A method and system for analyzing electricity consumption preference of low-voltage distributed photovoltaic users

Abstract

The present invention discloses a method and system for analyzing the electricity preference of low-voltage distributed photovoltaic users. The method of the present invention includes: obtaining historical data of low-voltage distributed photovoltaic users and performing data preprocessing; analyzing the electricity preference characteristics of low-voltage distributed photovoltaic users from three perspectives of user photovoltaic power generation status, electricity price coefficient and load selection inertia according to the preprocessed data set, and obtaining an electricity priority evaluation index; clustering the output curves of low-voltage distributed photovoltaic users based on an improved DTW algorithm to obtain a clustering result set; combining the electricity priority evaluation index and the clustering result set, constructing a quantum weighted neuron suitable for analyzing the electricity preference of photovoltaic users, and on this basis, establishing a photovoltaic user electricity preference analysis model based on the improved quantum weighted neuron, analyzing the preference degree of energy economy, energy quality assurance and energy safety, and thus obtaining the electricity preference of low-voltage distributed photovoltaic users.

Description

Technical Field

[0001] The present invention belongs to the technical field of distributed photovoltaics, and in particular to a method and system for analyzing electricity consumption preferences of low-voltage distributed photovoltaic users. Background Art

[0002] At present, there is a lack of a unified standard system for analyzing the electricity preferences of photovoltaic users, and the electricity preferences and business architectures of various users are complicated. Therefore, there are certain difficulties in sorting out the types of user preferences and exact needs, and it is difficult to accurately push data and interactive services based on user characteristics and preferences. As far as the construction of low-voltage distributed photovoltaic interactive services is concerned, it is necessary to further explore the technical path for analyzing the characteristics of photovoltaic users' electricity preferences. In order to promote two-way interaction between the power grid and distributed photovoltaic users, enhance the public's awareness of participation, and realize intelligent energy management, it is necessary to deeply analyze and study users' electricity preferences.

[0003] The existing analysis of user electricity preferences generally uses time series clustering analysis algorithms. The clustering methods for photovoltaic output and electricity load curves are mainly divided into direct clustering methods and indirect clustering methods. The direct clustering method refers to clustering the electricity load curve directly by modifying the metric distance, but the algorithm complexity is high and the calculation time is long when processing high-dimensional time series. The indirect clustering method is to construct user electricity characteristics and then use traditional clustering algorithms to cluster feature indicators. Although it reduces the calculation complexity, it ignores the rich information contained in the time series. In addition, there are currently few methods for feature extraction and analysis of photovoltaic user electricity preferences.

[0004] The intelligent clustering algorithm combines deep learning technology. Although it can quickly extract features, it is overly dependent on computing power and is not as accurate as traditional clustering algorithms. Summary of the invention

[0005] In response to the problems existing in the above-mentioned prior art, the present invention proposes a method and system for analyzing electricity usage preferences of low-voltage distributed photovoltaic users. The method adopts a photovoltaic user electricity usage preference analysis model to analyze the preference levels of energy economy, energy guarantee and energy safety, so as to obtain the electricity usage preferences of low-voltage distributed photovoltaic users. The method can accurately optimize power dispatching, reasonably allocate energy resources, reduce system load peak, and thus reduce energy waste and cost expenditure.

[0006] To this end, the present invention adopts the following technical solution.

[0007] In a first aspect, the present invention provides a method for analyzing electricity consumption preferences of low-voltage distributed photovoltaic users, the steps of which are as follows:

[0008] Step 1, obtaining historical data of low-voltage distributed photovoltaic users, and performing data preprocessing to obtain a preprocessed data set;

[0009] Step 2: For the preprocessed data set, the electricity preference characteristics of low-voltage distributed photovoltaic users are analyzed from three perspectives: the user's photovoltaic power generation status, the electricity price coefficient, and the load selection inertia, and the electricity priority evaluation index is obtained;

[0010] Step 3: cluster the output curves of low-voltage distributed photovoltaic users based on the improved DTW algorithm, divide the users with similar output conditions into the same cluster, and obtain a clustering result set;

[0011] Step 4: Combine the electricity priority evaluation index obtained in step 2 and the clustering result set obtained by clustering in step 3 to construct a quantum weighted neuron suitable for photovoltaic users' electricity preference analysis. On this basis, a photovoltaic users' electricity preference analysis model based on improved quantum weighted neurons is established to analyze the preference for energy economy, energy quality assurance and energy safety, thereby obtaining the electricity preference of low-voltage distributed photovoltaic users.

