network area is as follows:

[0038]

[0039] Among them, C current represents the grid capacity data of the distribution network area, and η safe represents the safety load rate threshold, and γ represents the risk weight coefficient.

[0040] Furthermore, based on the capacity gap index, an NSGA-II multi-objective optimization algorithm is used to generate a distribution network planning scheme, including:

[0041] When the capacity gap index is greater than a preset capacity index, a multi-objective optimization model is constructed, and the NSGA-II multi-objective genetic algorithm is used to solve the Pareto optimal solution set. Among them, the optimization variables of the multi-objective optimization model include the line capacity expansion ratio, the energy storage configuration capacity, and the new energy access point layout.

[0042] Furthermore, based on technical indicators, economic indicators, and environmental indicators, the distribution network planning scheme is evaluated, including:

[0043] For the distribution network planning scheme, technical index data, economic index data, and environmental index data of the low-voltage distribution network are obtained, and each of the index data is quantified to obtain a technical index quantification result, an economic index quantification result, and an environmental index quantification result;

[0044] According to the regional characteristic data of the distribution network area, combined with a preset weight distribution rule, the weight ratio of each index data is dynamically adjusted, and the weighted average method is used to calculate and obtain a comprehensive index result.

[0045] Furthermore, the method further includes:

[0046] Extract the power consumption characteristic values from the historical power consumption data of different user types, use the Isolation Forest algorithm to perform anomaly detection on the power consumption characteristic values, and obtain the anomaly scores;

[0047] According to the anomaly scores and the load characteristic patterns, use a linear regression model to establish a normal power consumption behavior model and determine the normal value of power consumption;

[0048] Calculate the deviation value between the power consumption and the normal value of power consumption, and determine whether the deviation value exceeds a preset threshold;

[0049] If the deviation value exceeds the threshold, mark the power consumption as an abnormal value and record the abnormal power consumption behavior.

[0050] In a second aspect, an embodiment of the present application provides a low-voltage distribution network planning device based on big data analysis, including:

[0051] The first processing module: used to classify the electricity consumption behaviors of different types of users in the distribution network area, partition the distribution network area according to the area where each user type is located, and obtain multiple sub-areas with similar electricity consumption behaviors;

[0052] The second processing module: used to generate a load characteristic pattern for each sub-area with similar electricity consumption behaviors by combining the seasonal change characteristics and holiday change characteristics of electricity consumption;

[0053] The third processing module: used to construct a multi-dimensional load change model, and based on the load characteristic pattern, predict the load patterns of different sub-areas with similar electricity consumption behaviors through the multi-dimensional load change model, and generate a comprehensive load pattern prediction result;

[0054] The fourth processing module: used to calculate the grid capacity demand of the distribution network area based on the comprehensive load pattern prediction result, and combine the grid capacity data of the distribution network area to generate a distribution network planning scheme;

[0055] The fifth processing module: used to evaluate the distribution network planning scheme based on technical indicators, economic indicators, and environmental indicators.

[0056] The embodiments of the present application have at least one of the following beneficial effects compared with the prior art:

[0057] The beneficial effects of this solution are mainly reflected in the following aspects:

[0058] (1) By classifying the electricity consumption behaviors of different types of users in the distribution network area and partitioning the area accordingly, it is possible to identify sub-areas with similar electricity consumption behaviors, which helps decision-makers more accurately understand the electricity consumption needs and habits of various users, and can also provide more personalized services and management strategies for different users, improving user satisfaction.

[0059] (2) By combining the seasonal change characteristics and holiday change characteristics of electricity consumption to generate a load characteristic pattern, it is possible to more comprehensively capture the law of load change. Furthermore, by constructing a multi-dimensional load change model for load prediction, the accuracy of load prediction can be significantly improved, which is of great significance for the dispatching operation, energy management, and equipment planning of the distribution network area.

[0060] (3) The grid capacity demand calculated based on the comprehensive load pattern prediction result can more accurately reflect the future load development trend of the power grid. By combining the grid capacity data of the distribution network area, a scientific and reasonable distribution network planning scheme can be formulated to ensure the reliable power supply and efficient operation of the distribution network area.

[0061] (4) By conducting multi-dimensional evaluations on the distribution network planning scheme, it is possible to ensure that the scheme is technically feasible, economically reasonable, and environmentally friendly, which helps to promote the sustainable development of the distribution network and achieve the coordinated unity of economic, social, and environmental benefits. Description of the Drawings

[0062] To more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0063] Figure 1 is a schematic flowchart of a low-voltage distribution network planning method based on big data analysis provided by an embodiment of the present invention;

[0064] Figure 2 is a schematic structural diagram of a low-voltage distribution network planning device based on big data analysis provided by an embodiment of the present invention. Detailed Embodiments

[0065] In the following description, specific details such as specific system structures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0066] It should be understood that when used in the specification of the present application and the appended claims, the term "comprising" indicates the presence of the described features, wholes, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components, and / or their combinations.

