A method and system for energy optimization of a township micro-balance area based on edge computing, a storage medium and an electronic device

Abstract

The application relates to the technical field of energy management, and particularly discloses a method and system for energy optimization of a township micro-balanced area based on edge computing, a storage medium and electronic equipment, the method comprising the following steps: extracting fluctuation characteristics of local net load time series data sets in the township micro-balanced area to obtain a net load fluctuation characteristic vector; periodically exchanging the net load fluctuation characteristic vector as a node identifier with adjacent nodes, dynamically time warping the exchange result to obtain fluctuation characteristic similarity; determining same-group nodes according to the fluctuation characteristic similarity and integrating the same-group nodes into a virtual dynamic cluster; performing virtual power distribution in the virtual dynamic cluster to obtain final power adjustment amounts of the edge nodes; determining power set values for the next moment according to the final power adjustment amounts and actual values of the local net load at the current moment, and issuing and executing instructions; and the application improves the real-time performance and adaptability of energy management of the township micro-balanced area.

Description

Technical Field

[0001] This invention relates to the field of energy management technology, and in particular to a method, system, storage medium, and electronic device for optimizing energy consumption in rural micro-balance zones based on edge computing. Background Technology

[0002] In rural micro-balance zones, distributed photovoltaic power generation, energy storage systems, and adjustable loads are widely distributed. Each edge node collects local net load time-series data to achieve real-time power balancing and scheduling within the region. Traditional energy management methods are typically based on fixed geographical partitions or topologies, dividing the area into several sub-regions and then performing power coordination control separately.

[0003] Existing methods have the following drawbacks: First, fixed partitioning is difficult to adapt to the real-time fluctuations in photovoltaic output and load demand, resulting in a lack of inherent similarity or complementarity in the net load curves of nodes within the same partition, thus affecting the power complementarity effect within the region. Second, power allocation within partitions often adopts centralized or linear consistency methods, relying on central nodes or global information interaction, resulting in high communication overhead and limited convergence speed. Therefore, there is an urgent need to develop an energy optimization method that can dynamically partition clusters based on net load fluctuation characteristics and achieve efficient distributed power coordination within the clusters, in order to improve the real-time performance and adaptability of energy management in rural micro-balance zones. Summary of the Invention

[0004] In view of the above problems, the purpose of this invention is to provide a method, system, storage medium and electronic device for optimizing energy consumption in rural micro-balance zones based on edge computing, so as to solve the problems mentioned in the background art.

[0005] This invention provides a method for optimizing energy consumption in rural micro-balance zones based on edge computing, comprising: M1: Extract fluctuation features from the local net load time series dataset within the township micro-balance zone to obtain the net load fluctuation feature vector; M2: The net load fluctuation feature vector is used as a node identifier and periodically exchanged with adjacent nodes within the communication range of the township micro-balance zone; and the results of the periodic exchange are dynamically time-normalized to obtain the similarity of fluctuation features between the current node and adjacent nodes. M3: Based on the similarity of fluctuation characteristics, determine the same group nodes of the township micro-equilibrium area and integrate them into a virtual dynamic cluster; M4: Within the virtual dynamic cluster, virtual power allocation is performed on each edge node within the virtual dynamic cluster to obtain the final power adjustment amount for each edge node; M5: Based on the final power adjustment of each edge node and the actual local net load value of the township micro-balance zone at the current moment, determine the power setting value of the distributed power unit in the township micro-balance zone at the next moment, and issue an execution command.

[0006] Preferably, the process of extracting fluctuation features from the local net load time-series dataset within the township micro-balance zone to obtain the net load fluctuation feature vector is as follows: Multiple edge nodes are deployed at each distributed power unit within the township micro-balance zone; Each edge node obtains the load power consumption sequence and distributed power output sequence of continuous historical time periods within its jurisdiction, and subtracts the distributed power output from the load power consumption at the same moment to generate a local net load time series dataset. The local net load time series dataset is divided into multiple net load subsequences with fixed window lengths by a sliding window, and the first-order difference sequence and the second-order difference sequence of each net load subsequence are used as the rate of change features of the net load subsequence. The formula for calculating the amplitude of the high-frequency component in the net load subsequence is as follows: in, The amplitude of the high-frequency component, The length of the sliding window. For the sampling point One index, The first subsequence of the net load Power values ​​at each sampling point The arithmetic mean of the net load subsequences. These are the preset high-pass filter coefficients; The waveform curvature of the net load subsequence is obtained by comparing the cumulative absolute value of the second difference with the cumulative absolute value of the first difference. The rate of change characteristic, the amplitude of the high-frequency component, and the waveform curvature corresponding to each net load subsequence are concatenated in chronological order to generate the net load fluctuation feature vector of the edge node at the current sliding window moment.

