Power distribution network coordination control method and system based on multiple flexible loads

Abstract

The invention relates to the technical field of power system operation control, in particular to a power distribution network coordination control method and system based on multiple flexible loads, and the method comprises the steps: collecting the power data of each flexible load and each distributed power supply in a power distribution network in real time; equally dividing each day into each time period, acquiring a distribution characteristic array and a power fluctuation degree of each load in each time period, acquiring a reference power utilization difference factor of each load in each time period, and acquiring a characteristic vector of each load in each period; classifying all loads in each period; and acquiring the power demand degree of all flexible loads in each type in all days in each period in each same time period, acquiring the output characteristic value of each distributed power supply in each time period in each day in the next period, and acquiring the power upper limit of each load in the next time period of each time period in each day in the next period. According to the method, the accuracy of flexible load classification is improved, a reasonable power adjustment strategy is formulated for the flexible loads, and the source load matching degree of the power distribution network is improved.

Description

Technical Field

[0001] This application relates to the field of power system operation and control technology, specifically to a distribution network coordinated control method and system based on multiple flexible loads. Background Technology

[0002] Flexible loads reduce power losses in the power system by participating in the power control of the distribution network. Currently, distributed generation (DG) accounts for a high proportion in urban distribution networks, and the output and location distribution of DG exhibit significant unevenness. Furthermore, urban distribution networks contain numerous flexible loads. By coordinating the control of multiple flexible loads and adapting to changes in the operating status of distributed generation within the distribution network, coordinated control of the distribution network can be achieved, effectively reducing power losses.

[0003] Current urban power distribution networks primarily coordinate the control of multiple flexible loads by setting fixed power limits for different types of flexible loads at different time periods. This limits the power regulation range of these loads during operation, aiming to achieve source-load matching and reduce energy consumption. However, considering the redundancy and non-uniform temporal and spatial distribution of electrical data generated by flexible loads in urban areas, the above methods lack accuracy in identifying flexible loads, resulting in poor control performance for flexible load power regulation. Specifically, on the one hand, when flexible loads have high electricity consumption levels, excessively low power limits negatively impact user experience; on the other hand, when distributed power sources have high output levels, fixed power limits reduce source-load matching, causing additional energy losses. Therefore, it is necessary to improve the accuracy of flexible load identification, thereby increasing the accuracy of setting power limits for flexible loads, improving source-load matching in the power distribution network, and reducing energy losses. Summary of the Invention

[0004] In view of the above, it is necessary to provide a distribution network coordinated control method and system based on multiple flexible loads. Compared with the traditional power system operation control method used for flue gas treatment, this method can adjust the response speed of the control system to parameters and adjust the response deviation in a timely manner, thereby improving the stability of incinerator combustion. In a first aspect, embodiments of this application provide a distribution network coordinated control method based on multiple flexible loads, the method comprising the following steps: Real-time collection of power data of each flexible load and each distributed power source in the distribution network of any urban area; the flexible load is referred to as load. The day is divided into time periods. By analyzing the distribution and dispersion of power data for each load in each time period, the distribution feature array and power fluctuation of each load in each time period are obtained. Then, by comparing the power data of each load in the same time period of the previous cycle, the reference electricity consumption difference factor of each load in each time period of each day in each cycle is obtained, as well as the distance of each load to each distributed power source access point, the feature vector of each load in each cycle is obtained. Preset filtering parameters are used to filter the components in the feature vector, thereby classifying all loads in each cycle. An intelligent optimization algorithm is used to adjust the filtering parameters to obtain the optimal classification results of all loads in each cycle. The filtering parameters include the filtering ratio of the components in the feature vector and the local maximum retention length. Based on the optimal classification results, the power demand of all flexible loads within each category is obtained by using the distribution feature array of all loads within each category in the same time period of each day in each cycle. Combined with the power data of each distributed power source in each cycle and the next cycle in the same time period, the output feature value of each distributed power source in each time period of each day in the next cycle is obtained, and the power upper limit of each load in the next time period of each day in the next cycle is obtained.