[0012] Furthermore, in step 1, the historical data includes historical photovoltaic data and load data, specifically photovoltaic user load power consumption attributes, electricity price coefficient, generated power, online power, power consumption, daily power consumption, weather type, season, temperature and humidity.

[0013] Furthermore, in step 1, data preprocessing is performed in sequence according to the steps of data cleaning, noise reduction, data integration and data standardization to obtain a preprocessed data set.

[0014] Furthermore, in step 2, the calculation formula of the power consumption priority evaluation index is as follows:

[0015] ,

[0016] ,

[0017] ,

[0018] ,

[0019] In the formula, express Photovoltaic users The photovoltaic power generation status; express Photovoltaic users The actual value of power generation; express Photovoltaic users The predicted value of power generation; express The electricity price coefficient of photovoltaic users at the moment; express Photovoltaic users Real-time electricity prices; express Photovoltaic users The power consumption; Indicates time Photovoltaic users The historical average of electricity costs; express Photovoltaic users Load selection inertia; N represents the number of all PV users participating in demand response; express Photovoltaic users The electricity consumption willingness coefficient is determined by the photovoltaic users. of The historical electricity consumption behavior at each moment is predicted; express Photovoltaic users The power consumption priority evaluation index is as follows: the larger the index value, the greater the probability of the user using electricity; , , are all weight coefficients.

[0020] Furthermore, in step 3, a distance matrix between each photovoltaic output curve is constructed, a graph spectrum algorithm is introduced to search for matching points, the similarity between different photovoltaic output curves is calculated and a similarity matrix is ​​established, so as to divide clusters of different types to achieve clustering.

[0021] Furthermore, the step 3 is specifically as follows:

[0022] First, given two photovoltaic output curves and , construct the distance matrix between two photovoltaic output curves , and There are two photovoltaic output curves. i Active power value of the data point; n Indicates the number of data points of the photovoltaic output curve;

[0023] Then, the matching point search is performed through the graph algorithm, and the low-voltage distributed photovoltaic user characteristics are taken as entities, and the relationship between entities is taken as edges. The user characteristics are weather type, electricity price coefficient, daily electricity consumption and power generation. The structured user information is extracted for eigenvalues, relationships and entities, and the "eigenvalue-relationship-entity" triple is generated using the knowledge graph; and according to the boundary conditions, continuity and monotonicity constraints, the distance matrix is ​​combined Constructing a collection of curved paths ,in For the first The position coordinates of the matching points, is the number of elements contained in the dynamic bending path;

[0024] Calculate the cumulative distance of the dynamic curved path that meets the above constraints, and take its minimum value as the distance between the two photovoltaic output curves. distance:

[0025] ,

[0026] In the formula, Indicates the difference between two photovoltaic output curves Distance, which is used as an indicator to measure the similarity between any two photovoltaic output curves; The smaller the distance, the more similar the two PV output curves are. The higher;

[0027] Constructing a similarity matrix , and calculate the adjacency matrix based on the similarity matrix With degree matrix ;

[0028] ,

[0029] ,

[0030] ,

[0031] ,

[0032] In the formula, Indicates similarity; represents the non-negative weights of the adjacency matrix; Represents the degree of any sample point;

[0033] Compute the Laplacian matrix , and perform standardization; calculate the matrix The eigenvectors corresponding to each eigenvalue form a characteristic matrix ; Each row in is regarded as a sample point. Perform clustering and obtain a set of clustering results .

[0034] Furthermore, the silhouette coefficient and Davidson-Botting index were used to evaluate the clustering results.

