Power period division method based on simulated annealing fuzzy C-means and Bayesian

Abstract

The invention relates to a power period division method based on simulated annealing fuzzy C-means and Bayesian, and belongs to the technical field of power load data analysis. According to the method, a genetic simulated annealing algorithm and a fuzzy C-means clustering technology are combined, and verification is carried out through a Bayesian evaluation mechanism. The method comprises the following steps: firstly, collecting and analyzing historical power load data, optimizing a fuzzy C-means clustering process by using a genetic simulated annealing algorithm, and accurately dividing the power load data into different peak, flat and valley periods; thirdly, performing posterior probability evaluation on a clustering result by adopting a Bayesian evaluation mechanism to ensure the accuracy and reliability of power peak, flat and valley time division; according to the method, the peak, flat and valley characteristics of the power load can be accurately judged, and a scientific basis is provided for optimal scheduling and energy management of a power system. The method not only can improve the accuracy of time division and better adapt to various complex power system environments, but also can effectively reduce energy waste and power supply cost.

Description

Technical Field

[0001] The present invention belongs to the technical field of power load data analysis, and in particular relates to a power time period division method based on simulated annealing fuzzy C-means and Bayesian. Background Art

[0002] With the continuous development of smart grid technology, it is particularly important to accurately divide the power load into peak, flat and valley periods. The division of power peak, flat and valley periods is not only related to the stable operation of the power grid, but also directly affects the pricing strategy of the power market and the electricity costs of users. Traditional methods for dividing power peak, flat and valley periods are mainly based on historical load data and expert experience. These methods often have problems such as strong subjectivity, low accuracy and poor adaptability.

[0003] At present, some advanced clustering algorithms have been applied to the division of power peak, flat and valley periods, such as the fuzzy C-means clustering algorithm. However, these algorithms still face some challenges when processing complex power load data. First, due to the high nonlinearity and uncertainty of power load data, the traditional fuzzy C-means clustering algorithm is difficult to accurately capture the intrinsic structure of the data. Secondly, the selection of cluster centers and the determination of the number of clusters often rely on experience or trial and error, which lacks objectivity and effectiveness. Finally, the verification and evaluation of clustering results is also an important issue, and current methods often lack a systematic evaluation mechanism. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and to provide a method for dividing power time periods based on simulated annealing fuzzy C-means and Bayesian, which can accurately and efficiently divide power peak, flat and valley time periods so as to improve the accuracy and efficiency of power peak, flat and valley time period division, help the power system to better adapt to changes in the power market and changes in user demand, achieve optimal allocation and efficient utilization of power resources, and thus improve the overall performance and competitiveness of the power system.

[0005] To achieve the above object, the technical solution of the present invention is: a method for dividing power time periods based on simulated annealing fuzzy C-means and Bayesian, comprising: Collect historical user load data in the required area and pre-process it; The genetic simulated annealing fuzzy C-means clustering method is used to divide the preprocessed historical load data into time periods and obtain the time period clustering results. The Bayesian evaluation model is used to evaluate the time period clustering results. Based on historical data and prior probability, the posterior probability of the time period clustering results is estimated. If errors occur in the time period clustering results for more than three consecutive times under the same evaluation parameters, the parameters of the Bayesian evaluation model are adjusted and the time period clustering results are re-evaluated. According to the output results of the Bayesian evaluation model, the sample discrimination accuracy of the Bayesian evaluation model is calculated; if the accuracy is higher than or equal to the accuracy threshold, the time period clustering result is reliable, and the final time period clustering result is output; if the accuracy is lower than the accuracy threshold, the membership of the cluster center point in the genetic simulated annealing fuzzy C-means clustering method is adjusted according to the sample discrimination accuracy of the Bayesian evaluation model and historical experience until the accuracy requirement is met.

[0006] In one embodiment of the present invention, the user's historical load data includes the user's electricity usage in different time periods, the user's electricity usage habits and electricity demand patterns; the method of preprocessing the user's historical load data includes removing outliers and noise data, and interpolating missing data, and at the same time, the time resolution and time period division method are selected for the user's historical load data after removing outliers and noise data and interpolating missing data.