[0005] In one embodiment, the process of obtaining the distribution feature array is as follows: The average power data of each load in each time period is used as the global average of each load in each time period; Each time period is divided into sub-time periods. The average power data of each load in each sub-time period is taken as the local average of each load in each sub-time period. For each load in each day of each cycle, the proportion of sub-time periods in which the local average is greater than the global average is calculated among all sub-time periods. The distribution feature array consists of the global mean and the quantity percentage.

[0006] In one embodiment, the process of obtaining the power fluctuation is as follows: Calculate the dispersion of power data for each load in each sub-period. For each load in each period of each day within each cycle, take the mean of the dispersion of each load in all sub-periods as the power fluctuation.

[0007] In one embodiment, the process of obtaining the reference electricity consumption difference factor is as follows: The power data of each load in each time period are arranged in chronological order to form the power sequence of each load in each time period; for each load in the j-th time period of all days in the previous cycle, the average distance between any two time periods is calculated as the reference power consumption difference factor of each load in the j-th time period of each day in each cycle.

[0008] In one embodiment, the process of obtaining the feature vector is as follows: The distribution characteristics of each load in each time period, power fluctuation and reference electricity consumption difference factor are used to form the time period vector of each load in each time period; the time period vectors of each load in all days of each cycle are sorted together according to time sequence, and the time period vectors are connected end to end in sequence to form the time vector of each load in each cycle. The distances of each load to all distributed power supply access points within each cycle are used to form the spatial vector of each load within each cycle. The feature vector is composed of the time vector and the spatial vector.

[0009] In one embodiment, the method for obtaining the fitness value in the intelligent optimization algorithm during the process of obtaining the optimal classification result is as follows: Based on the screening results of the feature vectors of all loads in each period, a clustering algorithm is used to classify all loads in each period. For each classification, the membership degree between each load and the center of each class is obtained. For each class, the sum of the similarity of the membership degrees between any two loads is calculated. The cumulative value of the sum of the values ​​of the membership degrees of all classes obtained in a single classification is used as the fitness value of the screening parameter corresponding to the single classification.

[0010] In one embodiment, the process of obtaining the electricity demand is as follows: For each category in the optimal classification result, the mean of the first component and the mean of the second component in the distribution characteristic array of all loads in each category for all days in each cycle at the same time period are denoted as the first mean and the second mean, respectively. The electricity demand is the product of the first mean and the second mean.

[0011] In one embodiment, the process of obtaining the output characteristic value is as follows: Calculate the average power data of each distributed power source in the j-th time period of all days in each cycle. Then, take the percentage of the power data of each distributed power source in the j-th time period of each day in the next cycle that is greater than the average value as the output characteristic value of each distributed power source in the j-th time period of each day in the next cycle.

[0012] In one embodiment, the expression for the power limit is: ; In the formula, This represents the upper limit of the power of the i-th load during the j+1th time period on day f within the T+1th cycle; This represents the maximum power value of the last sub-period within the j-th time period of the f-th day in the T+1-th cycle; , These represent the electricity demand of the i-th load during all days of the T-th cycle in the (j+1)-th and j-th time periods, respectively. This represents the output characteristic value of the distributed power source that is closest to the i-th load during the j+1th time period on day f within the (T+1)-th cycle.

[0013] Secondly, embodiments of this application also provide a distribution network coordination control system based on multiple flexible loads, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above-described distribution network coordination control methods based on multiple flexible loads.

