Optimization Method and System for Energy Consumption Load Management in Chain Gardens Based on Clustering Algorithm

Abstract

The present invention discloses an optimization method and system for the energy consumption load management of chain parks based on a clustering algorithm, which relates to the technical field of load management optimization. The method includes collecting data of multiple chain parks, constructing a load characteristic vector through a multi-source feature fusion technology; based on the load similarity and geographical distribution characteristics among chain parks, using a graph structure model to model the energy interaction and correlation characteristics among chain park nodes, forming a collaborative load network; aiming at the time-dependent characteristics of the energy consumption load of chain parks, segmenting the load curve according to time sequence and introducing a time weighting coefficient for clustering; designing a multi-objective collaborative optimization clustering algorithm, combining load clustering with optimization objectives to solve the optimal clustering scheme under different scenarios; and generating a distributed scheduling strategy adaptable to scenario changes according to the clustering results and the load collaboration relationship among chain parks. The present invention realizes energy sharing and load balancing among chain parks, and greatly improves the economy and environmental benefits of the regional energy system.

Description

Technical Field

[0001] The present invention relates to the technical field of load management optimization, and in particular to a method and system for optimizing energy load management in a chain park based on a clustering algorithm. Background Art

[0002] In recent years, with the rapid growth of global energy demand and the continuous development of distributed energy technology, campus energy management has gradually become an important research direction for smart grids and energy Internet. In a chain park (a complex energy consumption environment where multiple parks collaborate), due to the significant differences in load characteristics, equipment status and environmental parameters of different parks, the need for efficient management and optimized scheduling of energy loads has become more urgent. Traditional campus energy management methods usually rely on static load forecasting and independent optimization strategies for a single park. These methods have significant limitations in dealing with complex energy interactions between chain parks. For example, the load management method of a single park cannot fully utilize the potential for surplus power sharing and coordinated regulation between parks. At the same time, traditional load forecasting methods have large errors when dealing with nonlinear load fluctuations and complex timing characteristics.

[0003] In addition, since park load scheduling is usually affected by multiple factors such as geographical distribution, energy circulation restrictions, and time-dependency, there are still obvious research gaps in existing technologies on how to build cross-park energy coordination networks and design optimization strategies that adapt to multi-scenario load requirements. Summary of the invention

[0004] In view of the problems existing in the existing chain park energy load management technology in dealing with complex energy interactions and time-dependent load optimization, a chain park energy load management optimization method and system based on a clustering algorithm are proposed in the present invention.

[0005] Therefore, the problem to be solved by the present invention is how to generate an accurate and efficient distributed scheduling strategy.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, the present invention provides an optimization method for the energy consumption load management of a chain park based on a clustering algorithm, which includes collecting real-time load data, equipment status information, and environmental parameters of multiple chain parks, and constructing a load characteristic vector through a multi-source feature fusion technology; based on the load similarity and geographical distribution characteristics among the chain parks, using a graph structure model to model the energy interaction and correlation characteristics among the chain park nodes, forming a collaborative load network; aiming at the time-dependent characteristics of the energy consumption load of the chain park, adopting a dynamic time segmentation technology, segmenting the load curve according to time sequence and introducing a time weighting coefficient for clustering; designing a multi-objective collaborative optimization clustering algorithm, combining load clustering with optimization objectives to form a comprehensive optimization function, and using a hybrid optimization algorithm to solve the optimal clustering scheme under different scenarios; according to the clustering results and the load collaboration relationship among the chain parks, defining a distributed scheduling priority rule, and combining with the scheduling priority rule to generate a distributed scheduling strategy adaptable to scenario changes, realizing the sharing of surplus electricity among the chain parks, demand response regulation, and load peak-valley balancing.

[0008] As a preferred solution of the optimization method for the energy consumption load management of the chain park based on the clustering algorithm of the present invention, wherein: constructing the load characteristic vector through the multi-source feature fusion technology includes: adopting a feature splicing method to fuse the load characteristics, equipment status characteristics, and environmental characteristics into a unified characteristic vector:

[0009] ,

[0010] wherein, is the load characteristic, is the equipment status characteristic, is the environmental characteristic; applying a dimensionality reduction algorithm to reduce the dimension of the high-dimensional characteristic vector and remove redundant information; constructing the completed load characteristic vector matrix:

[0011] ,

[0012] wherein, m is the number of chain parks.