[0067] It should also be understood that the term "and / or" used in the specification of the present application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0068] As used in the specification and appended claims of the present application, the term "if" may be construed, depending on the context, as "when", "once", "in response to determining", or "in response to detecting". Similarly, the phrases "if determined" or "if [the described condition or event] is detected" may be construed, depending on the context, to mean "once determined", "in response to determining", "once [the described condition or event] is detected", or "in response to detecting [the described condition or event]".

[0069] In addition, in the description of the specification and appended claims of the present application, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be construed as indicating or implying relative importance.

[0070] Reference to "one embodiment" or "some embodiments" or the like described in the specification of the present application means that a specific feature, structure, or characteristic described in connection with that embodiment is included in one or more embodiments of the present application. Thus, statements such as "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification do not necessarily all refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized. The terms "comprising", "including", "having", and their variants all mean "including but not limited to", unless otherwise specifically emphasized.

[0071] Please refer to Figure 1 As shown, the present invention is a low-voltage distribution network planning method based on big data analysis, including the following steps:

[0072] S100. Classify the electricity consumption behaviors of different types of users in the distribution network area, partition the distribution network area according to the area where each user type is located, and obtain multiple sub-areas with similar electricity consumption behaviors;

[0073] In this embodiment, by classifying the electricity consumption behaviors of different types of users in the distribution network area and partitioning the area accordingly, sub-areas with similar electricity consumption behaviors can be identified, which helps decision-makers more accurately understand the electricity consumption needs and habits of various users, and can also provide more personalized services and management strategies for different users, improving user satisfaction.

[0074] In some of these embodiments, the above step S100 includes:

[0075] Based on the electricity consumption behavior feature vectors of each user type, divide each user in the distribution network area into the corresponding user type;

[0076] For the classified users, use the DBSCAN algorithm to generate physical power supply blocks;

[0077] For each physical power supply block, based on the geographical coordinate information of each user type, it is divided into multiple sub-areas with similar electricity consumption behaviors through spectral clustering, forming a double-layer partition structure of power supply grid - electricity consumption unit.

[0078] In this embodiment, according to the service requirements and data characteristics, the user types to be divided are determined, such as residential users, commercial users, industrial users, etc. For each user type, typical electricity consumption behavior feature vectors of each user type are defined through historical data, expert experience or data mining techniques, and the user behavior feature vectors of each user type are constructed into a feature vector space for subsequent user classification.

[0079] In this embodiment, the time-series data of the electricity consumption of each user in the distribution network area is uniformly collected through smart meters, and the electricity consumption characteristics of each user are extracted from the time-series data of electricity consumption, such as the average value of electricity consumption, standard deviation, peak value, valley value, etc. These characteristics can reflect the electricity consumption patterns and levels of users. Based on the above electricity consumption characteristics, an electricity consumption feature vector is formed. For each user in the distribution network area, the similarity between its electricity consumption feature vector and the electricity consumption behavior feature vectors of various user types is calculated. According to the calculated similarity, each user is classified into the user type that is most similar to it. For boundary cases or users that are difficult to clearly classify, the fuzzy clustering method is used for classification processing.

[0080] In this embodiment, for the classified users, geographical density clustering is performed based on the DBSCAN algorithm to generate physical power supply blocks. In each block, the spectral clustering algorithm is used to divide the standardized electricity consumption behavior feature vectors into electricity consumption behavior sub-areas, forming a double-layer partition structure of power supply grid - electricity consumption unit. Among them, the same sub-area needs to meet the voltage deviation ≤ 5% and the Pearson correlation coefficient of the load curve > 0.8. Specifically, in the DBSCAN algorithm, the neighborhood radius is 500 meters. Specifically speaking, if a user has enough other users within 500 meters, then these users will be regarded as the same physical power supply block. The selection of the neighborhood radius depends on the actual application scenario and the distribution characteristics of the data, and is used to establish connections between geographically close users. The spectral clustering algorithm divides the electricity consumption behavior feature vectors into electricity consumption behavior sub-areas:

[0081] In this embodiment, the physical power supply blocks formed by DBSCAN geographical density clustering constitute the power supply grid layer. Within each power supply grid, the sub-regions of electricity consumption behavior divided by the spectral clustering algorithm constitute the electricity consumption unit layer. The above double-layer partition structure helps to manage the power grid and the electricity consumption behavior of users more precisely. Users within the same sub-region of electricity consumption behavior need to meet two conditions: voltage deviation ≤ 5% and Pearson correlation coefficient of load curves > 0.8. The voltage deviation refers to the deviation between the actual voltage and the rated voltage of each user within the sub-region, thereby ensuring the voltage stability within the sub-region. The Pearson correlation coefficient of load curves reflects the similarity between the load curves of each user within the sub-region, ensuring the similarity of the electricity consumption behavior of users within the sub-region.