[0014] This application has at least the following beneficial effects: This application analyzes the power consumption and power demand of flexible loads by analyzing the distribution of their power data, analyzes the fluctuation of their power data by analyzing the dispersion of their power data, analyzes the periodic variation of flexible loads by combining the differences in power consumption of flexible loads at the same time on different days in the previous cycle, and considers the locational relationship between flexible loads and distributed power sources. This allows for a more accurate description of the characteristics of flexible loads, provides a basis for subsequent classification, and helps to formulate differentiated power adjustment strategies for flexible loads of different types and locations, thereby improving the coordinated control capability of the distribution network. The system filters components in the feature vector using preset screening parameters and adjusts these parameters through an intelligent optimization algorithm to obtain the optimal classification result. This effectively removes redundant components from the feature vector while retaining features that are important for classification, thereby improving the accuracy of flexible load classification. Based on the optimal classification result and the power demand of flexible loads within each type of load, as well as the output characteristics of distributed power sources, the system obtains the power upper limit for each load in each time period of the next cycle. This allows for the development of reasonable power regulation strategies for each flexible load based on its power demand and the output of distributed power sources, avoiding the problem of setting the power upper limit too high or too low. This ensures normal user operation while improving the source-load matching degree of the distribution network and reducing energy loss. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating the steps of a distribution network coordinated control method based on multiple flexible loads provided in one embodiment of this application; Figure 2 This is a schematic diagram illustrating the process of adjusting the power limit. Detailed Implementation

[0017] In the description of the embodiments in this application, the words "exemplary," "or," and "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary," "or," and "for example" is intended to present the relevant concepts in a specific manner.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. It should be understood that, unless otherwise stated, " / " in this application means "or".

[0019] It should also be noted that the terms "first" and "second" in this application are used to distinguish similar objects, rather than to describe a specific order or sequence.

[0020] The following description, in conjunction with the accompanying drawings, details the specific scheme of the distribution network coordinated control method and system based on multiple flexible loads provided in this application.

[0021] Please see Figure 1 The diagram illustrates a flowchart of a distribution network coordinated control method based on multiple flexible loads according to an embodiment of this application. The method includes the following steps: Step 1: Collect power data of each flexible load and each distributed power source in the distribution network of any urban area in real time. Flexible loads are referred to as loads.

[0022] Taking a distribution network in any urban area as an example, power sensors are installed at each flexible load and each distributed power source in the distribution network to collect power data of each flexible load and each distributed power source in real time. The power data of both flexible loads and distributed power sources are collected starting at midnight every day.

[0023] In this embodiment, the sampling period for the power data of the flexible load is 1 minute, and the sampling period for the power data of the distributed power source is 30 minutes. The sampling periods for the power data of the flexible load and the power data of the distributed power source are preset by humans. The implementer can set them according to the actual situation. This application does not impose any special restrictions.

[0024] Considering the changes in the electricity consumption patterns of flexible loads and the output patterns of distributed power sources, an adjustment period for the power dataset of flexible loads is preset. Taking the i-th flexible load as an example, all power data of the i-th flexible load in each period are used to form the power dataset of the i-th flexible load in each period, which is used to evaluate the power upper limit of the i-th flexible load in the next period of each period. Specifically, it evaluates the power upper limit of the i-th flexible load in each time period of each day in the next period of each period.

[0025] In this embodiment, the adjustment period of the power dataset of the flexible load is 7 days. The length of the adjustment period of the power dataset of the flexible load is preset by the user. The implementer can set it according to the actual situation. This application does not impose any special restrictions.

[0026] Step 2: Divide each day into time periods, obtain the distribution feature array and power fluctuation of each load in each time period, obtain the reference electricity consumption difference factor of each load in each period of each day in each cycle, and obtain the feature vector of each load in each cycle; classify all loads in each cycle and obtain the optimal classification result of all loads in each cycle.

[0027] Step 2.1: Divide each day into time periods. Based on the distribution and dispersion of power data of each load in each time period, obtain the distribution feature array and power fluctuation of each load in each time period. Then, by comparing the power data of each load in the same time period in the previous cycle, obtain the reference electricity consumption difference factor of each load in each day and time period in each cycle, as well as the distance of each load to each distributed power source access point, to obtain the feature vector of each load in each cycle.