[0035] Furthermore, in step 4, the quantum weighted neuron is:

[0036] ,

[0037] In the formula, The input is the clustering result set H of the photovoltaic user similarity division; is the activity value quantum state, is its phase; for The phase of is the aggregation operator; For the i weighted quantum states, is its phase; for m The quantum state vector of the weight value;

[0038] The photovoltaic user electricity preference analysis model based on improved quantum weighted neurons is:

[0039] ,

[0040] In the formula, For photovoltaic users i Output of energy usage scenario preference, ; For the A weight quantum state vector; ; ;

[0041] According to the above clustering result set based on the similarity division of distributed low-voltage photovoltaic users H As input to the user electricity preference analysis model , output variable The user's preference for energy economy, energy security and energy safety is obtained, thereby obtaining the electricity preference of distributed low-voltage photovoltaic users.

[0042] Furthermore, the architecture of the user electricity preference analysis model based on the improved quantum weighted neuron consists of an input layer, a hidden layer, a Dropout layer, a fully connected layer, and an output layer;

[0043] The input layer is the clustering result set after the distributed photovoltaic user load data is clustered. The hidden layer is a single hidden layer, and the Dropout value of the Dropout layer is set to 0.25. After the Dropout layer, a fully connected layer is passed to get the output. The mean square error is selected as the loss function, and the Adam algorithm is used to optimize the model parameters. The mean square error calculation formula is as follows:

[0044] ,

[0045] The Adam algorithm replaces the gradient with the exponential moving average of the gradient at each moment, that is, the first-order moment estimate of the gradient, and its calculation formula is as follows:

[0046] ,

[0047] In the formula, , Respectively represent the first-order moment estimates before and after optimization; is an exponentially weighted average; is the gradient of the weight to be optimized;

[0048] ,

[0049] , They are the second-order moment estimates before and after optimization respectively; is an exponentially weighted average; is the square of the gradient of the weight to be optimized;

[0050] Combine the first-order and second-order momentum together to update the weights. The update expression is as follows:

[0051] ,

[0052] , are the weights before and after the update respectively; is the learning rate; is a constant.

[0053] In a second aspect, the present invention provides a low-voltage distributed photovoltaic user electricity preference analysis system, which is used to implement the low-voltage distributed photovoltaic user electricity preference analysis method, and includes:

[0054] Data acquisition and processing module: obtain historical data of low-voltage distributed photovoltaic users and perform data preprocessing;

[0055] Electricity priority evaluation index acquisition module: for the pre-processed data, the electricity preference characteristics of low-voltage distributed photovoltaic users are analyzed from three perspectives: the user's photovoltaic power generation status, the electricity price coefficient, and the load selection inertia, and the electricity priority evaluation index is obtained;

[0056] Clustering result set acquisition module: cluster the output curves of low-voltage distributed photovoltaic users based on the improved DTW algorithm, divide the users with similar output conditions into the same cluster, and obtain the clustering result set;

[0057] Photovoltaic user electricity preference analysis model establishment module: Combined with the electricity priority evaluation index obtained in the electricity priority evaluation index acquisition module and the clustering result set obtained by clustering in the clustering result set acquisition module, a quantum weighted neuron suitable for photovoltaic user electricity preference analysis is constructed. On this basis, a photovoltaic user electricity preference analysis model based on improved quantum weighted neurons is established to analyze the preference degree of energy economy, energy quality assurance and energy safety, thereby obtaining the electricity preference of low-voltage distributed photovoltaic users.