[0082] S200. For each sub-region of similar electricity consumption behavior, combine the seasonal change characteristics and holiday change characteristics of electricity consumption to generate load characteristic patterns.

[0083] In some of these embodiments, the above step S200 includes:

[0084] Analyze the user load data within each sub-region of similar electricity consumption behavior using a time series decomposition algorithm to obtain the seasonal component, and based on the seasonal component, obtain the seasonal change characteristics.

[0085] Identify the holiday periods in the user load data within each sub-region of similar electricity consumption behavior to obtain holiday load data, and extract the holiday change characteristics from the holiday load data.

[0086] Based on the seasonal change characteristics and the holiday change characteristics, cluster the user load data within each sub-region of similar electricity consumption behavior using the K-shape clustering algorithm to generate multiple load characteristic patterns, which are used to describe the law of change of the user load over time within each sub-region of similar electricity consumption behavior.

[0087] In this embodiment, analyze the user load data within each sub-region of similar electricity consumption behavior using a time series decomposition algorithm to extract the trend component, seasonal component, and residual component. For the seasonal component, extract key seasonal characteristics from the seasonal component, such as seasonal peaks, seasonal valleys, and seasonal fluctuation amplitudes, etc. At the same time, identify and mark the holiday periods in the user load data, analyze the impact of holidays on the load, such as the decrease or increase in load during holidays, etc., and generate the holiday change characteristics.

[0088] In this embodiment, for the extracted seasonal features and holiday change features, the K-shape clustering algorithm is used to cluster the user load data in each sub-region with similar electricity consumption behaviors, generating load characteristic patterns representing different load characteristics. It can be understood that the load characteristic patterns are used to describe the law of the user load changing with time in each sub-region with similar electricity consumption behaviors, including seasonal fluctuations, holiday impacts, and long-term trends, providing data support for the planning, operation, and management of the distribution network.

[0089] In this embodiment, the application of the time series decomposition algorithm and the K-shape clustering algorithm, as well as the generation of the load characteristic patterns, are all based on in-depth analysis of historical load data to ensure the accuracy and practicality of the load characteristic patterns.

[0090] S300. Construct a multi-dimensional load change model. Based on the load characteristic patterns, use the multi-dimensional load change model to predict the load patterns of different user types, generating a comprehensive load pattern prediction result.

[0091] In this embodiment, by combining the seasonal change characteristics and holiday change characteristics of electricity consumption to generate load characteristic patterns, the law of load change can be captured more comprehensively. Furthermore, by constructing a multi-dimensional load change model for load forecasting, the accuracy of load forecasting can be significantly improved, which is of great significance for the dispatching operation, energy management, and equipment planning of the power grid.

[0092] In some of the embodiments, the constructing of the multi-dimensional load change model includes:

[0093] Obtain historical load data in the time dimension, space dimension, and load type dimension, preprocess the historical load data, and decompose the preprocessed data.

[0094] For the decomposition results, extract the features in the time dimension, space dimension, and load type dimension respectively, and construct a multi-dimensional feature matrix.

[0095] Analyze the multi-dimensional feature matrix through the PCA algorithm to obtain multiple first load feature components.

[0096] If there is a non-linear relationship between the first load feature components extracted by the PCA algorithm, then use the KPCA algorithm to perform secondary feature extraction on the multiple first load feature components to generate second load feature components.

[0097] Construct an ARIMA model, train the ARIMA model based on the second load feature components, generate a multi-dimensional load change model, and output a comprehensive load pattern prediction result for a preset time period through the multi-dimensional load change model.

[0098] In this embodiment, historical load data in the time dimension, space dimension, and load type dimension are obtained. Among them, the time dimension represents the time granularity, specifically years, months, days, hours, etc. The space dimension represents sub-areas with similar electricity consumption behaviors, and the load type dimension represents user types, such as residential electricity consumption, commercial electricity consumption, industrial electricity consumption, etc. The interpolation method is used to fill in the missing values in the obtained historical load data to ensure data integrity.

[0099] For the historical load data in the time dimension, the STL decomposition method is used for decomposition. Furthermore, the time series data is decomposed into three components: a trend term, a periodic term, and a random term. Among them, the trend term reflects the long-term change trend of the data, the periodic term reflects the periodic change of the data, and the random term is the remaining part after removing the trend and periodic components. Then, features in the time dimension are further extracted from the decomposition results, specifically including seasonal peaks, seasonal valleys, periodic fluctuation amplitudes, trend growth rates, etc.