[0028] Considering that the power changes of different flexible loads vary in different time periods, the day is divided evenly into a preset number of time periods, starting from 0:00, and each time period is further divided into a preset number of sub-time periods.

[0029] In this embodiment, the preset quantity and the preset number are 8 and 6 respectively. The preset quantity and the preset number are preset by humans. The implementer can set them according to the actual situation. This application does not impose any special restrictions.

[0030] For any given day, taking the j-th time period as an example, the average of all power data of the i-th flexible load in each sub-time period is recorded as the local average of the i-th flexible load in each sub-time period. The average of all power data of the i-th flexible load in the j-th time period is used as the global average of the i-th flexible load in the j-th time period. The global average is used to reflect the electricity consumption of the i-th flexible load in the j-th time period.

[0031] Within the j-th time period, calculate the proportion of sub-time periods where the local mean is greater than the global mean among all sub-time periods, and use this proportion as the power demand characteristic value of the i-th flexible load in the j-th time period; the larger the power demand characteristic value, the higher the power demand of the i-th flexible load in the j-th time period. Combine the global mean of the i-th flexible load in the j-th time period with the power demand characteristic value to form the distribution characteristic array of the i-th flexible load in the j-th time period.

[0032] Calculate the dispersion of all power data for the i-th flexible load in each sub-period. The mean of the dispersion of the i-th flexible load in all sub-periods within the j-th period is denoted as the power fluctuation of the i-th flexible load in the j-th period. The greater the power fluctuation, the greater the fluctuation of the power data of the i-th flexible load within the j-th period.

[0033] In this embodiment, the dispersion is the coefficient of variation. The calculation of the coefficient of variation is a well-known technique and will not be described in detail here. As other implementation methods, based on the ability to measure the unevenness of power data distribution, implementers may use other existing techniques, such as standard deviation, variance, etc. This application does not impose any special restrictions.

[0034] To enhance the coordinated control capability of the distribution network by setting different power adjustment strategies for flexible loads of different types and locations, it is necessary to improve the accuracy of the classification of flexible loads of different types and locations. Considering the differences in user usage patterns for different types of flexible loads within a week, based on the power dataset of the i-th flexible load in each cycle, the power consumption difference of the i-th flexible load between the same time periods on all days in each cycle is obtained. This is used to reflect the power consumption difference of the i-th flexible load in the same time period on different days in each cycle. The specific process is as follows: The power data of the i-th flexible load in each time period are arranged chronologically to form the power sequence of the i-th flexible load in each time period. For the same time period in all days of the previous cycle, the distance between any two time periods is calculated. The average of the distances between all any two time periods is used as the electricity consumption difference degree between the i-th flexible load in the same time period in all days of the previous cycle. Taking the j-th time period as an example, the electricity consumption difference degree between the j-th time period in all days of the previous cycle of the i-th flexible load is used as the reference electricity consumption difference factor for the j-th time period in all days of the previous cycle.

[0035] In this embodiment, the distance between power sequences is the DTW (Dynamic Time Warping) distance, which is a well-known technique and will not be described further in this application.

[0036] The global average value, power demand characteristic value, power fluctuation degree, and reference electricity consumption difference factor of the i-th flexible load in each time period are used to form the time period vector of the i-th flexible load in each time period. The time period vectors of the i-th flexible load in all days of each cycle are sorted together according to time sequence, and the time period vectors are connected end to end in sequence to form the time vector of the i-th flexible load in each cycle, reflecting the electricity consumption pattern characteristics of the i-th flexible load in each cycle.