[0007] In one embodiment of the present invention, the specific implementation method of using the genetic simulated annealing fuzzy C-means clustering method to divide the pre-processed historical load data into time periods is as follows: When determining the peak, flat and valley time periods of each time point on the typical daily load curve of the power system and the user, different membership functions are selected to calculate the membership of each time point according to the change trend of the typical daily load curve of the user; for the low points on the typical daily load curve of the user, that is, the areas with low load values, a relatively small semi-trapezoidal membership function is selected to evaluate the possibility of these points being in the valley period, and it is defined as the valley membership; for the high points on the load curve, that is, the areas with high load values, a relatively large semi-trapezoidal membership function is selected to evaluate the possibility of these points being in the peak period, which is called the peak membership; Order time point collection Represents the load value set at all times on the user's typical daily load curve Represents the corresponding load value; divide the time points in the set M into three subsets: represents the set of time points during the peak period, represents a set of time points during a normal period, The set of time points representing the valley period; a relatively large semi-trapezoidal membership function is used to calculate the possibility that each time point belongs to the peak period, i.e., the peak membership; a relatively small semi-trapezoidal membership function is used to calculate the possibility that each time point belongs to the valley period, i.e., the valley membership. The formula is as follows: In the formula, , are the peak and valley membership function values ​​respectively; Time point i The load value; b is the maximum value of the load value at each time point; aIt is the minimum value of the load at each time point.

[0008] In one embodiment of the present invention, the specific steps of using the genetic simulated annealing fuzzy C-means clustering method to divide the pre-processed historical load data into time periods are as follows: (1) Initialization: Setting the size of the population GenSize , Maximum number of evolutions MAXGen , crossover probability, mutation probability, temperature cooling coefficient , Annealing temperature , termination temperature ; (2) Initialization c Cluster centers are identified and the initial population is generated. The peak membership, valley membership, and fitness of each individual are calculated for each cluster center. , where n=1,2,…, GenSize ; (3) Set the loop count variable Gen=0 ; (4) Genetic operations including mutation, crossover, selection, and boundary processing are performed on the population, and the membership of each cluster center and each sample is calculated for the newly generated individuals using the following formula: In the formula, is the objective function of clustering; M IN is the load value set M l The total number of load values; is the number of clusters; For users k No. i The samples for r The membership degree of each cluster center; is the Euclidean distance, indicating that the user k No. i The samples and r The distance between cluster centers; is the weighting coefficient; For the r Cluster centers; calculate the fitness value of each individual ,like , then the new individual replaces the old individual; otherwise, with probability Accept new individuals and discard old ones; (5) If ,but , go to step (4); otherwise, go to step (6); (6) If the current temperature , the process ends and returns to the global optimal solution, otherwise, the cooling operation is performed , go to step (3).

[0009] In one embodiment of the present invention, the specific implementation method of evaluating the time period clustering result using the Bayesian evaluation model is as follows: The time period clustering results have different distribution density functions, which are recorded as , and the prior probability distribution of each period is known, expressed as ,and ; Face the time point collection When , it is necessary to determine which time period each time point belongs to; assuming that the covariance matrix corresponding to each time period is equal, k linear discriminant functions are constructed using g indicators : In the formula, is the discriminant coefficient of the kth linear discriminant function in the gth indicator; is the characteristic of the g-th indicator; Consider each time point as a p-dimensional Euclidean space A point in ; According to the Bayesian discriminant rule, it is expected to divide the sample space into k regions ; If a time point m falls in a certain area , , it is classified as a period ; According to the Bayesian formula, time point m comes from the period The conditional probability, or posterior probability, is expressed as: If time point m actually belongs to the period , but was mistakenly identified as a period , , then the resulting loss is calculated using the loss function Indicates; in the average sense, when it belongs to the period The time point was misjudged as the period The average loss It is expressed as: Assuming that the loss of each type of misclassification is the same, the goal is to maximize the posterior probability , the discriminant function is expressed as: The judgment rule is: when point m falls in the area , classify it as a time period .