[0100] For the historical load data in the space dimension, spatial Fourier transform is performed on the load data of all sub-areas with similar electricity consumption behaviors at the same moment to extract fundamental wave and harmonic components. Through Fourier transform, the load data is transformed from the time domain to the frequency domain, so as to analyze its frequency components. Furthermore, the spectral characteristics of the load data, such as the main frequency, harmonic distribution, etc., can be obtained, thereby understanding the operating state of the power grid.

[0101] For the historical load data in the load type dimension, a typical load curve is extracted as the reference load characteristic pattern.

[0102] In this embodiment, a multi-dimensional feature matrix is constructed. Specifically, the extracted features are represented in the form of a matrix. At the same time, the Z-score standardization method is used to standardize the feature matrix. Specifically, the data is converted into the form of a standard normal distribution with a mean of 0 and a standard deviation of 1 to eliminate the dimensional differences between different features.

[0103] In this embodiment, the standardized multi-dimensional feature matrix is used to extract the main load feature components through the PCA algorithm. Specifically, the PCA algorithm is a dimensionality reduction method used to convert high-dimensional data into low-dimensional data through linear transformation while retaining the main information of the data to generate the first load feature component. If there is a non-linear relationship between the feature components extracted by PCA, the KPCA algorithm is further used for secondary feature extraction. The data is mapped to a high-dimensional space through a kernel function, and then PCA dimensionality reduction is performed in the high-dimensional space to generate the second load feature component.

[0104] In this embodiment, an ARIMA model is constructed based on the extracted second load characteristic components. Through the ARIMA model, the autocorrelation and moving average characteristics of time series data can be captured. Specifically, by using historical data to train the parameters of the ARIMA model, including autoregressive coefficients, moving average coefficients, and differencing orders, etc., the trained ARIMA model can output the prediction results of the comprehensive load pattern in the future time period. The prediction results of the comprehensive load pattern include the predicted load quantity and the load quantity change trend in a preset time period.

[0105] In this application, the historical load data of similar sub-regions with different electricity consumption behaviors are input into the multi-dimensional load change model. Then, the multi-dimensional load change model analyzes the historical load data in the time dimension, space dimension, and load type dimension, and finally outputs the predicted load quantity and the load quantity change trend in the preset time period for the corresponding similar sub-regions of electricity consumption behaviors.

[0106] S400. Based on the prediction results of the comprehensive load pattern, calculate the grid capacity demand of the distribution network area, and combine the grid capacity data of the distribution network area to generate a distribution network planning scheme.

[0107] In this application, through accurate prediction, optimization calculation, and multi-objective optimization planning, the operation efficiency and safety of the distribution network can be significantly improved. By reasonably planning and configuring grid equipment, ensure the stable and reliable operation of the grid during peak hours and normal operation, which helps to promote the development of the low-voltage distribution network towards a more intelligent, efficient, and sustainable direction.

[0108] Furthermore, in this embodiment, the grid capacity demand calculated based on the prediction results of the comprehensive load pattern can more accurately reflect the future load development trend of the grid. Combining the grid capacity data of the distribution network area, a scientific and reasonable distribution network planning scheme can be formulated to ensure the reliable power supply and efficient operation of the grid.

[0109] In some of these embodiments, the above step S400 includes:

[0110] Obtain the predicted load quantity and the load quantity change trend of each similar sub-region of electricity consumption behavior in the distribution network area from the prediction results of the comprehensive load pattern.

[0111] Based on the load quantity change trend, use the rolling time window algorithm to calculate the peak capacity demand of the distribution network area.

[0112] Based on the predicted load quantity, obtain the annual average load rate of the distribution network area.

[0113] Based on the peak capacity demand, annual average load rate, and grid capacity data, calculate the capacity gap index of the distribution network area.

[0114] Based on the capacity gap index, the NSGA-II multi-objective optimization algorithm is used to generate a distribution network planning scheme.

[0115] In this embodiment, through the multi-dimensional load change model constructed in step S300, the predicted load and the load change trend of each sub-region with similar electricity consumption behaviors in the distribution network area can be obtained, which improves the accuracy of load forecasting and helps to implement more accurate power grid scheduling and management.

[0116] In addition, the rolling time window algorithm is used to calculate the peak capacity demand of the distribution network area, which can dynamically reflect the change trend of the power grid load, improve the accuracy and timeliness of peak capacity forecasting. At the same time, the annual average load rate of the distribution network area is calculated based on the predicted load, which reflects the utilization degree of the power grid equipment. By combining the peak capacity demand, the annual average load rate and the power grid capacity data, the capacity gap index of the distribution network area is calculated, and then the current capacity gap situation of the power grid is reflected, and it is judged whether the current distribution network area needs to carry out power grid planning. If power grid planning is required, the NSGA-II multi-objective optimization algorithm is used to generate a distribution network planning scheme. The NSGA-II algorithm can find a balance among multiple objectives and generate a distribution network planning scheme that meets the requirements of economy, reliability and greenness.

