A method for establishing a typical multi - energy user model

Abstract

The present invention discloses a method for establishing a typical multi-energy user model, which overcomes the problems in the prior art, such as the lack of multi-energy characteristic portraits and energy demand prediction for typical multi-energy users, the failure to consider the uncertainty of user behavior, and the limitations of the analysis method. The method includes: extracting and mining the historical energy consumption data of typical multi-energy users; collecting parameters, labeling user characteristics, constructing user portraits, and establishing a prediction model. The present invention comprehensively considers the subjective characteristics of users and the objective environmental factors of the market, and establishes a reasonable and effective user evaluation model; constructs user portraits through labels to realize the mapping from energy consumption characteristics to user labels; uses the fuzzy C-means clustering algorithm to extract the historical energy consumption parameters of typical multi-energy users, and conducts refined modeling of the energy consumption characteristics and core demands of various users from multiple perspectives for multivariate users.

Description

Technical Field

[0001] The present invention relates to the field of integrated energy systems, and particularly to a method for establishing a typical multi-energy user model. Background Art

[0002] Integrated energy systems involve various energy forms such as electricity, gas, cold / heat, etc. There are obvious characteristic differences among various energy forms in the links of production, transmission, consumption, and storage. At the same time, there are complex mutual conversions and coupling correlations among them. In addition, compared with traditional single-energy supply systems, the deep integration of energy systems and information and communication technologies has significantly changed the operation mode of integrated energy systems. The above characteristics have brought a series of problems to the modeling, algorithms, and evaluation indicators of the reliability assessment of integrated energy systems. The reliability assessment modeling based on big data, cloud computing, and machine learning, the reliability assessment algorithm that makes full use of information flow to accurately simulate and quickly evaluate the system operation state, and the reliability evaluation index system for different energy supply forms will be the future research direction of the reliable assessment of integrated energy systems; how to establish a digital model of typical energy-consuming equipment, and then realize the networked planning and operation simulation support of multi-energy systems, and accumulate long-term operation data is the key problem in the development of integrated energy systems, and urgent technical breakthroughs are needed.

[0003] For example, a "comprehensive demand response method based on the node energy price strategy of an integrated energy system" disclosed in a Chinese patent document, with the publication number CN 113077173A, includes constructing an operation framework for an electricity-gas-heat integrated energy system; proposing a calculation method for the node energy price of an electricity-gas-heat coupled multi-energy flow network including wind turbine output; obtaining the node energy price, building an energy consumption cost model considering the load change of user comprehensive demand response, and analyzing the response behaviors of various flexible loads under the guidance of the energy pricing mechanism of the integrated energy system; the above technical solutions lack the multi-energy characteristic portrait and energy demand prediction of typical multi-energy users, do not have differentiated and precise marketing strategies, do not consider the uncertainty of user behavior, the analysis method has limitations, and the considered energy consumption data is small. Summary of the Invention

[0004] The purpose of the present invention is to overcome the problems of the prior art that lack the multi-energy characteristic portrait and energy demand prediction of typical multi-energy users, do not consider the uncertainty of user behavior, and the analysis method has limitations, and provides a method for establishing a typical multi-energy user model. The reliability assessment modeling based on big data, cloud computing, and machine learning, the reliability assessment algorithm that makes full use of information flow to accurately simulate and quickly evaluate the system operation state, comprehensively considers the subjective characteristics of users and the objective environmental factors of the market, and establishes a reasonable and effective user assessment model.

[0005] To achieve the above object, the present invention adopts the following technical solutions: A method for establishing a typical multi-energy user model, comprising the following steps:

[0006] S1: Collect actual cases and demonstration projects of integrated energy systems, analyze and summarize their main application scenarios;

[0007] S2: Use the fuzzy C-means clustering algorithm to extract the historical energy consumption parameters of typical multi-energy users, preprocess and screen the energy consumption parameters, and analyze the typical data and indicators that can reflect the energy consumption characteristics;

[0008] S3: Based on the analytic hierarchy process, introduce the entropy weight method and the technique for order preference by similarity to an ideal solution, combine the operator's subjective judgment and objective information, and establish a multi-energy user evaluation model and indicators in the integrated energy market;

[0009] S4: Abstract the specific information of multi-energy users into labels according to the evaluation model and indicators, concretize the user image through the labels, and then construct a user portrait of the energy consumption characteristics of multi-energy users;

[0010] S5: Clarify the time and space characteristics of the energy demand of typical multi-energy users, and analyze its long-term growth trend;

[0011] S6: Establish an adaptive combined prediction model to predict the power load of the energy consumption demand of users with different labels, and improve the prediction accuracy through a data-driven accuracy feedback mechanism.