[0013] As a preferred solution of the optimization method for the energy consumption load management of the chain park based on the clustering algorithm of the present invention, wherein: modeling the energy interaction and correlation characteristics among the chain park nodes by using the graph structure model includes the following steps: constructing a set of chain park data nodes according to the load characteristic vector, where each data node represents the load characteristics of a single chain park; based on the load characteristic vector of the data nodes, using the dynamic weighted cosine similarity to measure the load similarity among the chain parks; using the geographical location information of the chain parks to construct a geographical distance matrix, and the matrix elements represent the geographical distance between any two chain parks, calculated by using the Haversine formula; combining the load similarity and the geographical distance, and adopting a graph structure-based modeling method to construct a chain park collaborative load network, where the nodes represent the chain park data nodes; the weight of the edge is jointly determined by the load similarity and the geographical distance.

[0014] As a preferred embodiment of the method for optimizing the energy consumption load management of the chain park based on the clustering algorithm of the present invention, the formula for measuring the load similarity between chain parks using dynamic weighted cosine similarity is as follows:

[0015] ,

[0016] wherein, and are the i-th dimensional load characteristics of chain park A and chain park B respectively, is the weight of the corresponding characteristic, dynamically adjusted by the importance of the characteristic, and n is the dimension number of the load characteristic vector.

[0017] As a preferred embodiment of the method for optimizing the energy consumption load management of the chain park based on the clustering algorithm of the present invention, the dynamic time segmentation technology is adopted to segment the load curve according to time sequence and introduce a time weighting coefficient for clustering, which includes the following steps: combining the load characteristic vector of the chain park and the real-time load curve, dividing the daily load data according to the time granularity; assigning a time weighting coefficient to each time segment to highlight the influence of key periods, and for the load data within each time segment, extracting the load characteristic vector; using the extracted segmented load characteristic vectors to construct an input matrix for dynamic clustering:

[0018] ,

[0019] wherein, the rows of the matrix represent different time segments; the columns of the matrix represent the dimensions of the segmented load characteristics, ensuring the integrity of the input data for dynamic clustering; based on the input matrix perform dynamic clustering using the time series clustering algorithm.

[0020] As a preferred embodiment of the method for optimizing the energy consumption load management of the chain park based on the clustering algorithm of the present invention, the design of the multi-objective collaborative optimization clustering algorithm combines load clustering with optimization objectives, which includes the following steps: based on the output result of dynamic clustering, define the objective function of multi-objective optimization clustering; perform weighted integration on the objective function to form a comprehensive optimization function; use a hybrid optimization algorithm to solve the objective function; according to the optimized clustering result, perform dynamic adjustment for different energy consumption scenarios.

[0021] As a preferred embodiment of the method for optimizing the energy consumption load management of the chain park based on the clustering algorithm of the present invention, the optimization objectives include the load balance objective, the collaborative correlation objective, and the load fluctuation adaptability objective.

[0022] In a second aspect, the present invention provides an optimized system for load management of chain parks based on a clustering algorithm, which includes a data collection and fusion module for collecting real-time load data, equipment status information, and environmental parameters of multiple chain parks, and constructing a load characteristic vector through multi-source feature fusion technology; a collaborative network modeling module for modeling the energy interaction and correlation characteristics between chain park nodes using a graph structure model based on the load similarity and geographical distribution characteristics between chain parks to form a collaborative load network; a time segmentation and clustering module for segmenting the load curve according to time sequence and introducing a time weighting coefficient for clustering for the time-dependent characteristics of the energy consumption load of chain parks by using dynamic time segmentation technology; an optimized clustering calculation module for designing a multi-objective collaborative optimization clustering algorithm to combine load clustering with optimization objectives and solve the optimal clustering scheme under different scenarios; and a scheduling strategy generation module for generating a distributed load scheduling strategy according to the clustering results and the load collaboration relationship between chain parks to achieve surplus power sharing, demand response regulation, and load peak-valley balancing between chain parks.

[0023] In a third aspect, the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program instructions are executed by the processor, the steps of the optimized method for load management of chain parks based on a clustering algorithm as described in the first aspect of the present invention are implemented.