[0117] In some of these embodiments, the calculation of the peak capacity demand of the distribution network area is specifically calculated by the following formula:

[0118] C peak =max(L pred )×1.2

[0119] Wherein, represents the comprehensive load change trend of the distribution network area at the unit time granularity within the preset time, k represents the number of sub-regions with similar electricity consumption behaviors, and w i is the weight of the i-th sub-region with similar electricity consumption behaviors;

[0120] The formula for calculating the annual average load rate of the distribution network area is as follows:

[0121]

[0122] Wherein, k represents the number of sub-regions with similar electricity consumption behaviors, L q_i represents the predicted load of the i-th sub-region with similar electricity consumption behaviors, and C tated represents the actual available capacity of the power grid equipment in the distribution network area;

[0123] The specific formula for calculating the capacity gap index of the distribution network area is as follows:

[0124]

[0125] Among them, C current represents the grid capacity data of the distribution network area, η safe represents the safety load rate threshold, and γ represents the risk weight coefficient.

[0126] In this embodiment, in the calculation formula of the peak capacity demand, the coefficient of 1.2 is to reserve a 20% safety margin in the distribution network planning, so as to cover short-term fluctuations such as extreme weather and emergencies or long-term prediction deviations, and then ensure the reliable operation of the power grid under complex working conditions. The 1.2-fold coefficient equivalent to a 20% margin is an empirical value in the power system planning and will not be described in detail here.

[0127] In this embodiment, for the calculation formula of the capacity gap index, represents the peak gap, and max(0, η avg -η safe ) represents the overload risk. Thus, this capacity gap index can measure the instantaneous capacity gap and long-term operation risk, and then improve the feasibility of quantifying the capacity gap. In addition, η safe represents the safety load rate threshold. In a preferred embodiment, the specific value of this threshold can be 75%.

[0128] In some of these embodiments, based on the capacity gap index, the NSGA-II multi-objective optimization algorithm is used to generate a distribution network planning scheme, including:

[0129] When the capacity gap index is greater than the preset capacity index, a multi-objective optimization model is constructed, and the NSGA-II multi-objective genetic algorithm is used to solve the Pareto optimal solution set. Among them, the optimization variables of the multi-objective optimization model include the line capacity expansion ratio, the energy storage configuration capacity, and the new energy access point layout.

[0130] In this embodiment, through the capacity gap index, it is measured whether the low-voltage distribution network in the current distribution network area needs to be planned. It can be understood that if the capacity gap index is greater than the preset capacity index, it means that the distribution network needs to be re-planned. The specific planning methods are the line capacity expansion ratio, the energy storage configuration capacity, and the new energy access point layout. At this time, by constructing a multi-objective optimization model and using the NSGA-II multi-objective genetic algorithm to solve the Pareto optimal solution set. In a preferred embodiment, the preset capacity gap index is 10%.

[0131] S500. Evaluate the distribution network planning scheme based on technical indicators, economic indicators, and environmental indicators.

[0132] In this embodiment, by evaluating the distribution network planning scheme from multiple dimensions, it can be ensured that the scheme is technically feasible, economically reasonable, and environmentally friendly, which helps to promote the sustainable development of the distribution network and achieve the coordinated unity of economic, social, and environmental benefits.

[0133] In some of these embodiments, the above step S500 includes:

[0134] For the distribution network planning scheme, obtain the technical index data, economic index data, and environmental index data of the low-voltage distribution network, perform quantization processing on each of the index data, and obtain the technical index quantization result, economic index quantization result, and environmental index quantization result;

[0135] According to the regional characteristic data of the distribution network area, combined with the preset weight distribution rule, dynamically adjust the weight ratio of each index data, and use the weighted average method to calculate and obtain the comprehensive index result.

[0136] In this embodiment, the technical index data includes voltage qualification rate, line loss rate, and load balance degree, the economic index data includes investment efficiency, operation and maintenance cost, and asset utilization rate, and the environmental index data includes equipment floor area, electromagnetic radiation, and noise control. By quantifying the above technical index data, economic index data, and environmental index data respectively, the corresponding quantization results are obtained, and then based on the quantization results, the weighted average method is used to calculate and obtain the comprehensive index result of the distribution network planning scheme.

[0137] For the technical index data, standardize the above three index data through a preset calculation formula. For the voltage qualification rate, its specific calculation formula is: Voltage qualification rate = Qualified voltage time / Total operation time × 100% to quantify the voltage qualification rate. For the line loss rate, its specific calculation formula is: L 线损 =(Power supply amount - Electricity sales amount) / Power supply amount × 100%. For the load balance degree, its specific calculation formula is: According to the importance of power grid operation and the influence degree of each technical index on the power grid performance, different weights are assigned to the voltage qualification rate, line loss rate, and load balance degree, and then the standardized index values are multiplied by the corresponding weights and then summed to obtain the technical index quantization result.