[0012] The present invention collects actual cases and demonstration projects of integrated energy systems, analyzes and summarizes their main application scenarios, extracts the historical energy consumption parameters of typical multi-energy users, cleans the data, improves the value density, and analyzes and extracts high-order indicators of typical energy consumption characteristics; according to the key indicators reflecting the energy consumption characteristics of multi-energy users, further classify typical users and concretize them into typical energy consumption labels; establish a user portrait through the labels to realize the mapping from energy consumption characteristics to user labels; clarify the time and space characteristics of the energy demand of typical multi-energy users, and analyze its long-term growth trend; establish an adaptive combined prediction model to differentially predict the energy consumption demand of users with different labels, and improve the prediction accuracy through a data-driven accuracy feedback mechanism.

[0013] Preferably, the step S2 specifically includes the following steps:

[0014] S21: Use data preprocessing-related technologies to clean, integrate, transform, and normalize the original data;

[0015] S22: On the basis of the database generated by the preprocessing, use the fuzzy C-means clustering algorithm to study the feature extraction of the energy consumption characteristics of multi-energy users under different industry categories, energy consumption, and reliability requirements;

[0016] S23: Use the FCM clustering algorithm for fuzzy partitioning to define the energy consumption parameters of each data object for each integrated energy user.

[0017] By solving the objective function of the minimization clustering algorithm, the energy consumption data, economic data, and range of variation of typical users in the multi-energy user group can be obtained, the energy consumption patterns of different users can be analyzed, and the energy consumption characteristics of multi-energy users can be extracted; further, by combining the characteristics of multi-energy users themselves, such as industry category, energy consumption, and reliability requirements, with the clustering results for comparison, the integrated energy consumption characteristics of multi-energy users under different industries, energy consumption levels, and reliability requirements can be extracted.

[0018] Preferably, the step S3 specifically includes the following steps:

[0019] S31: Use the analytic hierarchy process to determine the influencing factors of user evaluation based on the electricity improvement process and the analysis results of the electricity sales market environment;

[0020] S32: Establish a judgment matrix based on the data of the user's electricity characteristics, demand characteristics, and industrial development prospects;

[0021] S33: Introduce the entropy weight method to correct the weights of the influencing factors.

[0022] Preferably, the step S4 specifically includes the following steps:

[0023] S41: Conduct feature classification and grading based on the differences in the user's basic attributes, electricity consumption behavior, payment behavior, and appeal behavior;

[0024] S42: Extract typical features from each type and assign threshold values to the labels;

[0025] S43: Based on the final labels and combined with the business demand scenario, conduct power user profiling.

[0026] Preferably, the energy consumption parameters refer to the total energy consumption of the user, energy consumption time, energy price sensitivity, energy consumption expectation, and industrial scale.

[0027] Preferably, in the step S23, the FCM clustering algorithm represents the degree to which all data objects belong to each cluster using membership degrees between [0,1], and the objective function is:

[0028]

[0029] In the formula, J m (U,P) represents the degree to which data objects belong to each cluster, and μ ik ∈1 represents the degree to which the kth data object belongs to the ith clustering center, P iDenote the cluster center of cluster i; m ∈ [0, 2] represents the weighting exponent; d ki represents the Euclidean distance between the i-th cluster center and the k-th data object.

[0030] Preferably, the step S6 specifically includes the following steps:

[0031] S61: Select the load in the past period as the training sample and perform regression processing;

[0032] S62: Construct the network structure and establish a neural network model;

[0033] S63: Predict the power load through the fitting degree of the neural network model.

[0034] Preferably, the process of regression processing in the step S61 is:

[0035] Y t = b0 + b1X t1 + b2X t2 + … + b n X tn

[0036] In the formula, X t1 , X t2 , … X tn represent the factors affecting the load change, b0, b1, … b n represent the parameter variables, and Y t represents the power load.

[0037] Preferably, the neural network model in the step S62 is:

[0038] Y (i) = F(W i , Y (i-1) , M (t-1) )

[0039] In the formula, Y i = {Y(i, t)|t = 1, 2, …, 24} represents the load vector of the i-th day; Y(i, t) represents the load of the i-th day at the t-th hour; W i represents the weight vector; M (t-1) = (m (t-1) , m (t-2) , …, m (t-k) ) represents the factors affecting the load change, and k is the data length.

[0040] Preferably, the factors affecting the load change include weather conditions, temperature, and humidity.

[0041] Preferably, the user portrait includes an individual portrait and a group portrait; the individual portrait is to attach exclusive labels to each customer according to the actual situation according to the labels in the user label library; the group portrait is composed of individual portraits that simultaneously meet the selected labels screened from the user system through the known labels.