[0010] In one embodiment of the present invention, judging whether an error occurs in the time period clustering result is achieved by setting an error threshold for judging continuity.

[0011] In one embodiment of the present invention, the accuracy threshold is set to 90%.

[0012] In one embodiment of the present invention, the number of time points N is 24.

[0013] The present invention also provides a power time period division system based on simulated annealing fuzzy C-means and Bayesian evaluation, comprising a memory, a processor, and computer program instructions stored in the memory and capable of being executed by the processor. When the processor executes the computer program instructions, the method steps described above can be implemented.

[0014] The present invention also provides a computer-readable storage medium, on which computer program instructions that can be executed by a processor are stored. When the processor executes the computer program instructions, the method steps described above can be implemented.

[0015] Compared with the prior art, the present invention has the following beneficial effects: The present invention significantly improves the accuracy and efficiency of the division of power peak, flat and valley time periods by combining the genetic simulated annealing algorithm and the fuzzy C-means clustering algorithm. The method can adaptively handle the complexity and uncertainty of power load data, accurately identify the characteristics of peak, flat and valley time periods, and accurately judge the clustering results through the Bayesian evaluation model, further ensuring the reliability of time period division. In addition, the method is flexible and adaptive, and can adapt to the needs of different power systems, load characteristics and optimization goals, and provide personalized solutions for the scheduling and optimization of power systems. Its application not only improves the energy utilization efficiency of the power system, but also enhances the power supply reliability, providing strong support for the stable operation and power supply security of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is the flowchart of genetic simulated annealing fuzzy C-means clustering in the example.

[0017] Figure 2 Solve the flow chart for the method in the example.

[0018] Figure 3 is the typical daily load curve of the user in the example.

[0019] Figure 4 This is the clustering result diagram of the method in the example.

[0020] Figure 5 This is a comparison chart of the Bayesian discriminant results in the example.

[0021] Figure 6 is the Bayesian discriminant posterior probability curve in the example. DETAILED DESCRIPTION

[0022] In order to make the advantages and features of the present invention easier to understand, the preferred embodiments will be described in detail below in conjunction with the accompanying drawings. In this way, the protection scope of the present invention can be more clearly defined and made easier to be understood by those skilled in the art.

[0023] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0024] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0025] The present invention provides a method for dividing power time periods based on simulated annealing fuzzy C-means and Bayesian, comprising: Collect historical user load data in the required area and pre-process it; The genetic simulated annealing fuzzy C-means clustering method is used to divide the preprocessed historical load data into time periods and obtain the time period clustering results. The Bayesian evaluation model is used to evaluate the time period clustering results. Based on historical data and prior probability, the posterior probability of the time period clustering results is estimated. If errors occur in the time period clustering results for more than three consecutive times under the same evaluation parameters, the parameters of the Bayesian evaluation model are adjusted and the time period clustering results are re-evaluated. According to the output results of the Bayesian evaluation model, the sample discrimination accuracy of the Bayesian evaluation model is calculated; if the accuracy is higher than or equal to the accuracy threshold, the time period clustering result is reliable, and the final time period clustering result is output; if the accuracy is lower than the accuracy threshold, the sample discrimination accuracy of the Bayesian evaluation model and historical experience are used.

[0026] The following are specific implementation examples of the present invention.