[0138] It can be understood that the quantization processes of the economic index data and the environmental index data are the same as those of the technical index data, and will not be elaborated here.

[0139] Specifically, in the economic index data, the calculation formula for its investment efficiency is: E 投资 = Annual income increment / Initial investment × 100%, where the calculation formula for the income increment is Income = Electricity price × Increased power supply amount - Operation and maintenance cost, and the calculation formula for the operation and maintenance cost is: C 运维= ∑(labor cost + equipment maintenance + energy consumption), and the calculation formula for asset utilization rate is: U 资产 = actual load / rated capacity × 100%.

[0140] Specifically, in the environmental index data, the calculation formula for equipment floor area is: S floor area = ∑(equipment projected area × safety distance coefficient), the calculation formula for electromagnetic radiation is: R radiation = max(measured value / national standard limit) × 100%, and the quantization standard for noise control is:

[0141] In this embodiment, according to the regional characteristic data of different user types in the distribution network area, a weight allocation rule is preset. In a possible embodiment, the preset weight allocation rule is shown in the following table:

[0142]

[0143]

[0144] In this embodiment, based on the above weight allocation rule, a weighted average calculation is performed on the distribution network planning scheme to obtain a comprehensive index result. According to the quantitative results of the technical, economic, and environmental indexes of the distribution network planning scheme, a detailed evaluation report is generated to provide data support for decision-making. At the same time, the key data in the evaluation report, such as line loss rate, voltage qualification rate, investment efficiency, etc., are stored in the database for subsequent distribution network planning optimization and historical data comparison analysis.

[0145] In some of these embodiments, the low-voltage distribution network planning method based on big data analysis further includes:

[0146] Extract the power consumption characteristic values from the historical power consumption data of different user types, and use the isolation forest algorithm to perform anomaly detection on the power consumption characteristic values to obtain anomaly scores;

[0147] According to the anomaly scores and the load characteristic patterns, use a linear regression model to establish a normal power consumption behavior model and determine the normal value of power consumption;

[0148] Calculate the deviation value between the power consumption and the normal value of power consumption, and determine whether the deviation value exceeds a preset threshold;

[0149] If the deviation value exceeds the threshold, mark the power consumption as an abnormal value and record the abnormal power consumption behavior.

[0150] In this embodiment, historical power consumption data of different user types is obtained. The historical power consumption data includes key information such as timestamps, user IDs, and power consumption. The above historical power consumption data is de-duplicated, processed for missing values, and timestamp corrected. Power consumption characteristic values of different user types are extracted, such as daily power consumption, weekly power consumption, monthly power consumption, etc. The extracted power consumption characteristic values are used as input data, and an isolation forest algorithm is used to train the data to obtain an anomaly detection model. The power consumption characteristic values are input into the trained anomaly detection model, and the anomaly detection model can output the anomaly score for each power consumption characteristic value. It can be understood that this anomaly score is used to measure the normal degree of power consumption of different user types at different time granularities.

[0151] In this embodiment, the anomaly score is associated with the load characteristic pattern features to more comprehensively describe the normal and abnormal states of power consumption behavior. The seasonal change characteristics and the holiday change characteristics under the load characteristic pattern are used as input to train a linear regression model. It can be understood that the goal of the linear regression model is to learn the relationship between normal power consumption behavior and the load characteristic pattern, and predict the normal value power consumption of each user type at different time granularities. Through the prediction results of the linear regression model, the normal value power consumption is determined for each user type, and this normal value power consumption will be used to judge whether the power consumption deviates from normal subsequently.

[0152] In this embodiment, based on the normal value power consumption, the power consumption of different user types at different time granularities is obtained, compared with the normal value power consumption, and the deviation value between the two is calculated. If the deviation value exceeds the preset threshold, it means that this power consumption is an abnormal value, and the power consumption behavior at this time granularity is marked as an abnormal power consumption behavior.

[0153] In a possible embodiment, the setting of the preset threshold of the deviation amount is shown in the following table:

[0154]

[0155] In this embodiment, by identifying abnormal power consumption behaviors in the distribution network area, such as the activation of sudden high-power equipment, seasonal production cycles, etc., and synchronizing the abnormal power consumption behaviors into a multi-dimensional load change model, comprehensive consideration can be carried out when predicting the load patterns of different user types, and through refined load characteristic analysis, accurate planning of the distribution network can be achieved.

[0156] Please refer to Figure 2 As shown, the present invention also provides a low-voltage distribution network planning device based on big data analysis. The device includes:

[0157] The first processing module 201: It is used to classify the electricity consumption behaviors of different types of users in the distribution network area, partition the distribution network area for each area where the user type is located, and obtain multiple sub-areas with similar electricity consumption behaviors.