[0024] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program instructions are executed by the processor, the steps of the optimized method for load management of chain parks based on a clustering algorithm as described in the first aspect of the present invention are implemented.

[0025] The beneficial effects of the present invention are as follows: The present invention integrates load characteristics, equipment status, and environmental impacts, realizing the deep integration of data and models; constructs dynamic clustering adapted to time dependence and scenario-adaptive optimization strategies, enhancing the system's adaptability to complex scenarios; and realizes energy sharing and load balancing between chain parks, greatly improving the economic and environmental benefits of the regional energy system. Description of the Drawings

[0026] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for description in the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0027] Figure 1 It is a flowchart of the optimized method for load management of chain parks based on a clustering algorithm.

[0028] Figure 2It is a structural diagram of an energy load management optimization system for a chain park based on a clustering algorithm. Detailed implementation manners

[0029] To make the above objects, features and advantages of the present invention more obvious and understandable, the following will describe in detail the specific implementation manners of the present invention with reference to the accompanying drawings of the specification.

[0030] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0031] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure or characteristic that may be included in at least one implementation manner of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments.

[0032] Embodiment 1

[0033] Referring to Figures 1 - 2 , this is the first embodiment of the present invention. This embodiment provides an energy load management optimization method for a chain park based on a clustering algorithm. As Figure 1 shown, it includes

[0034] S1: Collect real-time load data, equipment status information, and environmental parameters (such as temperature, humidity, wind speed, light, etc.) of multiple chain parks, and construct a load characteristic vector through multi-source feature fusion technology.

[0035] S1.1: Install smart meters or sensors at the power distribution nodes and main electrical equipment of the chain park to collect real-time load data such as power consumption, voltage, and current. The data sampling frequency is set to per minute or per hour, specifically selected according to the load fluctuation frequency. Use the Industrial Internet of Things (IIoT) platform to centrally collect data and upload it to the central database in real time.

[0036] Collect the switch status, running time, efficiency, and energy consumption data of key equipment in the chain park (such as air conditioners, production line equipment); install operation status monitoring sensors on the equipment, or collect data through the built-in management system of the equipment (such as PLC control system).

[0037] Arrange environmental monitoring sensors in the chain park to collect external environmental data such as temperature, humidity, wind speed, and light intensity; existing weather station data can be used as a supplement, especially the wide-area weather data covering multiple chain park areas.

[0038] S1.2: Preprocess the collected data, including filling in missing data, denoising and smoothing the data, and feature standardization.

[0039] S1.3: Extract basic features from the real-time load data, including peak value, valley value, average value, load factor, etc.; extract time-series features: the periodic pattern of the load data (extracting spectral features based on Fourier transform) and volatility (based on statistical indicators such as standard deviation and skewness).

[0040] Construct a device characteristic vector using parameters such as the device working state duration, start-stop frequency, and energy consumption efficiency, and establish a device-load mapping relationship by combining the device operation state and the corresponding load.

[0041] By analyzing the correlation between temperature, light and load changes, determine the impact degree of the external environment on the energy consumption load; and introduce a meteorological time window to convert the hourly changing environmental parameters into short-term average characteristics (such as 3-hour moving average).

[0042] S1.4: Adopt a feature splicing method to fuse the load characteristics, device state characteristics and environmental characteristics into a unified characteristic vector:

[0043] ,

[0044] where, is the load characteristic, is the device state characteristic, is the environmental characteristic.

[0045] Furthermore, apply a dimensionality reduction algorithm (such as principal component analysis PCA) to reduce the dimension of the high-dimensional characteristic vector, remove redundant information, and retain the main features.

[0046] Construct the completed load characteristic vector matrix:

[0047] ,

[0048] where, m is the number of chain gardens.

[0049] S2: Based on the load similarity and geographical distribution characteristics between chain gardens, use a graph structure model (Graph-Based Model) to model the energy interaction and association characteristics between chain garden nodes, and form a collaborative load network.

[0050] S2.1: According to the load characteristic vector generated in step S1, construct a set of chain garden data nodes, where each data node represents the load characteristics of a single chain garden, including real-time load parameters, device state characteristics, and environmental parameter characteristics. Among them, the set of data nodes constitutes the basis for collaborative modeling.