[0042] Therefore, the present invention has the following beneficial effects:

[0043] 1. A reasonable and effective user evaluation model is established by comprehensively considering the subjective characteristics of users and the objective environmental factors of the market;

[0044] 2. By establishing a user portrait through labels, the mapping from energy consumption characteristics to user labels is realized;

[0045] 3. The fuzzy C-means clustering algorithm is used to extract the historical energy consumption parameters of typical multi-energy users, and the energy consumption parameters are preprocessed and screened to analyze the typical data and indicators that can reflect the energy consumption characteristics, and a refined model of the energy consumption characteristics and core demands of various users for multi-energy users is built from multiple perspectives. Description of the Drawings

[0046] Figure 1 is the establishment process of the typical multi-energy user model. Detailed Embodiment

[0047] The following further describes this embodiment in conjunction with the drawings and the detailed embodiment.

[0048] This embodiment provides a method for establishing a typical multi-energy user model, Figure 1 which is the establishment process of the typical multi-energy user model, including the following steps:

[0049] S1: Collect actual cases and demonstration projects of the integrated energy system, and analyze and summarize their main application scenarios;

[0050] S2: Use the fuzzy C-means clustering algorithm to extract the historical energy consumption parameters of typical multi-energy users, preprocess and screen the energy consumption parameters, and analyze the typical data and indicators that can reflect the energy consumption characteristics;

[0051] S3: Based on the analytic hierarchy process, introduce the entropy weight method and the technique for order preference by similarity to an ideal solution, and combine the subjective judgment and objective information of the operator to establish a multi-energy user evaluation model and indicators in the integrated energy market;

[0052] S4: Abstract the specific information of multi-energy users into labels according to the evaluation model and indicators, concretize the user image through the labels, and then construct a user portrait of the energy consumption characteristics of multi-energy users;

[0053] S5: Clarify the time and space characteristics of the energy demand of typical multi-energy users, and analyze its long-term growth trend;

[0054] S6: Establish an adaptive combined prediction model to predict the power load of users with different tags, and improve the prediction accuracy through a data-driven accuracy feedback mechanism.

[0055] Step S2 specifically includes the following steps:

[0056] S21: Use data preprocessing-related technologies to clean, integrate, transform, and reduce the original data;

[0057] S22: Based on the database generated by the preprocessing, use the fuzzy C-means clustering algorithm to study the energy consumption characteristics of multi-energy users under different industry categories, energy consumption, and reliability requirements for feature extraction;

[0058] S23: Use the FCM clustering algorithm for fuzzy partitioning to define the energy consumption parameters of each data object for each integrated energy user.

[0059] In step S23, the FCM clustering algorithm represents the degree of all data objects belonging to each cluster using membership degrees between [0,1], and the objective function is:

[0060]

[0061] In the formula, J m (U, P) represents the degree of data objects belonging to each cluster, μ ik ∈1 represents the degree of the k-th data object belonging to the i-th cluster center, P i represents the cluster center of cluster i; m ∈ [0,2] represents the weighting exponent; d ki represents the Euclidean distance between the i-th cluster center and the k-th data object.

[0062] Step S3 specifically includes the following steps:

[0063] S31: Use the analytic hierarchy process to determine the influencing factors of user evaluation based on the analysis results of the electricity improvement process and the electricity sales market environment;

[0064] S32: Establish a judgment matrix based on the data of the user's electricity characteristics, demand characteristics, and industrial development prospects;

[0065] S33: Introduce the entropy weight method to correct the weights of the influencing factors.

[0066] Step S4 specifically includes the following steps:

[0067] S41: Carry out feature classification and grading according to the differences in the user's basic attributes, electricity consumption behavior, payment behavior, and appeal behavior;

[0068] S42: Extract typical features from each type and assign threshold values to the labels;

[0069] S43: Based on the final label and combined with the business requirement scenario, develop a power user profile.

[0070] Step S6 specifically includes the following steps:

[0071] S61: Select the load over a past period as the training sample and perform regression processing;

[0072] S62: Construct the network structure and establish a neural network model;

[0073] S63: Predict the power load through the fitting degree of the neural network model.

[0074] The process of regression processing in step S61 is:

[0075] Y t = b0 + b1X t1 + b2X t2 + … + b n X tn

[0076] In the formula, X t1 , X t2 , … X tn represent the factors affecting the load change, b0, b1, … b n represent the parameter variables, and Y t represents the power load.