[0158] The second processing module 202: For each sub-area with similar electricity consumption behaviors, it is used to generate a load characteristic pattern by combining the seasonal change characteristics and holiday change characteristics of the electricity consumption.

[0159] The third processing module 203: It is used to construct a multi-dimensional load change model, and based on the load characteristic pattern, predict the load patterns of different sub-areas with similar electricity consumption behaviors through the multi-dimensional load change model, and generate a comprehensive load pattern prediction result.

[0160] The fourth processing module 204: It is used to calculate the grid capacity demand of the distribution network area based on the comprehensive load pattern prediction result, and combine the grid capacity data of the distribution network area to generate a distribution network planning scheme.

[0161] The fifth processing module 205: It is used to evaluate the distribution network planning scheme based on technical indicators, economic indicators, and environmental indicators.

[0162] It can be understood that the content in the embodiment of the low-voltage distribution network planning method based on big data analysis as Figure 1 shown is applicable to the embodiment of this low-voltage distribution network planning device based on big data analysis. The functions specifically implemented in the embodiment of this low-voltage distribution network planning device based on big data analysis are the same as those in the embodiment of the low-voltage distribution network planning method based on big data analysis as Figure 1 shown, and the beneficial effects achieved are also the same as those in the embodiment of the low-voltage distribution network planning method based on big data analysis as Figure 1 shown.

[0163] It should be noted that for the information interaction, execution process, etc. between the above systems, because they are based on the same concept as the method embodiment of the present invention, for their specific functions and the technical effects brought, please refer to the method embodiment part for details, and will not be elaborated here.

[0164] Those skilled in the art can clearly understand that, for the convenience and conciseness of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the system is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiments can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working processes of the units and modules in the above system can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0165] The above-described embodiments are only used to illustrate the technical solutions of this application and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of each embodiment of this application and should all be included in the protection scope of this application.

Claims

1. A low voltage distribution network planning method based on big data analysis, characterized in that: include: Classifying the electricity consumption behaviors of different types of users in the distribution network area, partitioning the distribution network area according to the area where each user type is located, and obtaining multiple sub-areas with similar electricity consumption behaviors; For each sub-area with similar electricity consumption behavior, the load characteristic pattern is generated by combining the seasonal variation characteristics and holiday variation characteristics of electricity consumption; Constructing a multi-dimensional load change model, based on the load characteristic pattern, predicting the load pattern of sub-areas with similar different power consumption behaviors through the multi-dimensional load change model, and generating a comprehensive load pattern prediction result; Based on the comprehensive load pattern prediction results, the power grid capacity demand of the distribution network area is calculated, and the distribution network planning scheme is generated in combination with the power grid capacity data of the distribution network area; The distribution network planning scheme is evaluated based on technical indicators, economic indicators and environmental indicators.

2. The method according to claim 1, characterized in that The power consumption behaviors of different types of users in the distribution network area are classified, and the distribution network area is partitioned according to the area where each type of user is located to obtain multiple sub-areas with similar power consumption behaviors, including: Based on the electricity consumption behavior feature vector of each user type, each user in the distribution network area is divided into corresponding user types; For the classified users, the DBSCAN algorithm is used to generate physical power supply blocks; For each physical power supply block, based on the geographic coordinate information of each user type, it is divided into multiple sub-areas with similar power consumption behaviors through spectral clustering, forming a two-layer partition structure of power supply grid-power consumption unit.

3. The method according to claim 2, characterized in that For each sub-area with similar electricity consumption behavior, a load characteristic pattern is generated by combining the seasonal variation characteristics and holiday variation characteristics of electricity consumption, including: The user load data in each sub-area with similar power consumption behavior is analyzed by using a time series decomposition algorithm to obtain seasonal components, and seasonal variation characteristics are obtained based on the seasonal components; Identify the holiday period in the user load data in each sub-area with similar electricity consumption behavior, obtain the holiday load data, and extract the holiday change characteristics from the holiday load data; Based on the seasonal variation characteristics and the holiday variation characteristics, the user load data in each sub-area with similar electricity consumption behavior is clustered by a K-shape clustering algorithm to generate a plurality of load characteristic patterns, and the load characteristic patterns are used to describe the law of change of user load over time in each sub-area with similar electricity consumption behavior.