[0051] S2.2: Based on the load characteristic vectors of data nodes, the dynamic weighted cosine similarity is used to measure the load similarity between chain gardens.

[0052] The specific calculation formula is as follows:

[0053] ,

[0054] where, and are the i-th dimensional load characteristics of chain garden A and chain garden B respectively, is the weight of the corresponding characteristic, dynamically adjusted by the importance of the characteristic, and n is the dimension number of the load characteristic vector.

[0055] S2.3: Using the geographical location information of chain gardens, a geographical distance matrix is constructed. The matrix elements represent the geographical distance between any two chain gardens, and the Haversine formula is used for calculation:

[0056] ,

[0057] where, and are the latitudes of the two chain gardens, is the longitude, and are the differences in latitude and longitude respectively, and r is the radius of the earth.

[0058] It can be seen that by constructing the geographical distance matrix, the physical positions between chain gardens are associated with the load similarity, providing spatial constraint conditions for subsequent network modeling.

[0059] S2.4: Combining the load similarity and geographical distance, a graph structure-based modeling method is adopted to construct a chain garden collaborative load network, where the nodes represent the chain garden data nodes; the weight of the edge is jointly determined by the load similarity and geographical distance, and the weight calculation formula of the edge is:

[0060] ,

[0061] where, is the weight between chain gardens A and B, and are the weighted coefficients of similarity and geographical distance, is the geographical distance between two chain gardens A and B.

[0062] S3: Aiming at the time-dependent characteristics of the energy consumption load of chain gardens, a dynamic time segmentation technology is adopted to segment the load curve according to time sequence and introduce a time weighting coefficient for clustering to adapt to load fluctuations and seasonal changes.

[0063] Combine the load characteristic vector of the chain garden and the real-time load curve, divide the daily load data according to the time granularity, and set the segmentation rules:

[0064] Define the basic segmentation period (such as 1 hour or 30 minutes) to ensure sufficient refinement of the load change characteristics.

[0065] Introduce a dynamic segmentation factor to adjust the segmentation length according to the load fluctuation amplitude. When the load fluctuates violently, shorten the segmentation period, and when the fluctuation is stable, extend the segmentation period.

[0066] Furthermore, assign a time weighting coefficient to each time segment to highlight the influence of key periods, and for the load data within each time segment, extract the load characteristic vector, including:

[0067] Statistical characteristics: the average load within the segment, the fluctuation range, the load growth rate, etc.;

[0068] Temporal characteristics: the load change trend (represented by the slope of linear fitting);

[0069] Correlation characteristics: the correlation coefficient between the load and environmental parameters (such as temperature, humidity).

[0070] Use the extracted segmented load characteristic vectors to construct an input matrix for dynamic clustering:

[0071] ,

[0072] Among them, the rows of the matrix represent different time segments; the columns of the matrix represent the dimensions of the segmented load characteristics, ensuring the integrity of the input data for dynamic clustering.

[0073] Even further, based on the input matrix Adopt a time series clustering algorithm (such as K-Means based on Dynamic Time Warping DTW) to perform dynamic clustering:

[0074] Use DTW to calculate the similarity between time series to adapt to the situation of inconsistent time segment lengths; the clustering outputs multiple clusters, and each cluster contains time segments with similar time-dependent load characteristics.

[0075] S4: Design a multi-objective collaborative optimization clustering algorithm, combine load clustering with optimization objectives, and solve the optimal clustering scheme under different scenarios.

[0076] S4.1: Based on the output results of the dynamic clustering in step S3, define the objective function of multi-objective optimization clustering, and comprehensively consider the following optimization objectives:

[0077] Load balance objective: Maximize the uniform distribution of the load between different clusters, defined as:

[0078] ,

[0079] where k is the number of clustering clusters, is the global load mean value, is the average load value of the i-th clustering cluster.

[0080] Collaborative relevance objective: Maximize the collaborative load network weight between chain gardens within a clustering cluster, defined as:

[0081] ,

[0082] where, is the i-th clustering cluster.

[0083] Load fluctuation adaptability objective: Minimize the variance of the segmented load characteristics within each clustering cluster, defined as:

[0084] ,

[0085] where, is the variance of the time-segmented load characteristic matrix.