[0077] The neural network model in step S62 is:

[0078] Y (i) = F(W i , Y (i-1) , M (t-1) )

[0079] In the formula, Y i = {Y(i,t)|t = 1, 2, …, 24} represents the load vector of the i-th day; Y(i,t) represents the load of the i-th day at the t-th hour; W i represents the weight vector; M (t-1) = (m (t-1) , m (t-2) , …, m (t-k) ) represents the factors affecting the load change, and k is the data length.

[0080] First, relevant data preprocessing techniques need to be adopted to clean, integrate, transform, and reduce the original data. Based on the database generated by the preprocessing, the Fuzzy C-Means (FCM) clustering algorithm is used to study the feature extraction of the energy consumption characteristics of multi-energy users under different factors such as industry categories, energy consumption, and reliability requirements. The FCM clustering algorithm uses fuzzy partitioning to define each data object as the data information of the total energy consumption, energy consumption time, energy price sensitivity, energy consumption expectation, and industrial scale of each integrated energy user. All data objects are represented by membership degrees between [0,1] to indicate the degree to which they belong to each cluster.

[0081] Based on the analysis of the energy consumption characteristics and core demands of diversified users, determine the user energy consumption evaluation index system, and establish a reasonable and effective user evaluation model by comprehensively considering user subjective characteristics and market objective environmental factors.

[0082] Based on the Analytic Hierarchy Process (AHP), the Entropy Weight Method, and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS method), combined with the subjective judgment of the operator and objective information, a multi-energy user evaluation model and index in the integrated energy market can be established. The evaluation model mainly consists of three steps. First, use the AHP to determine the influencing factors of user evaluation (determined by the analysis results of the electricity improvement process and the electricity sales market environment), and establish a judgment matrix (the data for determining the judgment matrix comes from the electricity characteristics, demand characteristics, and industrial development prospects of users, etc.). Then, introduce the Entropy Weight Method to correct the weights of the influencing factors. The TOPSIS method is a commonly used method for multi-objective decision-making analysis in systems engineering. It is based on the normalized original data matrix to find the optimal and worst solutions in the limited solutions (represented by the optimal vector and the worst vector respectively), and then calculate the distances between each evaluation object and the optimal and worst solutions respectively to obtain the relative closeness of the evaluation object to the optimal solution, which is used as the basis for judging the quality of the evaluation object.

[0083] The TOPSIS method is the abbreviation of Technique for Order Preference by Similarity to Ideal Solution, that is, the technique for approaching the ideal solution. It is a multi-objective decision-making method. The basic idea of the method is to define the ideal solution and the negative ideal solution of the decision problem, and then find a solution in the feasible solutions that is closest to the ideal solution and farthest from the negative ideal solution. The ideal solution is generally the best solution imagined, and the corresponding attributes of it reach at least the best values of each solution. The negative ideal solution is the assumed worst solution, and its corresponding attributes are at least not better than the worst values of each solution. The decision rule for ranking the solutions is to compare the actual feasible solutions with the ideal solution and the negative ideal solution. If a certain feasible solution is closest to the ideal solution and farthest from the negative ideal solution at the same time, then this solution is the satisfactory solution of the solution set.

[0084] The working process of the present invention is as follows: collect the actual cases and demonstration projects of the integrated energy system, analyze and summarize their main application scenarios; use the fuzzy C-means clustering algorithm to extract the historical energy consumption parameters of typical multi-energy users, preprocess and screen the energy consumption parameters, and analyze the typical data and indicators that can reflect the energy consumption characteristics; based on the analytic hierarchy process, introduce the entropy weight method and the technique for order preference by similarity to an ideal solution, combine the subjective judgment of the operator with objective information, and establish an evaluation model and indicators for multi-energy users in the integrated energy market; abstract the specific information of multi-energy users into labels according to the evaluation model and indicators, concretize the user image through the labels, and then construct a user portrait of the energy consumption characteristics of multi-energy users; clarify the time and space characteristics of the energy demand of typical multi-energy users, and analyze their long-term growth trend; establish an adaptive combined prediction model to predict the power load of the energy consumption demand of users with different labels, and improve the prediction accuracy through a data-driven accuracy feedback mechanism.

[0085] The present invention is not limited to the embodiments described above. The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent transformations, improvements, etc. made based on the technical essence of the present invention shall fall within the protection scope claimed by the present invention.