4. The method according to claim 1, characterized in that The multi-dimensional load change model is constructed, including: Acquire historical load data in time dimension, space dimension and load type dimension, preprocess the historical load data, and decompose the preprocessed data; Based on the decomposition results, the features of time dimension, space dimension and load type dimension are extracted respectively to construct a multi-dimensional feature matrix; Analyze the multi-dimensional characteristic matrix by using a PCA algorithm to obtain a plurality of first load characteristic components; If there is a nonlinear relationship between the first load characteristic components extracted by the PCA algorithm, a KPCA algorithm is used to perform secondary feature extraction on a plurality of the first load characteristic components to generate a second load characteristic component; An ARIMA model is constructed, and based on the second load characteristic component, the ARIMA model is trained to generate a multi-dimensional load change model, and a comprehensive load pattern prediction result for a preset time period is output through the multi-dimensional load change model.

5. The method according to claim 1, characterized in that The method of calculating the grid capacity demand of the distribution network area based on the comprehensive load pattern prediction result and generating a distribution network planning scheme in combination with the grid capacity data of the distribution network area includes: From the comprehensive load pattern prediction results, the predicted load and load change trend of each sub-area with similar power consumption behavior in the distribution network area are obtained; Based on the load variation trend, a rolling time window algorithm is used to calculate the peak capacity demand of the distribution network area; Based on the predicted load, obtaining an annual average load rate of the distribution network area; Based on the peak capacity demand, the average annual load rate and the grid capacity data, a capacity gap index of the distribution network area is calculated; Based on the capacity gap index, the NSGA-II multi-objective optimization algorithm is used to generate a distribution network planning scheme.

6. The method according to claim 5, characterized in that The specific calculation formula for calculating the peak capacity demand of the distribution network area is as follows: C peak =max(L pred )×1.2 in, represents the change trend of the comprehensive load in the distribution network area at the unit time granularity within the preset time, k represents the number of sub-areas with similar power consumption behaviors, and w i is the weight of the ith sub-area with similar electricity consumption behavior; The calculation formula for the annual average load rate of the distribution network area is as follows: Where k represents the number of sub-areas with similar electricity consumption behaviors, L q_i represents the predicted load of the ith sub-area with similar electricity consumption behavior, C tated Indicates the actual available capacity of the grid equipment in the distribution network area; The capacity gap index of the distribution network area is obtained by calculation, and the specific calculation formula is as follows: Among them, C current Represents the grid capacity data of the distribution network area, η safe represents the safety load rate threshold, and γ represents the risk weight coefficient.

7. The method according to claim 6, characterized in that The method of generating a distribution network planning scheme based on the capacity gap index using the NSGA-II multi-objective optimization algorithm includes: When the capacity gap index is greater than the preset capacity index, a multi-objective optimization model is constructed, and the NSGA-II multi-objective genetic algorithm is used to solve the Pareto optimal solution set, wherein the optimization variables of the multi-objective optimization model include line expansion ratio, energy storage configuration capacity and new energy access point layout.

8. The method according to claim 1, characterized in that The distribution network planning scheme is evaluated based on technical indicators, economic indicators and environmental indicators, including: According to the distribution network planning scheme, the technical indicator data, economic indicator data and environmental indicator data of the low-voltage distribution network are obtained, and the data of each indicator is quantified to obtain the quantified results of the technical indicators, the quantified results of the economic indicators and the quantified results of the environmental indicators; According to the regional characteristic data of the distribution network area, combined with the preset weight allocation rules, the weight ratio of each indicator data is dynamically adjusted, and the weighted average method is used to calculate and obtain the comprehensive indicator result.

9. The method according to claim 1, characterized in that The method further comprises: Extract electricity consumption feature values ​​from historical electricity consumption data of different user types, use the isolation forest algorithm to detect anomalies in the electricity consumption feature values, and obtain anomaly scores; According to the abnormal score and the load characteristic pattern, a normal power consumption behavior model is established using a linear regression model to determine a normal value of power consumption; Calculate the deviation between the power consumption and the normal power consumption, and determine whether the deviation exceeds a preset threshold; If the deviation value exceeds the threshold, the power consumption is marked as an abnormal value and the abnormal power consumption behavior is recorded.

10. A low voltage distribution network planning device based on big data analysis, characterized in that: The first processing module is used to classify the electricity consumption behaviors of different types of users in the distribution network area, and partition the distribution network area according to the area where each user type is located to obtain multiple sub-areas with similar electricity consumption behaviors; The second processing module is used to generate a load characteristic pattern for each sub-area with similar electricity consumption behavior, combining the seasonal variation characteristics and holiday variation characteristics of electricity consumption; The third processing module is used to construct a multi-dimensional load change model, based on the load characteristic mode, predict the load mode of sub-areas with similar power consumption behaviors with the multi-dimensional load change model, and generate a comprehensive load mode prediction result; The fourth processing module is used to calculate the power grid capacity demand of the distribution network area based on the comprehensive load pattern prediction result, and generate a distribution network planning scheme in combination with the power grid capacity data of the distribution network area; The fifth processing module is used to evaluate the distribution network planning scheme based on technical indicators, economic indicators and environmental indicators.

Images