[0037] Considering the source-load matching of flexible loads and distributed generation in the distribution network, the distances from the i-th flexible load to all distributed generation connection points in each cycle are used to form the spatial vector of the i-th flexible load in each cycle, reflecting the positional relationship between the i-th flexible load and the distributed generation connection points. Following the method for obtaining the spatial vector of the i-th flexible load, the spatial vectors of the remaining flexible loads are obtained. The component at the same location in the spatial vectors of all flexible loads represents the distance from all flexible loads to the same distributed generation connection point. In this embodiment, the distance from the flexible load to the distributed generation connection point is the Euclidean distance.

[0038] To comprehensively consider the long-term electricity consumption patterns of flexible loads and the locational relationship between flexible loads and distributed power supply access points, this embodiment connects the time vector and spatial vector of the i-th flexible load in each cycle to form the feature vector of the i-th flexible load in each cycle, which is used as the basis for classifying flexible loads in urban areas. The time vector comes first, followed by the spatial vector.

[0039] Step 2.2: Preset filtering parameters to filter the components in the feature vector, and then classify all loads in each cycle. Use an intelligent optimization algorithm to adjust the filtering parameters to obtain the optimal classification results of all loads in each cycle. The filtering parameters include the filtering ratio of the components in the feature vector and the local maximum retention length.

[0040] Because different users have a certain degree of randomness when using flexible loads, and the output and location distribution of distributed power sources in urban areas are non-uniform, there are a lot of redundant components in the feature vector, which affects the accurate classification of flexible loads of different types and locations.

[0041] To improve the accuracy of flexible load segmentation in urban areas, this embodiment constructs a feature mask vector with the same length as the feature vector. The components in the feature mask vector take values ​​of 0 or 1, used to filter out redundant components in the feature vector. The construction method is as follows: First, filter parameters are set, including the filtering ratio of components in the feature vector and the local maximum retention length. Specifically, the percentage of 0 components in the feature mask vector is used as the filtering ratio, and the maximum number of consecutive 1 components is used as the local maximum retention length. The filtering ratio and the local maximum retention length are combined to form the parameter vector of the feature mask vector. Based on the parameter vector, feature mask vectors that conform to the filtering ratio and the local maximum retention length are randomly generated.

[0042] In this embodiment, the screening ratio ranges from [0,1) and the local maximum retention length ranges from [5,100]. The screening ratio and the local maximum retention length ranges are preset by the user and can be set by the user according to the actual situation. This application does not impose any special restrictions.

[0043] When dividing all flexible loads within an urban area into a single cycle, the feature mask vector remains consistent.

[0044] The method for filtering components in a feature vector using a feature mask vector is as follows: multiply the feature mask vector with the components at the same positions in the feature vector, and then reassemble the multiplication result into a new feature vector according to the original positions. This is called the filtered feature vector.

[0045] Based on the screening feature vectors of all flexible loads in each period, a clustering algorithm is used to obtain the clusters of all flexible loads in each period, as well as the membership degree between each flexible load and the center of each cluster.

[0046] In this embodiment, the Fuzzy C-means (FCM) clustering algorithm is used to obtain the clusters of all flexible loads in each period, as well as the membership degree between each flexible load and the center of each cluster. The FCM algorithm is a well-known technology and will not be described in detail in this application. As another implementation method, based on the ability to obtain the clusters of all flexible loads in each period and the membership degree between each flexible load and the center of each cluster according to the screening feature vector of all flexible loads in each period, the implementer may use other existing feasible technologies.

[0047] To improve the accuracy of flexible load segmentation, a particle swarm optimization algorithm is used to adjust the parameter vector of the feature mask vector, thereby optimizing the screening process for redundant components in the feature vector and improving the accuracy of flexible load segmentation.

[0048] A random number is selected within the value range [0,1) as the initial screening ratio in the parameter vector, and an integer is randomly selected from 5 to 100 as the initial local maximum retention length in the parameter vector. The obtained initial screening ratio and the initial local maximum retention length are then combined to form the initial parameter vector of the feature mask vector. Using the above method, a preset number of initial parameter vectors are obtained. In this embodiment, the number of initial parameter vectors is 30. The number of initial parameter vectors is preset by the user, and the implementer can set it according to the actual situation. This application does not impose any special restrictions.