[0007] Preferably, the net load fluctuation feature vector is used as a node identifier and periodically exchanged with neighboring nodes within the communication range of the township micro-balance zone; and the results of the periodic exchange are dynamically time-warped to obtain the similarity of fluctuation features between the current node and its neighboring nodes. The process is as follows: The net load fluctuation feature vector is encapsulated into a data packet, and the data packet is broadcast to all neighboring nodes within the communication range according to a preset communication period to generate a local neighborhood feature vector set. The optimal curved path is obtained by performing point-by-point matching on the net load fluctuation feature vector in the local neighborhood feature vector set. The Euclidean distance between each pair of matching points of two adjacent feature vectors in the optimal curved path is used as the original cumulative distance value. The original cumulative distance value after length normalization is compared with the preset similarity conversion threshold, and the similarity of fluctuation features is determined based on the comparison result.

[0008] Preferably, the process of determining the same-group nodes of the township micro-equilibrium zone based on the similarity of fluctuation characteristics and integrating them into a virtual dynamic cluster is as follows: The similarity of the fluctuation features of each adjacent node is compared with a preset clustering threshold. Adjacent nodes with a similarity greater than the clustering threshold are selected and marked as candidate cluster nodes. The current edge node sends a clustering confirmation request to the candidate co-group node and receives a confirmation response from the candidate co-group node; If the confirmation response includes information that the current edge node is also marked as a candidate peer node, then the current candidate peer node will be finally determined as a peer node of this node. All the finally determined same-group nodes are integrated into a virtual dynamic cluster, and a cluster member list of the virtual dynamic cluster is generated. The cluster member list includes the node identifier of each same-group node and the corresponding fluctuation feature similarity. After each communication cycle, each edge node repeats the virtual dynamic cluster confirmation step based on the similarity of the fluctuation characteristics of newly received neighboring nodes, and dynamically updates the virtual dynamic cluster. If the similarity of fluctuation characteristics of nodes in the same group in the virtual dynamic cluster is continuously lower than the clustering threshold, then the nodes in the same group will be removed from the current virtual dynamic cluster.

[0009] Preferably, within the virtual dynamic cluster, virtual power allocation is performed on each edge node to obtain the final power adjustment amount for each edge node. The process is as follows: Obtain the current state of charge of the energy storage devices and the current reserve capacity of the distributed power units under the jurisdiction of each edge node in the current virtual dynamic cluster; Calculate the virtual potential energy value of the current edge node based on the state of charge and the reserve capacity; Each edge node exchanges the virtual potential energy value with its neighboring nodes in the virtual dynamic cluster, and dynamically updates the power adjustment amount of each edge node according to the potential energy difference between the current edge node and each neighboring node. Each edge node adds the updated power adjustment to the current virtual potential energy value to obtain the updated virtual potential energy value, and repeats the process of potential energy value exchange, potential energy difference calculation and power adjustment update. The power adjustment is taken as the final power adjustment of the edge node when the absolute value of the potential energy between all adjacent node pairs is less than the preset convergence threshold.

[0010] Preferably, the formula for calculating the final power adjustment is: in, For nodes The final power adjustment amount, The preset convergence step size factor, The value of the hyperbolic tangent function of the potential energy difference. For nodes The potential energy difference between the set of adjacent nodes. For nodes The set of adjacent nodes.

[0011] Preferably, the process of determining the next-time power setting value of the distributed power unit in the township micro-balance zone based on the final power adjustment amount of each edge node and the actual local net load value of the township micro-balance zone at the current moment, and issuing the execution instruction, is as follows: The actual local net load value of the township micro-balance zone at the current moment is collected. The actual local net load value is the difference between the power consumption of the load at the current moment and the output power of the distributed power unit. The final power adjustment of each edge node is added to the actual value of the local net load to obtain the preliminary power setting value of the distributed power unit under the jurisdiction of this node at the next moment. Each edge node sorts the distributed power units within its jurisdiction according to their response time from shortest to longest, thus obtaining a priority queue of the distributed power units. Each edge node allocates the initial power setting value to each distributed power unit in sequence according to the priority queue. Based on the power setting value finally allocated to each distributed power unit, a corresponding power execution command is generated, and the power execution command is simultaneously sent to all distributed power units according to a preset timing synchronization mechanism.