[0086] The present invention improves and optimizes the overall applicability of the clustering result by introducing multi-objective trade-offs of load balance, collaborative relevance, and fluctuation adaptability.

[0087] S4.2: Weight and integrate the objective function in step S4.1 to form a comprehensive optimization function:

[0088] ,

[0089] where, ~ are the objective weight coefficients, which are dynamically adjusted according to the actual scenario requirements (such as increasing the load balance weight during peak hours).

[0090] S4.3: Use a hybrid optimization algorithm to solve the objective function , and the specific process is as follows:

[0091] Use the dynamic clustering result in step S3 as the initial solution set, and each solution represents the clustering scheme of the chain garden;

[0092] Combine the particle swarm optimization (PSO) and genetic algorithm (GA) to iteratively optimize the objective function:

[0093] The particle swarm optimization improves the global search ability and ensures the rapid convergence of the objective function;

[0094] The crossover and mutation operations of the genetic algorithm enhance the local optimization ability and improve the clustering quality;

[0095] When the change in the objective function value is less than the set threshold or the maximum number of iterations is reached, stop the optimization and output the optimal clustering solution.

[0096] S4.4: According to the optimized clustering results, perform dynamic adjustment for different energy consumption scenarios (such as peak load, valley load, seasonal demand, etc.):

[0097] Identify the scenario type (such as judging whether it is the peak period currently by the change rate of load characteristics);

[0098] Adjust the target weight coefficient according to the scenario ~ and re - execute the optimization solution process, and output the optimal clustering solution that meets the current scenario requirements.

[0099] S5: According to the clustering results and the load coordination relationship between the industrial parks, generate a distributed load scheduling strategy to achieve surplus power sharing, demand response regulation, and load peak - valley balancing between the industrial parks.

[0100] Specifically, according to the clustering results optimized in S4, extract the load characteristic vectors and time - segmented characteristics of each clustering cluster, and combine historical data to analyze the changing trend of the energy consumption demand of each industrial park; introduce the time - sliding window method to predict the electricity demand and available surplus electricity of each industrial park in the next time period.

[0101] Based on the collaborative load network constructed in S2, extract the strong - weak relationship and priority of energy interaction between the industrial parks, establish a resource scheduling matrix between the industrial parks, and conduct a hierarchical analysis of the coordination relationship to identify key node industrial parks (industrial parks with high power supply capacity) and weakly - associated industrial parks (industrial parks with high demand), providing a basis for distributed scheduling.

[0102] Define the distributed scheduling priority rules according to the clustering results and the load coordination relationship between the industrial parks:

[0103] Give priority to transferring surplus power to industrial parks with strong coordination relationships to reduce energy loss;

[0104] During the peak load period, give priority to coordinating key node industrial parks to participate in the response to smooth the peak load;

[0105] During the non - peak period, disperse the scheduling of surplus power to optimize the overall load balance of the industrial parks.

[0106] Furthermore, combined with the scheduling priority rules, generate a distributed scheduling strategy adapted to scenario changes, for example:

[0107] Peak scenario: Initiate surplus power sharing between high - priority industrial parks to balance the regional load and avoid overload;

[0108] Valley scenario: Through the distribution of surplus power between industrial parks, fill the low - load areas and reduce energy waste;

[0109] Emergency scenario: In case of equipment failure or extreme weather, etc., quickly reallocate resources to maintain the load balance among the chain parks.

[0110] Furthermore, this embodiment also provides an optimized system for load management of chain park energy consumption based on the clustering algorithm, including

[0111] A data collection and fusion module, which is used to collect real-time load data, equipment status information, and environmental parameters of multiple chain parks, and construct a load characteristic vector through multi-source feature fusion technology;

[0112] A collaborative network modeling module, which is used to model the energy interaction and correlation characteristics between chain park nodes by using a graph structure model based on the load similarity and geographical distribution characteristics among the chain parks, and form a collaborative load network;

[0113] A time segmentation and clustering module, which is used to adopt dynamic time segmentation technology for the time-dependent characteristics of chain park energy consumption load, segment the load curve according to time sequence, and introduce a time weighting coefficient for clustering;

[0114] An optimized clustering calculation module, which is used to design a multi-objective collaborative optimization clustering algorithm, combine load clustering with optimization objectives, and solve the optimal clustering scheme under different scenarios;

[0115] A scheduling strategy generation module, which is used to generate a distributed load scheduling strategy according to the clustering results and the load collaboration relationship among the chain parks, and realize the sharing of surplus electricity, demand response regulation, and load peak-valley balance among the chain parks.