[0058] The beneficial effects of the present invention are as follows: the present invention divides distributed photovoltaic users with the maximum similarity output curves into the same cluster, introduces a graph algorithm to search for matching points, calculates the similarity index between output curves, and effectively reduces the division error between clusters. The present invention establishes a user power preference analysis model based on improved quantum weighted neurons, which can better capture the nonlinear relationship between photovoltaic user clusters; introduces quantum computing theory, improves the distinction between cluster features, and highlights the power preferences of distributed low-voltage photovoltaic users, so as to accurately optimize power dispatching, reasonably allocate energy resources, reduce system load peaks, and thus reduce energy waste and cost expenditures. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0060] Figure 1 It is a flow chart of the method for analyzing the electricity consumption preference of low-voltage distributed photovoltaic users of the present invention;

[0061] Figure 2 This is an architecture diagram of the low-voltage distributed photovoltaic user electricity preference analysis system of the present invention. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0063] Example 1

[0064] This embodiment provides a method for analyzing the electricity consumption preference of low-voltage distributed photovoltaic users. Figure 1 As shown, the steps are as follows:

[0065] Step 1: Obtain historical data of low-voltage distributed photovoltaic users (including historical photovoltaic data and load data), and perform data preprocessing in sequence according to the steps of data cleaning, noise reduction, data integration and data standardization to obtain a preprocessed data set.

[0066] The specific contents of step 1 are as follows:

[0067] Obtain historical photovoltaic, load, environment and meteorological data. The data set includes photovoltaic user load power consumption attributes, electricity price coefficient, power generation, online power, power consumption, daily power consumption, weather type, season, temperature, humidity and other data for one year, with an interval of 15 minutes. And perform data preprocessing according to the following steps.

[0068] (1) Data cleaning: The main function and purpose of data cleaning is to deal with anomalies in the data. Missing values ​​can usually be filled, deleted or interpolated. The detection and processing of outliers can be done with the help of statistical methods, visualization tools or professional domain knowledge. Delete data that deviates greatly from the normal value. For data with a large number of missing values, delete them directly.

[0069] (2) Noise reduction: When there are outliers, noise interference or other abnormal data in the data, if the proportion of these abnormal data exceeds a certain threshold, it will significantly affect the accuracy and reliability of the analysis results. Removing noise or unnecessary information in the data can improve data quality and accuracy. Common noise reduction measures include: noise reduction in small blocks, detection of isolated discrete points, etc.

[0070] (3) Data integration: Integrate data from different data sources into a unified data set for subsequent analysis. Solve the inconsistency of format, unit or naming between different data sources. Data integration mainly includes data consolidation and data matching.

[0071] (4) Data standardization: Eliminate the impact of different scales or dimensions, normalize the data, ensure that different data have similar scales, make the model easier to converge, and improve the performance and stability of the model.

[0072] Step 2: The electricity consumption behavior of low-voltage distributed photovoltaic users will change with factors such as weather, temperature, electricity price and production arrangement, forming different user electricity consumption behaviors and demand preferences. In order to coordinate the uncertainty of user behavior and demand planning, and at the same time ensure the comfort of users in the demand response process, the electricity preference characteristics of low-voltage distributed photovoltaic users are analyzed from three perspectives: user photovoltaic power generation status, electricity price coefficient and load selection inertia, and the electricity priority evaluation index is obtained. By understanding the user's electricity consumption behavior and demand preferences, resources can be reasonably allocated, resource utilization efficiency can be improved, costs can be reduced, and competitiveness can be enhanced.

[0073] The calculation formula of the power consumption priority evaluation index is as follows:

[0074] ,

[0075] ,

[0076] ,

[0077] ,

[0078] In the formula, express Photovoltaic users The photovoltaic power generation status; express Photovoltaic users The actual value of power generation; express Photovoltaic users The predicted value of power generation; express The electricity price coefficient of photovoltaic users at the moment; express Photovoltaic users Real-time electricity prices; express Photovoltaic users The power consumption; Indicates time Photovoltaic users The historical average of electricity costs; express Photovoltaic users Load selection inertia; N represents the number of all PV users participating in demand response; express Photovoltaic users The electricity consumption willingness coefficient is determined by the photovoltaic users. of The historical electricity consumption behavior at each moment is predicted; express Photovoltaic users The power consumption priority evaluation index is as follows: the larger the index value, the greater the probability of the user using electricity; , , are all weight coefficients.

[0079] With the changes in factors such as working days / non-working days, day / night, season, temperature, and real-time electricity prices, the characteristics of different types of photovoltaic users are also different.