[0027] like Figure 1-6 As shown, this embodiment provides a method for dividing power time periods based on simulated annealing fuzzy C-means and Bayesian, comprising the following steps: S1. Collect historical user data in the required area, and record in detail the load of each type of user in each time period; S2. Use the genetic simulated annealing fuzzy C-means clustering method to divide the preprocessed historical load data into time periods. Combining the global search capability of the genetic algorithm and the local optimization capability of the simulated annealing algorithm, the user's net load is accurately divided into time periods by iteratively optimizing the cluster center and the number of clusters, and the time period clustering results are obtained; S3. Use Bayesian discriminant theory to evaluate the time period clustering results. Based on historical data and prior probability, estimate the posterior probability of the clustering results to determine the rationality of the clustering. S4. If errors occur in the time period division results for more than three consecutive times under the same evaluation parameters, the parameters of the Bayesian evaluation model will be adjusted, and S3 will be re-executed to evaluate the clustering results. Otherwise, S5 will be executed. S5. Calculate the sample discrimination accuracy of the Bayesian discriminant model based on the results of the Bayesian evaluation model. If the accuracy is higher than 90%, the time period division result is reliable, and the final time period division result is output. If the accuracy is lower than 90%, adjust the membership of the cluster center point in the genetic simulated annealing fuzzy C-means clustering method based on the Bayesian discriminant result and historical experience, and re-execute the entire process from S1 until the accuracy requirement is met.

[0028] In step S1, firstly, the scope of the required area is clarified, which is usually defined based on the needs of grid dispatch and management. Secondly, the load data of each type of user in each time period is obtained through the power monitoring system, smart meter or other data acquisition equipment. These data not only record the power usage of users in different time periods, but also reflect the power usage habits and power demand patterns of users. To ensure the accuracy and completeness of the data, the collected data needs to be cleaned and preprocessed, outliers and noise data need to be removed, and missing data needs to be reasonably interpolated. In addition, it is necessary to select the appropriate time resolution and time period division method according to the actual situation of the data in order to better reflect the changing trend and characteristics of the power load.

[0029] In step S2, the time period division method based on genetic simulated annealing fuzzy C-means clustering refers to: when determining the peak, flat and valley time periods of each time point on the typical daily load curve of the power system and the user, different membership functions are selected to calculate the membership of each time point according to the change trend of the load curve. For the low points on the load curve (i.e., the area with lower load values), a relatively small semi-trapezoidal membership function is selected to evaluate the possibility of these points being in the valley period, and it is defined as the valley membership. For the high points on the load curve (i.e., the area with higher load values), a relatively large semi-trapezoidal membership function is selected to evaluate the possibility of these points being in the peak period, which is called the peak membership.

[0030] In this example, the number of time points is 24, and the time point set Represents the load value set at all times on the user's typical daily load curve Represents the corresponding load value; divide the time points in the set M into three subsets: represents the set of time points during the peak period, represents a set of time points during a normal period, The set of time points representing the valley period; a relatively large semi-trapezoidal membership function is used to calculate the possibility that each time point belongs to the peak period, i.e., the peak membership; a relatively small semi-trapezoidal membership function is used to calculate the possibility that each time point belongs to the valley period, i.e., the valley membership. The formula is as follows: In the formula, , are the peak and valley membership function values ​​respectively; Time point i The load value; b is the maximum value of the load value at each time point; a It is the minimum value of the load at each time point.

[0031] like Figure 1 As shown, the specific steps of using the genetic simulated annealing fuzzy C-means clustering method to divide the pre-processed historical load data into time periods are as follows: (1) Initialization: Setting the size of the population GenSize , Maximum number of evolutions MAXGen , crossover probability, mutation probability, temperature cooling coefficient , Annealing temperature , termination temperature ; (2) Initialization c Cluster centers are identified and the initial population is generated. The peak membership, valley membership, and fitness of each individual are calculated for each cluster center. , where n=1,2,…, GenSize ; (3) Set the loop count variable Gen=0 ; (4) Genetic operations including mutation, crossover, selection, and boundary processing are performed on the population, and the membership of each cluster center and each sample is calculated for the newly generated individuals using the following formula: In the formula, is the objective function of clustering; M IN is the load value setM l The total number of load values; is the number of clusters; For users k No. i The samples for r The membership degree of each cluster center; is the Euclidean distance, indicating that the user k No. i The samples and r The distance between cluster centers; is the weighting coefficient; For the r Cluster centers; calculate the fitness value of each individual ,like , then the new individual replaces the old individual; otherwise, with probability Accept new individuals and discard old ones; (5) If ,but , go to step (4); otherwise, go to step (6); (6) If the current temperature , the process ends and returns to the global optimal solution, otherwise, the cooling operation is performed , go to step (3).