[0116] This embodiment also provides a computer device, which is applicable to the situation of the optimized method for load management of chain park energy consumption based on the clustering algorithm, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the optimized method for load management of chain park energy consumption based on the clustering algorithm as proposed in the above embodiment.

[0117] The computer device can be a terminal, which includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used to communicate with external terminals in a wired or wireless manner, and the wireless manner can be implemented through WIFI, carrier network, NFC (Near Field Communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or buttons, trackballs or touchpads provided on the outer shell of the computer device, or an external keyboard, touchpad or mouse, etc.

[0118] This embodiment also provides a storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the method for optimizing the energy consumption load management of the chain park based on the clustering algorithm as proposed in the above embodiment.

[0119] In summary, the present invention integrates load characteristics, equipment status, and environmental impacts, realizing the deep integration of data and models; constructs a dynamic clustering adapted to time dependence and an optimization strategy adapted to scenarios, improving the system's adaptability to complex scenarios; realizes energy sharing and load balancing among chain parks, greatly improving the economy and environmental benefits of the regional energy system.

[0120] Embodiment 2

[0121] Referring to Table 1, this is the second embodiment of the present invention. This embodiment provides a method for optimizing the energy consumption load management of the chain park based on the clustering algorithm. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0122] In this experiment, 6 typical chain parks in a certain area are selected as the test objects, and the test time is 7 consecutive days, recording relevant data continuously for 24 hours every day. The 6 chain parks cover different types of energy consumption scenarios such as manufacturing, warehousing and logistics, agricultural greenhouses, and commercial parks. In the experiment, the real-time load data, equipment status information, and environmental parameters of each chain park are collected, and optimization analysis is carried out in combination with the method of the present invention.

[0123] The test preparations include:

[0124] Install smart meters at the main power distribution nodes and key electrical equipment (such as air conditioners, production line equipment) of each chain park to collect load data, and the data sampling frequency is set to 1 minute;

[0125] Deploy environmental monitoring sensors to record data such as temperature, humidity, wind speed, and light intensity, and supplement the wide-area weather data of the meteorological station;

[0126] Upload the collected data to the central database through the Industrial Internet of Things (IIoT) platform;

[0127] For data preprocessing, the interpolation method is used to fill in the missing values, the wavelet denoising method is used to remove noise, and at the same time, the eigenvalues are standardized.

[0128] Calculate the load characteristics of each chain garden, including statistical characteristics such as peak value, valley value, and average value, as well as volatility and periodicity characteristics; extract the equipment characteristic vector in combination with the equipment operation status data; analyze the impact of environmental parameters on the load and determine the short-term average characteristics through the meteorological time window, and apply principal component analysis (PCA) to the above characteristic vectors to remove redundant information and obtain the reduced-dimensional load characteristic vector matrix.

[0129] Calculate the load similarity between chain gardens, use the dynamic weighted cosine similarity to conduct similarity analysis on the characteristic vectors, use the Haversine formula to construct the geographical distance matrix based on the geographical location of the chain gardens, and comprehensively construct the collaborative load network by combining the load similarity and the geographical distance.

[0130] Dynamically segment the load time series of the chain garden, and dynamically adjust the segmentation period according to the load volatility (shorten to 15 minutes when the load fluctuates violently and extend to 1 hour when the fluctuation is stable);

[0131] Extract the statistical and time series characteristics of each segment of the load, construct the dynamic clustering input matrix, and use the DTW-based K-Means algorithm for clustering to obtain the time segments with similar load characteristics for each chain garden.

[0132] Combining the three optimization objectives of balance, collaborative relevance, and fluctuation adaptability, optimize the clustering scheme through the PSO-GA hybrid algorithm, and output the optimal clustering results under different load scenarios.