Claims

1. A method for establishing a typical multi - energy user model, characterized in that, It includes the following steps: S1: Collect the actual cases and demonstration projects of the integrated energy system, and analyze and summarize their main application scenarios; S2: Use the fuzzy C-means clustering algorithm to extract the historical energy consumption parameters of typical multi-energy users, preprocess and screen the energy consumption parameters, and analyze the typical data and indicators that can reflect the energy consumption characteristics; S3: Based on the analytic hierarchy process, determine the influencing factors of user evaluation, establish a judgment matrix through the data of users' electricity characteristics, demand characteristics and industrial development prospects, introduce the entropy weight method to correct the weights of the influencing factors, and establish an evaluation model and indicators for multi-energy users in the integrated energy market in combination with the technique for order preference by similarity to an ideal solution; S4: Abstract the basic attributes, electricity consumption behavior, payment behavior and appeal behavior information of multi-energy users into labels according to the evaluation model and indicators, concretize the user image through the labels, and construct a user portrait of the energy consumption characteristics of multi-energy users; S5: Clarify the time and space characteristics of the energy demand of typical multi-energy users, and analyze their long-term growth trends; S6: Establish an adaptive combined prediction model to predict the electricity load of the energy consumption demand of users with different labels, and improve the prediction accuracy through a data-driven accuracy feedback mechanism.

2. The method for establishing a typical multi-capable user model according to claim 1, wherein The specific steps of step S2 include the following steps: S21: Use data preprocessing-related technologies to clean, integrate, transform and reduce the original data; S22: Based on the database generated by the preprocessing, use the fuzzy C-means clustering algorithm to study the feature extraction of the energy consumption characteristics of multi-source users under different industry categories, energy consumption and reliability requirements; S23: Use the FCM clustering algorithm for fuzzy division, and define the energy consumption parameters of each data object for each integrated energy user.

3. The method for establishing a typical multi-capable user model according to claim 1, characterized in that, The specific steps of step S3 include the following steps: S31: Use the analytic hierarchy process to determine the influencing factors of user evaluation based on the analysis results of the electricity improvement process and the electricity sales market environment; S32: Establish a judgment matrix through the data of users' electricity characteristics, demand characteristics and industrial development prospects; S33: Introduce the entropy weight method to correct the weights of the influencing factors.

4. A method for establishing a typical multi-capable user model according to claim 1, characterized in that, The specific steps of step S4 include the following steps: S41: Carry out feature classification and grading according to the differences in users' basic attributes, electricity consumption behavior, payment behavior and appeal behavior; S42: Extract typical features from each type and assign threshold values to the labels; S43: Carry out power user portrait according to the final labels in combination with the business demand scenario. The energy consumption parameters refer to the total energy consumption of users, energy consumption time, energy price sensitivity, energy consumption expectation and industrial scale.

5. A method for establishing a typical multi-capable user model according to claim 2, characterized in that, In the step S23, the FCM clustering algorithm represents the degree of all data objects belonging to each cluster by the membership degree between [0,1], and the objective function is: where J m (U, P) represents the degree to which a data object belongs to each cluster, and μ ik ∈ 1 represents the degree to which the k-th data object belongs to the i-th cluster center, P i represents the cluster center of cluster i; m ∈ [0, 2] represents the weighting exponent; d ki represents the Euclidean distance between the i-th cluster center and the k-th data object.

6. The method for establishing a typical multi-capable user model according to claim 1, wherein The specific steps of step S6 include the following steps: S61: Select the load in the past period as the training sample and carry out regression processing; S62: Construct the network structure and establish a neural network model; S63: Predict the electricity load through the fitting degree of the neural network model.

7. A method for establishing a typical multi-capable user model according to claim 6, characterized in that The process of regression processing in step S61 is: Y t = b0 + b1X t1 + b2X t2 + … + b n X tn Wherein, X t1 , X t2 , … X tn represent the factors affecting the load change, b0, b1, … b n represent the parameter variables, and Y t represents the electric power load.

8. A method for establishing a typical multi-capable user model according to claim 6, characterized in that The neural network model in step S62 is: Y (i) = F(W i , Y (i-1) , M (t-1) ) where Y i ={Y(i,t)|t = 1, 2, …, 24} represents the load vector of the i-th day; Y(i,t) represents the load at the t-th hour of the i-th day; W i represents the weight vector; M (t-1) =(m (t-1) , m (t-2) , …, m (t-k) ) represents the factors affecting the load change, and k is the data length.

9. A method for establishing a typical multi-capable user model according to claim 7 or 8, characterized in that The factors affecting the load change include weather conditions, temperature and humidity.

10. A method for establishing a typical multi-capable user model according to claim 1 or 4, characterized in that, The user portrait includes an individual portrait and a group portrait; the individual portrait is to attach exclusive labels to each customer according to the labels in the user label library according to the actual situation; The group portrait is composed of individual portraits that meet the selected tags simultaneously, screened from the user system through known tags.

Images