[0032] In step S3, the specific implementation method of evaluating the time period clustering result using the Bayesian evaluation model is as follows: The time period clustering results have different distribution density functions, which are recorded as , and the prior probability distribution of each period is known, expressed as ,and ; Face the time point collection When , it is necessary to determine which time period each time point belongs to; assuming that the covariance matrix corresponding to each time period is equal, k linear discriminant functions are constructed using g indicators : In the formula, is the discriminant coefficient of the kth linear discriminant function in the gth indicator; is the characteristic of the g-th indicator; Consider each time point as a p-dimensional Euclidean space A point in ; According to the Bayesian discriminant rule, it is expected to divide the sample space into k regions ; If a time point m falls in a certain area , , it is classified as a period ; According to the Bayesian formula, time point m comes from the period The conditional probability, or posterior probability, is expressed as: If time point m actually belongs to the period , but was mistakenly identified as a period , , then the resulting loss is calculated using the loss function Indicates; in the average sense, when it belongs to the period The time point was misjudged as the period The average loss It is expressed as: Determining the loss function in practice is a challenge, so the goal is to maximize the posterior probability, assuming that the loss for each type of misclassification is the same. , the discriminant function is expressed as: The judgment rule is: when point m falls in the area , classify it as a time period .

[0033] In step S4, an error threshold of 90% is set for judging continuity, that is, errors occur in the judgment period division results for more than three consecutive times. When the threshold is met, the system will trigger a parameter adjustment mechanism, which will adjust the relevant parameters in the Bayesian evaluation model to optimize the performance of the model. After adjusting the parameters, the system will re-execute step S3, that is, judge the clustering results based on the genetic simulated annealing fuzzy C-means clustering algorithm. If there are no errors in the judgment period division results for more than three consecutive times, the system will continue to execute the subsequent step S5; In step S5, the sample discrimination accuracy of the Bayesian discriminant model is first calculated according to the output result of the Bayesian evaluation model. In this method, the accuracy threshold is set to 90%. If the calculated discrimination accuracy is higher than or equal to 90%, it indicates that the time period division result is reliable, and the system will output the final time period division result as the basis for the peak, flat and valley time period division of the power system. If the discrimination accuracy is lower than 90%, the system will adjust the membership of the cluster center point in the genetic simulated annealing fuzzy C-means clustering method according to the Bayesian discrimination result and historical experience, change the degree of attribution of the sample data to the cluster center point, and affect the clustering result. By continuously optimizing the membership parameters, the discrimination accuracy of the model can be improved, thereby obtaining a more accurate time period division result. After adjusting the membership, the system will start the entire process from step S1 again, including data preprocessing, genetic simulated annealing fuzzy C-means clustering, Bayesian evaluation and other steps. Through iterative optimization, the system will continue to improve the performance of the model until the discrimination accuracy requirement is met.

[0034] The solution flow chart of the method proposed by the present invention is as follows: Figure 2 shown.

[0035] In order to verify the accuracy of the method and system proposed in this invention, this paper uses the typical daily load data of users in a certain industrial park in a certain province as an example for analysis. The typical daily load curve of users is as follows: Figure 3 The proposed method is solved using MATLAB R2022b, and the results are shown in Figures 4 to 6 shown.

[0036] Depend on Figure 4 As shown, the method proposed in the present invention combines the global search capability of the genetic simulated annealing algorithm and the data classification advantage of the fuzzy C-means clustering, ensuring the accuracy and rationality of the time period division. Through the optimization of the genetic simulated annealing algorithm, the optimal clustering center can be found in the complex power load data, providing a solid foundation for the subsequent fuzzy C-means clustering. The fuzzy C-means clustering method fully considers the fuzziness and uncertainty of the power load, and describes the degree of membership of the load data to each time period through the fuzzy membership function. This method not only improves the accuracy of clustering, but also makes the time period division more in line with the actual situation. The obtained time period division results are shown in Table 1.