[0012] This invention also provides a rural micro-balance zone energy optimization system based on edge computing, the system comprising: The fluctuation feature extraction module is used to extract fluctuation features from the local net load time series dataset within the township micro-balance zone to obtain the net load fluctuation feature vector. The similarity calculation module is used to periodically exchange the net load fluctuation feature vector with neighboring nodes within the communication range of the township micro-balance zone, using the net load fluctuation feature vector as a node identifier; and to dynamically time-normalize the results of the periodic exchange to obtain the fluctuation feature similarity between the current node and neighboring nodes. The dynamic cluster formation module is used to determine the same-group nodes of the township micro-balance zone based on the similarity of fluctuation characteristics, and integrate them into a virtual dynamic cluster. The virtual power allocation module is used to perform virtual power allocation on each edge node within the virtual dynamic cluster to obtain the final power adjustment amount for each edge node. The execution control module is used to determine the next power setting value of the distributed power unit in the township micro-balance zone based on the final power adjustment amount of each edge node and the actual local net load value of the township micro-balance zone at the current moment, and to issue execution instructions.

[0013] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention uses the high-frequency component amplitude, rate of change, and waveform curvature of net load time-series data as multidimensional fluctuation feature vectors, and obtains the fluctuation feature similarity between nodes based on dynamic time warping. This enables the virtual dynamic cluster to adaptively reconstruct according to the inherent similarity of the net load waveform, thereby ensuring that the power complementarity between nodes in the same cluster is dynamically optimized as the fluctuation pattern changes, significantly improving the accuracy and flexibility of distributed resource coordination.

[0014] 2. This invention employs a virtual potential field coordination algorithm based on the hyperbolic tangent function for power allocation within a virtual dynamic cluster, overcoming the drawbacks of traditional centralized or linear consensus methods, such as high communication overhead and slow convergence speed. Specifically, by mapping the energy storage state of charge and the distributed power reserve capacity to virtual potential energy values, and utilizing the nonlinear saturation characteristics of the hyperbolic tangent function to iterate the power adjustment amount driven by the neighborhood potential energy difference, each node only needs to exchange potential energy values ​​with its neighboring nodes to achieve consistent convergence of potential energy within the cluster, without requiring a central node or global information. This mechanism effectively reduces communication bandwidth consumption, improves convergence speed and system robustness, and avoids oscillation problems that may arise from linear update rules, ultimately realizing dynamic self-optimizing control of energy and autonomous absorption of distributed resources within the township micro-balance zone. Attached Figure Description

[0015] Other objects and results of the invention will become more apparent and readily understood by referring to the following description taken in conjunction with the accompanying drawings, and with a more complete understanding of the invention. In the drawings: Figure 1This is a flowchart illustrating a method for optimizing energy consumption in a township micro-balance zone based on edge computing, according to an embodiment of the present invention. Figure 2 This is a functional block diagram of a township micro-balance zone energy optimization system based on edge computing, according to an embodiment of the present invention. Figure 3 This is a schematic diagram of the composition structure of an electronic device for implementing the energy consumption optimization method for township micro-balance zones based on edge computing, as described in an embodiment of the present invention. Figure 4 This is a functional block diagram of the similarity calculation module in an embodiment of the present invention; Figure 5 This is a graph showing the amplitude variation of the net load high-frequency component of the fluctuation feature extraction module in this embodiment of the invention. Detailed Implementation

[0016] To address the aforementioned problems, this invention provides a method, system, storage medium, and electronic device for optimizing energy consumption in rural micro-balance zones based on edge computing. Specific embodiments of this invention will be described in detail below with reference to the accompanying drawings.

[0017] To illustrate the energy optimization method, system, storage medium, and electronic device for rural micro-balance zones based on edge computing provided by this invention, Figure 1 An exemplary illustration is provided for a method for optimizing energy consumption in township micro-balance zones based on edge computing, according to an embodiment of the present invention. Figure 2 An exemplary illustration is provided for a township micro-balance zone energy optimization system based on edge computing according to an embodiment of the present invention.

[0018] The following description of exemplary embodiments is merely illustrative and is in no way intended to limit the invention or its application or use. Technologies and electronic devices known to those skilled in the art may not be discussed in detail, but where appropriate, such technologies and electronic devices should be considered part of the specification.