[0133] According to the optimized clustering results, generate a distributed load scheduling strategy, give priority to realizing the sharing of surplus power and load balance between chain gardens, and adjust the scheduling priority for peak load and emergency scenarios respectively. Some experimental data are as follows:

[0134] Table 1 Experimental data of energy consumption optimization for chain gardens based on clustering algorithm

[0135] Parameter Name Chain Park A - Manufacturing Chain Park B - Warehousing and Logistics Chain Park C - Agricultural Greenhouse Chain Park D - Commercial Park Chain Park E - Comprehensive Park Chain Park F - Science and Technology Park Peak Load (kW) 850 620 400 720 780 600 Average Load (kW) 550 480 320 500 520 450 Fluctuation Coefficient (%) 35.6 28.7 25.4 31.5 33.2 29.8 Similarity Weight 0.92 0.89 0.85 0.87 0.90 0.88 Geographical Distance (km) 2.5 5.3 8.2 3.8 4.1 6.7 Scheduling Priority High Medium Medium High Medium Low Energy Saving Rate after Optimization (%) 18.2 15.7 12.9 16.3 17.8 14.5

[0136] It can be seen from the above data that the load management optimization method of the present invention significantly improves the energy utilization efficiency of the chain garden and realizes collaborative scheduling:

[0137] The tabular data shows that through clustering and optimized scheduling, the load balance of each chain garden has been improved. For example, the peak load of Chain Garden A was as high as 850 kW before optimization, while the peak load decreased significantly after optimization, and the energy-saving rate reached 18.2%. This indicates that the optimization method can effectively smooth the load curve and reduce the pressure on the power grid caused by peak loads.

[0138] For chain gardens with relatively high fluctuation coefficients (such as Chain Garden A and Chain Garden E), through dynamic segmentation and collaborative network modeling, their load fluctuations have been significantly alleviated, and the fluctuation coefficient has dropped to a reasonable range (about 30%). At the same time, the load similarity weights based on dynamic weighted cosine similarity are mostly higher than 0.85, indicating that the clustering algorithm can accurately capture the collaborative relationships between chain gardens.

[0139] The significant increase in the energy-saving rate after optimization further verifies the innovation of the present invention. The energy-saving rate of each chain garden has reached more than 12%, with the highest being 18.2%, showing that through the comprehensive effects of load balancing, surplus power sharing, and demand response regulation, the overall energy efficiency has been effectively improved.

[0140] Compared with traditional independent load management methods, the present invention solves the problem of difficulty in balancing the dispersibility and dynamics in load management through feature vector fusion, dynamic clustering analysis, and multi-objective optimization. At the same time, the combination of the PSO-GA hybrid optimization algorithm significantly improves the global optimization efficiency.