[0080] Step 3: Cluster the output curves of low-voltage distributed photovoltaic users based on the improved DTW algorithm, divide the output conditions of users into the same cluster, and obtain a set of clustering results. Specifically, construct a distance matrix between each photovoltaic output curve, introduce a graph algorithm to search for matching points, calculate the similarity between different photovoltaic output curves, and establish a similarity matrix, so as to divide clusters of different types and achieve clustering.

[0081] The specific contents of step 3 are as follows:

[0082] First, given two photovoltaic output curves and , construct the distance matrix between two photovoltaic output curves , and There are two photovoltaic output curves. i Active power value of the data point; n Indicates the number of data points of the photovoltaic output curve;

[0083] Then, the matching point search is performed through the graph algorithm, and the low-voltage distributed photovoltaic user characteristics are taken as entities, and the relationship between entities is taken as edges. The user characteristics are weather type, electricity price coefficient, daily electricity consumption and power generation. The structured user information is extracted for eigenvalues, relationships and entities, and the "eigenvalue-relationship-entity" triple is generated using the knowledge graph; and according to the boundary conditions, continuity and monotonicity constraints, the distance matrix is ​​combined Constructing a collection of curved paths ,in For the first The position coordinates of the matching points, is the number of elements contained in the dynamic bending path;

[0084] Calculate the cumulative distance of the dynamic curved path that meets the above constraints, and take its minimum value as the distance between the two photovoltaic output curves. distance:

[0085] ,

[0086] In the formula, Indicates the difference between two photovoltaic output curves Distance, which is used as an indicator to measure the similarity between any two photovoltaic output curves; The smaller the distance, the more similar the two PV output curves are. The higher;

[0087] Constructing a similarity matrix , and calculate the adjacency matrix based on the similarity matrix With degree matrix ;

[0088] ,

[0089] ,

[0090] ,

[0091] ,

[0092] In the formula, represents the non-negative weights of the adjacency matrix; Represents the degree of any sample point;

[0093] Compute the Laplacian matrix , and perform standardization; calculate the matrix The eigenvectors corresponding to each eigenvalue form a characteristic matrix ; Each row in is regarded as a sample point. Perform clustering and obtain a set of clustering results .

[0094] The silhouette coefficient and Davidson-Botting index are used to evaluate the clustering results. The specific calculation formula is as follows:

[0095] ,

[0096] In the formula, For the The silhouette coefficient of a vector; For the The average distance from a vector to other vectors in its class; For the The average distance of a vector to other vectors that do not belong to its class.

[0097] ,

[0098] In the formula, is the Davidson-Botting Index; , They are the average Euclidean distances from each type of sample to the class center; For the Class and The Euclidean distance of the class center of the class sample.

[0099] Step 4: Combine the electricity priority evaluation index obtained in step 2 and the clustering result set obtained by clustering in step 3 to construct a quantum weighted neuron suitable for photovoltaic users' electricity preference analysis. On this basis, a photovoltaic users' electricity preference analysis model based on improved quantum weighted neurons is established to analyze the preference for energy economy, energy quality assurance and energy safety, thereby obtaining the electricity preference of low-voltage distributed photovoltaic users.

[0100] In step 4, the quantum weighted neuron is:

[0101] ,

[0102] In the formula, The input is the clustering result set H of the photovoltaic user similarity division; is the activity value quantum state, is its phase; for The phase of is the aggregation operator; For the i weighted quantum states, is its phase; for m The quantum state vector of weight values.