[0049] In clustering algorithms, the distribution of membership degrees between each flexible load and the centers of different clusters reflects the accuracy of the partitioning. Based on the above analysis, in each cycle, the membership degrees between each flexible load and all cluster centers are used to form the membership vector of each flexible load. For a single clustering, the sum of the similarities between the membership vectors of any two flexible loads in each cluster is calculated; the cumulative sum of these sums for all clusters obtained from a single clustering is used as the fitness value of the parameter vector corresponding to that single clustering. The greater the similarity, the better the flexible load partitioning, which facilitates assigning the same power upper limit to flexible loads of the same type and in similar locations, thereby improving the source-load matching of multiple flexible loads during subsequent coordinated control.

[0050] In this embodiment, the similarity between membership vectors is cosine similarity. The calculation of cosine similarity is a well-known technique and will not be described in detail here. As other implementation methods, based on the ability to measure the similarity between components in membership vectors, implementers may use other existing techniques, such as the reciprocal of Euclidean distance, etc. This application does not impose any special restrictions.

[0051] In each cycle, a predetermined number of initial parameter vectors are used as input to the particle swarm optimization algorithm. The parameter vectors are iteratively optimized, and the optimal parameter vector and the corresponding optimal clusters are output. The particle swarm optimization algorithm is a well-known technique and will not be described further in this application.

[0052] In this embodiment, the maximum number of iterations of the particle swarm optimization algorithm is set to 30. The value of the maximum number of iterations is preset by the user and can be set by the implementer according to the actual situation. This application does not impose any special restrictions.

[0053] Step 3: Based on the optimal classification results, obtain the electricity demand of all flexible loads within each category in each cycle in each same time period by using the distribution feature array of all loads in each category in each cycle in each same time period. Combined with the power data of each distributed power source in each cycle and the next cycle in each same time period by comparing the power data of each distributed power source in each cycle in each cycle in each same time period, obtain the output feature value of each distributed power source in each day in each time period in the next cycle in each cycle, and obtain the power upper limit of each load in each day in each time period in the next cycle in each cycle.

[0054] The global average reflects the electricity consumption of flexible loads in each time period, while the power demand characteristic value reflects the power demand of flexible loads in each time period.

[0055] For each optimal cluster obtained by dividing all flexible loads in each period, the mean of the global mean and the mean of the power demand characteristic value of all flexible loads in each cluster for all days in each period are recorded as the first mean and the second mean, respectively. The product of the first mean and the second mean is taken as the power demand of all flexible loads in each cluster for all days in each period for all days in each period, reflecting the power demand of flexible loads in each cluster in each period.

[0056] Considering the power variations of distributed power sources and the differences in the electricity demand of flexible loads within different clusters at different times, the upper limit of the power of flexible loads within different clusters is adjusted.

[0057] Based on the above analysis, the average power data of each distributed power source in the j-th time period of all days in each cycle is calculated. The proportion of the number of distributed power sources with power data greater than the average value in the j-th time period of each day in the next cycle is used as the output characteristic value of each distributed power source in the j-th time period of each day in the next cycle, reflecting the output level of the distributed power source.

[0058] The power variation of flexible loads exhibits a certain degree of continuity. By analyzing the electricity demand of all flexible loads within each cluster at the same time of day in each cycle, and combining this with the output characteristics of each distributed power source at each time of day in the next cycle, the upper limit of power for each flexible load in the next time of day in the next cycle can be obtained. The expression is as follows: In the formula, This represents the upper limit of the power of the i-th load during the j+1th time period on day f within the T+1th cycle; This represents the maximum power value of the last sub-period within the j-th time period of the f-th day in the T+1-th cycle; , These represent the electricity demand of the i-th load during all days of the T-th cycle in the (j+1)-th and j-th time periods, respectively. This represents the output characteristic value of the distributed power source closest to the i-th load during the j+1th time period on day f within the (T+1)-th cycle. Here, "load" is an abbreviation for flexible load.