[0037] Table 1 Time period division results After clustering the time periods corresponding to each time point, the corresponding posterior probability is calculated by substituting the maximum function value into the above method. Figure 5-6As shown in the figure, the evaluation results show that there is no misjudgment in the time period division of a typical day, and the overall accuracy rate is as high as 98.84%, which fully meets the preset error standard. The results strongly verify that the time period division method based on genetic simulated annealing fuzzy C-means clustering established by the present invention can accurately identify the time period in which the load is located and output accurate time period division results accordingly.

[0038] The present invention provides a method for dividing power time periods based on simulated annealing fuzzy C-means and Bayesian. The method collects and analyzes historical power load data, uses genetic simulated annealing algorithm to optimize the fuzzy C-means clustering process, and realizes accurate classification of power loads, thereby dividing accurate power peak, flat and valley time periods. Based on the clustering results, the time period division is verified and optimized using the Bayesian evaluation mechanism to ensure the accuracy and reliability of the time period division. This method not only improves the operating efficiency of the power dispatching system, but also effectively improves the reliability of power supply, and provides a more accurate energy management strategy for the power system. It has significant technical advantages and broad application prospects for solving the problem of power peak, flat and valley time period division.

[0039] The present invention also provides a power time period division system based on simulated annealing fuzzy C-means and Bayesian evaluation, comprising a memory, a processor, and computer program instructions stored in the memory and capable of being executed by the processor. When the processor executes the computer program instructions, the method steps described above can be implemented.

[0040] The present invention also provides a computer-readable storage medium, on which computer program instructions that can be executed by a processor are stored. When the processor executes the computer program instructions, the method steps described above can be implemented.

[0041] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0042] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0043] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0044] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0045] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.

Claims

1. A method for dividing power time periods based on simulated annealing fuzzy C-means and Bayesian, characterized in that: include: Collect historical user load data in the required area and pre-process it; The genetic simulated annealing fuzzy C-means clustering method is used to divide the preprocessed historical load data into time periods and obtain the time period clustering results. The Bayesian evaluation model is used to evaluate the time period clustering results. Based on historical data and prior probability, the posterior probability of the time period clustering results is estimated. If errors occur in the time period clustering results for more than three consecutive times under the same evaluation parameters, the parameters of the Bayesian evaluation model are adjusted and the time period clustering results are re-evaluated. According to the output results of the Bayesian evaluation model, the sample discrimination accuracy of the Bayesian evaluation model is calculated; If the accuracy is higher than or equal to the accuracy threshold, the time period clustering result is reliable, and the final time period clustering result is output; If the accuracy is lower than the accuracy threshold, the membership of the cluster center point in the genetic simulated annealing fuzzy C-means clustering method is adjusted according to the sample discrimination accuracy of the Bayesian evaluation model and historical experience until the accuracy requirement is met; The specific implementation method of using the genetic simulated annealing fuzzy C-means clustering method to divide the pre-processed historical load data into time periods is as follows: When determining the peak, flat and valley time periods of each time point on the typical daily load curve of the power system and the user, different membership functions are selected to calculate the membership of each time point according to the change trend of the typical daily load curve of the user; for the low points on the typical daily load curve of the user, that is, the areas with low load values, a relatively small semi-trapezoidal membership function is selected to evaluate the possibility of these points being in the valley period, and it is defined as the valley membership; for the high points on the load curve, that is, the areas with high load values, a relatively large semi-trapezoidal membership function is selected to evaluate the possibility of these points being in the peak period, which is called the peak membership; Order time point collection Represents the load value set at all times on the user's typical daily load curve Represents the corresponding load value; divide the time points in the set M into three subsets: represents the set of time points during the peak period, represents a set of time points during a normal period, The set of time points representing the valley period; a relatively large semi-trapezoidal membership function is used to calculate the possibility that each time point belongs to the peak period, i.e., the peak membership; a relatively small semi-trapezoidal membership function is used to calculate the possibility that each time point belongs to the valley period, i.e., the valley membership. The formula is as follows: In the formula, , are the peak and valley membership function values ​​respectively; Time point i The load value; b is the maximum value of the load value at each time point; a is the minimum value of the load value at each time point; The specific steps of using the genetic simulated annealing fuzzy C-means clustering method to divide the pre-processed historical load data into time periods are as follows: (1) Initialization: Setting the size of the population GenSize , Maximum number of evolutions MAXGen , crossover probability, mutation probability, temperature cooling coefficient , Annealing temperature , termination temperature ; (2) Initialization c Cluster centers are identified and the initial population is generated. The peak membership, valley membership, and fitness of each individual are calculated for each cluster center. , where n=1,2,…, GenSize ; (3) Set the loop count variable Gen=0 ; (4) Implement genetic operations including mutation, crossover, selection, and boundary processing on the population, and calculate the membership of each cluster center and each sample for the newly generated individuals using the following formula: In the formula, is the objective function of clustering; M IN is the load value set M l The total number of load values; is the number of clusters; For users k No. i For the sample r The membership degree of each cluster center; is the Euclidean distance, indicating that the user k No. i The samples and r The distance between cluster centers; is the weighting coefficient; For the r Cluster centers; calculate the fitness value of each individual ,like , then the new individual replaces the old individual; otherwise, with probability Accept new individuals and discard old ones; (5) If ,but , go to step (4); otherwise, go to step (6); (6) If the current temperature , the process ends and returns to the global optimal solution, otherwise, the cooling operation is performed , go to step (3).