[0103] The photovoltaic user electricity preference analysis model based on improved quantum weighted neurons is:

[0104] ,

[0105] In the formula, For photovoltaic users i Output of energy usage scenario preference, ; For the A weight quantum state vector; ; ;

[0106] The architecture of the user electricity preference analysis model based on improved quantum weighted neurons consists of an input layer, a hidden layer, a dropout layer, a fully connected layer, and an output layer;

[0107] The input layer is the clustering result set after the distributed photovoltaic user load data is clustered. The hidden layer is a single hidden layer, and the Dropout value of the Dropout layer is set to 0.25. After the Dropout layer, a fully connected layer is passed to get the output. The mean square error is selected as the loss function, and the Adam algorithm is used to optimize the model parameters. The mean square error calculation formula is as follows:

[0108] ,

[0109] The Adam algorithm replaces the gradient with the exponential moving average of the gradient at each moment, that is, the first-order moment estimate of the gradient, and its calculation formula is as follows:

[0110] ,

[0111] In the formula, , Respectively represent the first-order moment estimates before and after optimization; is an exponentially weighted average; is the gradient of the weight to be optimized;

[0112] ,

[0113] , They are the second-order moment estimates before and after optimization respectively; is an exponentially weighted average; is the square of the gradient of the weight to be optimized;

[0114] Combine the first-order and second-order momentum together to update the weights. The update expression is as follows:

[0115] ,

[0116] , are the weights before and after the update respectively; is the learning rate; is a constant.

[0117] According to the above clustering result set based on the similarity division of distributed low-voltage photovoltaic users H As input to the user electricity preference analysis model , output variable The user's preference for energy economy, energy security and energy safety is obtained, thereby obtaining the electricity preference of distributed low-voltage photovoltaic users.

[0118] The present invention collects the power generation and consumption information of distributed photovoltaic users, analyzes the personalized and refined power consumption behavior and power demand preferences of different types of users, and constructs a low-voltage distributed photovoltaic user power consumption preference model. It can deeply analyze the power consumption behavior of photovoltaic users in different time periods and scenarios, and explore the energy consumption patterns and preferences of photovoltaic users. The model can identify the power consumption patterns during peak and trough periods, and predict the energy demand trends of users.

[0119] By conducting an in-depth study on the electricity usage preferences of photovoltaic users, the present invention can accurately optimize power dispatch, reasonably allocate energy resources, reduce system load peaks, and thus reduce energy waste and cost expenditures. Analyzing the electricity usage preferences of photovoltaic users can not only provide customized energy solutions for individual users to meet their specific needs and improve user experience and satisfaction, but also help optimize the overall energy system.

[0120] With the help of a typical user electricity preference model, the present invention can guide users to use energy rationally and promote a green and low-carbon lifestyle; it helps to improve the overall performance and stability of the photovoltaic system, and is also conducive to promoting the energy industry to move towards a more intelligent, efficient and sustainable development direction.

[0121] Example 2

[0122] This embodiment provides a low-voltage distributed photovoltaic user power preference analysis system. Figure 2 As shown, the method for implementing the method described in Example 1 includes:

[0123] Data acquisition and processing module: obtain historical data of low-voltage distributed photovoltaic users and perform data preprocessing;

[0124] Electricity priority evaluation index acquisition module: for the pre-processed data, the electricity preference characteristics of low-voltage distributed photovoltaic users are analyzed from three perspectives: the user's photovoltaic power generation status, the electricity price coefficient, and the load selection inertia, and the electricity priority evaluation index is obtained;

[0125] Clustering result set acquisition module: cluster the output curves of low-voltage distributed photovoltaic users based on the improved DTW algorithm, divide the users with similar output conditions into the same cluster, and obtain the clustering result set;

[0126] Photovoltaic user electricity preference analysis model establishment module: Combined with the electricity priority evaluation index obtained in the electricity priority evaluation index acquisition module and the clustering result set obtained by clustering in the clustering result set acquisition module, a quantum weighted neuron suitable for photovoltaic user electricity preference analysis is constructed. On this basis, a photovoltaic user electricity preference analysis model based on improved quantum weighted neurons is established to analyze the preference degree of energy economy, energy quality assurance and energy safety, thereby obtaining the electricity preference of low-voltage distributed photovoltaic users.

[0127] It should be noted that each module in the above-mentioned low-voltage distributed photovoltaic user electricity preference analysis system can be fully or partially implemented by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware form, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above modules. For the specific definition of a low-voltage distributed photovoltaic user electricity preference analysis system, please refer to the definition of a low-voltage distributed photovoltaic user electricity preference analysis method above. The two have the same functions and effects, which will not be repeated here.