[0059] It should be noted that the higher the output of distributed power sources and the higher the electricity demand of flexible loads, the higher the power limit should be set to improve the source-load matching of the distribution network and reduce energy loss. A schematic diagram of the power limit adjustment process is shown below. Figure 2 As shown.

[0060] Furthermore, the power of each flexible load within any urban area is controlled to be below its own power limit.

[0061] Based on the same inventive concept as the above methods, this application also provides a distribution network coordination control system based on multiple flexible loads, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described distribution network coordination control methods based on multiple flexible loads.

[0062] In summary, this application analyzes the power consumption and power demand of flexible loads by analyzing the distribution of their power data, analyzes the fluctuation of their power data by analyzing the dispersion of their power data, analyzes the periodic variation of flexible loads by combining the differences in power consumption of flexible loads at the same time on different days in the previous cycle, and considers the locational relationship between flexible loads and distributed power sources. This approach can more accurately describe the characteristics of flexible loads, provide a basis for subsequent classification, and help to formulate differentiated power adjustment strategies for flexible loads of different types and locations, thereby improving the coordinated control capability of the distribution network. The system filters components in the feature vector using preset screening parameters and adjusts these parameters through an intelligent optimization algorithm to obtain the optimal classification result. This effectively removes redundant components from the feature vector while retaining features that are important for classification, thereby improving the accuracy of flexible load classification. Based on the optimal classification result and the power demand of flexible loads within each type of load, as well as the output characteristics of distributed power sources, the system obtains the power upper limit for each load in each time period of the next cycle. This allows for the development of reasonable power regulation strategies for each flexible load based on its power demand and the output of distributed power sources, avoiding the problem of setting the power upper limit too high or too low. This ensures normal user operation while improving the source-load matching degree of the distribution network and reducing energy loss.

[0063] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. In the descriptions corresponding to the flowcharts and block diagrams in the accompanying drawings, the operations or steps corresponding to different blocks may also occur in a different order than disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. Each block in a block diagram and / or flowchart, and combinations of blocks in a block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0064] It will be apparent to those skilled in the art that this application is not limited to the details of the exemplary embodiments described above, and that this application can be implemented in other specific forms without departing from its essential characteristics. Therefore, the embodiments described above should be considered exemplary and non-limiting in all respects.

Claims

1. A distribution network coordinated control method based on multiple flexible loads, characterized in that, The method includes the following steps: Real-time collection of power data of each flexible load and each distributed power source in the distribution network of any urban area; the flexible load is referred to as load. The day is divided into time periods. By analyzing the distribution and dispersion of power data for each load in each time period, the distribution feature array and power fluctuation of each load in each time period are obtained. Then, by comparing the power data of each load in the same time period of the previous cycle, the reference electricity consumption difference factor of each load in each time period of each day in each cycle is obtained, as well as the distance of each load to each distributed power source access point, the feature vector of each load in each cycle is obtained. Preset filtering parameters are used to filter the components in the feature vector, thereby classifying all loads in each cycle. An intelligent optimization algorithm is used to adjust the filtering parameters to obtain the optimal classification results of all loads in each cycle. The filtering parameters include the filtering ratio of the components in the feature vector and the local maximum retention length. Based on the optimal classification results, the power demand of all flexible loads within each category is obtained by using the distribution feature array of all loads within each category in the same time period of each day in each cycle. Combined with the power data of each distributed power source in each cycle and the next cycle in the same time period, the output feature value of each distributed power source in each time period of each day in the next cycle is obtained, and the power upper limit of each load in the next time period of each day in the next cycle is obtained.