[0141] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A chain park energy load management optimization method based on clustering algorithm, characterized by: include: Collect real-time load data, equipment status information, and environmental parameters of multiple chain parks, and construct load characteristic vectors through multi-source feature fusion technology; Based on the load similarity and geographical distribution characteristics between chain parks, the graph structure model is used to model the energy interaction and correlation characteristics between chain park nodes to form a collaborative load network; In view of the time-dependent characteristics of the energy load in the chain park, the dynamic time segmentation technology is used to segment the load curve according to the time sequence and introduce the time weighting coefficient for clustering; Design a multi-objective collaborative optimization clustering algorithm, combine load clustering with optimization objectives to form a comprehensive optimization function, and use a hybrid optimization algorithm to solve the optimal clustering solution in different scenarios; According to the clustering results and the load coordination relationship between the chain parks, the distributed scheduling priority rules are defined. Combined with the scheduling priority rules, a distributed scheduling strategy that adapts to scene changes is generated to achieve surplus power sharing, demand response regulation and load peak-valley balance among chain parks; Based on the load similarity and geographical distribution characteristics between the chain parks, the energy interaction and correlation characteristics between the chain park nodes are modeled using a graph structure model to form a collaborative load network, which specifically includes the following steps: S1: Based on the generated load characteristic vector, a chain park data node set is constructed, where each data node represents the load characteristics of a single chain park, including real-time load parameters, equipment status characteristics, and environmental parameter characteristics, wherein the data node set constitutes the basis for collaborative modeling; S2: Based on the load characteristic vector of the data node, the dynamic weighted cosine similarity is used to measure the load similarity between the chains; The specific calculation formula is: ， in, and are the i-th dimension load characteristics of chain park A and chain park B respectively, is the weight of the corresponding characteristic, which is dynamically adjusted by the importance of the characteristic, and n is the number of dimensions of the load characteristic vector; S3: Use the geographical location information of the chain parks to construct a geographical distance matrix. The matrix elements represent the geographical distance between any two chain parks, which is calculated using the Haversine formula: ， in, and is the latitude of the two chain parks, is the longitude, and are the latitude and longitude differences respectively, and r is the radius of the earth; By constructing a geographic distance matrix, the physical location between chain parks is associated with load similarity, providing spatial constraints for subsequent network modeling; S4: Combining load similarity and geographical distance, a graph-based modeling method is used to construct a chain park collaborative load network, where the nodes represent chain park data nodes; the edge weight is determined by load similarity and geographical distance, and the edge weight calculation formula is: ， in, is the weight between chain parks A and B, and is the weighting coefficient of similarity and geographical distance, is the geographical distance between the two chain parks A and B; The construction of the load characteristic vector by multi-source feature fusion technology includes: The feature splicing method is used to integrate the load characteristics, equipment status characteristics and environmental characteristics into a unified feature vector: ， in, is the load characteristic, is the device status characteristic, For environmental characteristics; Apply dimensionality reduction algorithms to reduce the dimensionality of high-dimensional feature vectors and remove redundant information; The constructed load characteristic vector matrix is: ， Wherein, m is the number of chain gardens; The method of adopting dynamic time segmentation technology to segment the load curve according to time sequence and introducing time weighting coefficient for clustering includes the following steps: Combine the load characteristic vector and real-time load curve of the chain park and divide the daily load data according to the time granularity; Assign a time weighting coefficient to each time segment to highlight the impact of the key period, and extract the load characteristic vector from the load data in each time segment; Using the extracted segment load characteristic vector, construct the input matrix for dynamic clustering: ， The rows of the matrix represent different time segments; the columns of the matrix represent the dimensions of the segment load characteristics, which ensures the integrity of the dynamic clustering input data; Based on the input matrix Dynamic clustering is performed using a time series clustering algorithm.

2. The method for optimizing chain park energy load management based on clustering algorithm according to claim 1, characterized in that: The design of a multi-objective collaborative optimization clustering algorithm to combine load clustering with optimization objectives includes the following steps: Based on the output results of dynamic clustering, define the objective function of multi-objective optimization clustering; Performing weighted integration on the objective function to form a comprehensive optimization function; A hybrid optimization algorithm is used to solve the objective function; According to the optimized clustering results, dynamic adjustments are made for different energy usage scenarios.

3. The method for optimizing chain park energy load management based on clustering algorithm according to claim 2, characterized in that: The optimization objectives include a load balancing objective, a collaborative correlation objective, and a load fluctuation adaptability objective.

4. A chain park energy load management optimization system based on a clustering algorithm, based on the chain park energy load management optimization method based on a clustering algorithm according to any one of claims 1 to 3, characterized in that: Also includes: Data acquisition and fusion module, which is used to collect real-time load data, equipment status information and environmental parameters of multiple chain parks, and construct load characteristic vectors through multi-source feature fusion technology; The collaborative network modeling module is used to model the energy interaction and correlation characteristics between the nodes of the chain parks based on the load similarity and geographical distribution characteristics between the chain parks, using the graph structure model to form a collaborative load network; The time segmentation and clustering module is used to target the time-dependent characteristics of the energy load in the chain park. It uses dynamic time segmentation technology to segment the load curve according to the time sequence and introduces time weighting coefficients for clustering. The optimization clustering calculation module is used to design a multi-objective collaborative optimization clustering algorithm, combining load clustering with optimization objectives to solve the optimal clustering solution in different scenarios; The scheduling strategy generation module is used to generate distributed load scheduling strategies based on the clustering results and the load coordination relationship between the chain parks, so as to realize the sharing of surplus power between the chain parks, demand response regulation and load peak-valley balance.

5. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the chain park energy load management optimization method based on clustering algorithm described in any one of claims 1 to 3 are implemented.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the chain park energy load management optimization method based on clustering algorithm described in any one of claims 1 to 3 are implemented.

Images