[0128] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A method for analyzing electricity consumption preferences of low-voltage distributed photovoltaic users, characterized in that: include: Step 1, obtaining historical data of low-voltage distributed photovoltaic users, and performing data preprocessing to obtain a preprocessed data set; Step 2: For the preprocessed data set, the electricity preference characteristics of low-voltage distributed photovoltaic users are analyzed from three perspectives: the user's photovoltaic power generation status, the electricity price coefficient, and the load selection inertia, and the electricity priority evaluation index is obtained; In step 2, the calculation formula of the power consumption priority evaluation index is as follows: ， ， ， ， In the formula, express Photovoltaic users The photovoltaic power generation status; express Photovoltaic users The actual value of power generation; express Photovoltaic users The predicted value of power generation; express The electricity price coefficient of photovoltaic users at the moment; express Photovoltaic users Real-time electricity prices; express Photovoltaic users The power consumption; Indicates time Photovoltaic users The historical average of electricity costs; express Photovoltaic users Load selection inertia; N represents the number of all PV users participating in demand response; express Photovoltaic users The electricity consumption willingness coefficient is determined by the photovoltaic users. of The historical electricity consumption behavior at each moment is predicted; express Photovoltaic users The power consumption priority evaluation index is as follows: the larger the index value, the greater the probability of the user using electricity; , , All are weight coefficients; Step 3: cluster the output curves of low-voltage distributed photovoltaic users based on the improved DTW algorithm, divide the users with similar output conditions into the same cluster, and obtain a clustering result set; The step 3 is to construct a distance matrix between each photovoltaic output curve, introduce a graph algorithm to search for matching points, calculate the similarity between different photovoltaic output curves and establish a similarity matrix, so as to divide clusters of different types to achieve clustering; Step 4: Combine the electricity priority evaluation index obtained in step 2 and the clustering result set obtained by clustering in step 3 to construct a quantum weighted neuron suitable for analyzing the electricity preference of photovoltaic users. On this basis, a photovoltaic user electricity preference analysis model based on improved quantum weighted neurons is established to analyze the preference degree of energy economy, energy quality assurance and energy safety, so as to obtain the electricity preference of low-voltage distributed photovoltaic users. In step 4, the quantum weighted neuron is: ， In the formula, The input is the clustering result set H of the photovoltaic user similarity division; is the activity value quantum state, is its phase; for The phase of is the aggregation operator; For the i weighted quantum states, is its phase; for m The quantum state vector of the weight value; The photovoltaic user electricity preference analysis model based on improved quantum weighted neurons is: ， In the formula, For photovoltaic users i Output of energy usage scenario preference, ; For the A weight quantum state vector; ; ; According to the above clustering result set based on the similarity division of low-voltage distributed photovoltaic users H As input to the user electricity preference analysis model , output variable The user's preference for energy economy, energy security and energy safety is obtained, thereby obtaining the electricity preference of distributed low-voltage photovoltaic users.

2. The method for analyzing the electricity consumption preference of low-voltage distributed photovoltaic users according to claim 1 is characterized in that: In step 1, the historical data includes photovoltaic user load power consumption attributes, electricity price coefficient, generated power, online power, power consumption, daily power consumption, weather type, season, temperature and humidity.

3. The method for analyzing the electricity consumption preference of low-voltage distributed photovoltaic users according to claim 1 is characterized in that: In step 1, data preprocessing is performed in sequence according to the steps of data cleaning, noise reduction, data integration and data standardization to obtain a preprocessed data set.