2. The distribution network coordinated control method based on multiple flexible loads as described in claim 1, characterized in that, The process of obtaining the distribution feature array is as follows: The average power data of each load in each time period is used as the global average of each load in each time period; Each time period is divided into sub-time periods. The average power data of each load in each sub-time period is taken as the local average of each load in each sub-time period. For each load in each day of each cycle, the proportion of sub-time periods in which the local average is greater than the global average is calculated among all sub-time periods. The distribution feature array consists of the global mean and the quantity percentage.

3. The distribution network coordinated control method based on multiple flexible loads as described in claim 2, characterized in that, The process of obtaining the power fluctuation is as follows: Calculate the dispersion of power data for each load in each sub-period. For each load in each period of each day within each cycle, take the mean of the dispersion of each load in all sub-periods as the power fluctuation.

4. The distribution network coordinated control method based on multiple flexible loads as described in claim 1, characterized in that, The process for obtaining the reference electricity consumption difference factor is as follows: The power data of each load in each time period are arranged in chronological order to form the power sequence of each load in each time period; for each load in the j-th time period of all days in the previous cycle, the average distance between any two time periods is calculated as the reference power consumption difference factor of each load in the j-th time period of each day in each cycle.

5. The distribution network coordinated control method based on multiple flexible loads as described in claim 1, characterized in that, The process of obtaining the feature vector is as follows: The distribution characteristics of each load in each time period, power fluctuation and reference electricity consumption difference factor are used to form the time period vector of each load in each time period; the time period vectors of each load in all days of each cycle are sorted together according to time sequence, and the time period vectors are connected end to end in sequence to form the time vector of each load in each cycle. The distances of each load to all distributed power supply access points within each cycle are used to form the spatial vector of each load within each cycle. The feature vector is composed of the time vector and the spatial vector.

6. The distribution network coordinated control method based on multiple flexible loads as described in claim 1, characterized in that, In the process of obtaining the optimal classification result, the method for obtaining the fitness value in the intelligent optimization algorithm is as follows: Based on the screening results of the feature vectors of all loads in each period, a clustering algorithm is used to classify all loads in each period. For each classification, the membership degree between each load and the center of each class is obtained. For each class, the sum of the similarity of the membership degrees between any two loads is calculated. The cumulative value of the sum of the values ​​of the membership degrees of all classes obtained in a single classification is used as the fitness value of the screening parameter corresponding to the single classification.

7. The distribution network coordinated control method based on multiple flexible loads as described in claim 1, characterized in that, The process of obtaining the electricity demand is as follows: For each category in the optimal classification result, the mean of the first component and the mean of the second component in the distribution characteristic array of all loads in each category for all days in each cycle at the same time period are denoted as the first mean and the second mean, respectively. The electricity demand is the product of the first mean and the second mean.

8. The distribution network coordinated control method based on multiple flexible loads as described in claim 1, characterized in that, The process for obtaining the output characteristic value is as follows: Calculate the average power data of each distributed power source in the j-th time period of all days in each cycle. Then, take the percentage of the power data of each distributed power source in the j-th time period of each day in the next cycle that is greater than the average value as the output characteristic value of each distributed power source in the j-th time period of each day in the next cycle.

9. The distribution network coordinated control method based on multiple flexible loads as described in claim 2, characterized in that, The expression for the upper limit of power is: ; In the formula, This represents the upper limit of the power of the i-th load during the j+1th time period on day f within the T+1th cycle; This represents the maximum power value of the last sub-period within the j-th time period of the f-th day in the T+1-th cycle; , These represent the electricity demand of the i-th load during all days of the T-th cycle in the (j+1)-th and j-th time periods, respectively. This represents the output characteristic value of the distributed power source that is closest to the i-th load during the j+1th time period on day f within the (T+1)-th cycle.

10. A distribution network coordinated control system based on multiple flexible loads, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the distribution network coordinated control method based on multiple flexible loads as described in any one of claims 1-9.

Images