2. The method for dividing power time periods based on simulated annealing fuzzy C-means and Bayesian according to claim 1 is characterized in that: User historical load data includes the user's electricity usage in different time periods, the user's electricity usage habits and electricity demand patterns; The method of preprocessing the user historical load data includes removing outliers and noise data and interpolating missing data. At the same time, the time resolution and time period division method are selected for the user historical load data after removing outliers and noise data and interpolating missing data.

3. The method for dividing power time periods based on simulated annealing fuzzy C-means and Bayesian according to claim 1 is characterized in that: The specific implementation method of using the Bayesian evaluation model to evaluate the time period clustering results is as follows: The time period clustering results have different distribution density functions, which are recorded as , and the prior probability distribution of each period is known, expressed as ,and ; Face the time point collection When , it is necessary to determine which time period each time point belongs to; assuming that the covariance matrix corresponding to each time period is equal, k linear discriminant functions are constructed using g indicators : In the formula, is the discriminant coefficient of the kth linear discriminant function in the gth indicator; is the characteristic of the g-th indicator; Consider each time point as a p-dimensional Euclidean space A point in ; According to the Bayesian discriminant rule, it is expected to divide the sample space into k regions ; If a time point m falls in a certain area , , it is classified as a period ; According to the Bayesian formula, time point m comes from the period The conditional probability, or posterior probability, is expressed as: If time point m actually belongs to the period , but was mistakenly identified as a period , , then the resulting loss is calculated using the loss function Indicates; in the average sense, when it belongs to the period The time point was misjudged as the period The average loss It is expressed as: Assuming that the loss of each type of misclassification is the same, the goal is to maximize the posterior probability , the discriminant function is expressed as: The judgment rule is: when point m falls in the area , classify it as a time period .

4. The method for dividing power time periods based on simulated annealing fuzzy C-means and Bayesian according to claim 1 is characterized in that: The determination of whether the time period clustering result is wrong is achieved by setting an error threshold for judging continuity.

5. The method for dividing power time periods based on simulated annealing fuzzy C-means and Bayesian according to claim 1 is characterized in that: The accuracy threshold is set to 90%.

6. The method for dividing power time periods based on simulated annealing fuzzy C-means and Bayesian according to claim 1 or 3, characterized in that: The number of time points N is set to 24.

Images