4. The method for analyzing the electricity consumption preference of low-voltage distributed photovoltaic users according to claim 1 is characterized in that: The step 3 is as follows: First, given two photovoltaic output curves and , construct the distance matrix between two photovoltaic output curves , and There are two photovoltaic output curves. i Active power value of the data point; n Indicates the number of data points of the photovoltaic output curve; Then, the matching point search is performed through the graph algorithm, and the low-voltage distributed photovoltaic user characteristics are taken as entities, and the relationship between entities is taken as edges. The user characteristics are weather type, electricity price coefficient, daily electricity consumption and power generation. The structured user information is extracted for eigenvalues, relationships and entities, and the "eigenvalue-relationship-entity" triple is generated using the knowledge graph; and according to the boundary conditions, continuity and monotonicity constraints, the distance matrix is ​​combined Constructing a collection of curved paths ,in For the first The position coordinates of the matching points, is the number of elements contained in the dynamic bending path; Calculate the cumulative distance of the dynamic curved path that meets the above constraints, and take its minimum value as the distance between the two photovoltaic output curves. distance: ， In the formula, Indicates the difference between two photovoltaic output curves Distance, which is used as an indicator to measure the similarity between any two photovoltaic output curves; The smaller the distance, the more similar the two PV output curves are. The higher; Constructing a similarity matrix , and calculate the adjacency matrix based on the similarity matrix With degree matrix ; ， ， ， ， In the formula, represents the non-negative weights of the adjacency matrix; Represents the degree of any sample point; Compute the Laplacian matrix , and perform standardization; calculate the matrix The eigenvectors corresponding to each eigenvalue form a characteristic matrix ; Each row in is regarded as a sample point. Perform clustering and obtain a set of clustering results .

5. The method for analyzing the electricity consumption preference of low-voltage distributed photovoltaic users according to claim 4 is characterized in that: The silhouette coefficient and Davidson-Boulding index were used to evaluate the clustering results.

6. The method for analyzing the electricity consumption preference of low-voltage distributed photovoltaic users according to claim 1 is characterized in that: The architecture of the user electricity preference analysis model based on improved quantum weighted neurons consists of an input layer, a hidden layer, a dropout layer, a fully connected layer, and an output layer; The input layer is the clustering result set after the distributed photovoltaic user load data is clustered. The hidden layer is a single hidden layer, and the Dropout value of the Dropout layer is set to 0.

25. After the Dropout layer, a fully connected layer is passed to get the output. The mean square error is selected as the loss function, and the Adam algorithm is used to optimize the model parameters. The mean square error calculation formula is as follows: ， The Adam algorithm replaces the gradient with the exponential moving average of the gradient at each moment, that is, the first-order moment estimate of the gradient, and its calculation formula is as follows: ， In the formula, , Respectively represent the first-order moment estimates before and after optimization; is an exponentially weighted average; is the gradient of the weight to be optimized; ， , They are the second-order moment estimates before and after optimization respectively; is an exponentially weighted average; is the square of the gradient of the weight to be optimized; Combine the first-order and second-order momentum together to update the weights. The update expression is as follows: ， , are the weights before and after the update respectively; is the learning rate; is a constant.

7. A low-voltage distributed photovoltaic user electricity preference analysis system, used to implement the low-voltage distributed photovoltaic user electricity preference analysis method according to any one of claims 1 to 6, characterized in that: include: Data acquisition and processing module: obtain historical data of low-voltage distributed photovoltaic users and perform data preprocessing; Electricity priority evaluation index acquisition module: for the pre-processed data, the electricity preference characteristics of low-voltage distributed photovoltaic users are analyzed from three perspectives: the user's photovoltaic power generation status, the electricity price coefficient, and the load selection inertia, and the electricity priority evaluation index is obtained; Clustering result set acquisition module: cluster the output curves of low-voltage distributed photovoltaic users based on the improved DTW algorithm, divide the users with similar output conditions into the same cluster, and obtain the clustering result set; Photovoltaic user electricity preference analysis model establishment module: Combined with the electricity priority evaluation index obtained in the electricity priority evaluation index acquisition module and the clustering result set obtained by clustering in the clustering result set acquisition module, a quantum weighted neuron suitable for photovoltaic user electricity preference analysis is constructed. On this basis, a photovoltaic user electricity preference analysis model based on improved quantum weighted neurons is established to analyze the preference degree of energy economy, energy quality assurance and energy safety, thereby obtaining the electricity preference of low-voltage distributed photovoltaic users.

Images