[0019] Reference Figure 1 The diagram shown is a flowchart illustrating a method for optimizing energy consumption in a township micro-balance zone based on edge computing, according to an embodiment of the present invention. In this embodiment, the method includes: M1: Extract fluctuation features from the local net load time series dataset within the township micro-balance zone to obtain the net load fluctuation feature vector; In this embodiment of the invention, the process of extracting fluctuation features from the local net load time-series dataset within the township micro-balance zone to obtain the net load fluctuation feature vector is as follows: Multiple edge nodes are deployed at each distributed power unit within the township micro-balance zone; Each edge node obtains the load power consumption sequence and distributed power output sequence of continuous historical time periods within its jurisdiction, and subtracts the distributed power output from the load power consumption at the same moment to generate a local net load time series dataset. The local net load time series dataset is divided into multiple net load subsequences with fixed window lengths by a sliding window, and the first-order difference sequence and the second-order difference sequence of each net load subsequence are used as the rate of change features of the net load subsequence. The formula for calculating the amplitude of the high-frequency component in the net load subsequence is as follows: in, The amplitude of the high-frequency component, The length of the sliding window. For the sampling point One index, The first subsequence of the net load Power values ​​at each sampling point The arithmetic mean of the net load subsequences. These are the preset high-pass filter coefficients; The waveform curvature of the net load subsequence is obtained by comparing the cumulative absolute value of the second difference with the cumulative absolute value of the first difference. The rate of change characteristic, the amplitude of the high-frequency component, and the waveform curvature corresponding to each net load subsequence are concatenated in chronological order to generate the net load fluctuation feature vector of the edge node at the current sliding window moment.

[0020] Multiple edge nodes are deployed at each distributed power unit within the township micro-balance zone. Each edge node is connected to the control ports of the photovoltaic inverters, energy storage converters, and adjustable loads within its jurisdiction via communication cables.

[0021] Each edge node acquires the load power consumption sequence and distributed power output sequence for continuous historical periods within its jurisdiction. The load power consumption sequence is read and reported to the edge node by the electricity meter installed on the user side at fixed time intervals, and the distributed power output sequence is reported to the edge node by the photovoltaic inverter or wind turbine controller at the same time intervals.

[0022] Each edge node subtracts the distributed power generation output from the load power consumption value at the same time to obtain the net load power value at that time. The net load power values ​​at all times are arranged in chronological order to generate a local net load time series dataset.

[0023] Each edge node performs a sliding window segmentation on the local net load time series dataset. The window length of the sliding window is preset by the system administrator based on the typical cycle of net load fluctuations in the micro-balance zone of the township and stored in the edge node. The sliding window moves backward by one sampling point each time, and a fixed number of overlapping sampling points are maintained between adjacent windows, thereby obtaining multiple net load subsequences with fixed window lengths.

[0024] For each net load subsequence, the edge nodes calculate their first-order difference sequence: subtract the power values ​​of two adjacent sampling points in the subsequence, the difference value of the first sampling point is missing, and from the second sampling point onwards, obtain the difference between the current power value and the previous power value at each position. All the differences are arranged in order to form a first-order difference sequence, which serves as a rate of change feature of the net load subsequence.

[0025] For each net load subsequence, the edge nodes further calculate their second-order difference sequence: subtract two adjacent difference values ​​in the already obtained first-order difference sequence, and obtain the difference between the current difference value and the previous difference value at each position starting from the second difference value. All differences are arranged in order to form a second-order difference sequence, which serves as another rate of change feature of the net load subsequence.

[0026] The edge node subtracts the arithmetic mean of the power values ​​of all sample points in each subsequence from the power value of each sample point in each net load subsequence to obtain the centering deviation value of each sample point.

[0027] The edge node multiplies the centering deviation value of each sampling point by its corresponding preset high-pass filter coefficient. The high-pass filter coefficient is a pre-calculated numerical sequence with a length equal to the sliding window length. Each coefficient value is between zero and one, and the coefficient value corresponding to the high-frequency component is greater than the coefficient value corresponding to the low-frequency component. All coefficient values ​​are determined and stored in the node by looking up a table based on the sampling frequency and the preset cutoff frequency of the edge node during system initialization.

[0028] The edge node squares the centered deviation value of each sampling point after multiplying by a coefficient, then sums up the square values ​​of all sampling points within the sliding window, divides by the length of the sliding window, and finally performs a square root operation on the division result to obtain the high-frequency component amplitude of the net load subsequence.

[0029] The sliding window length is a pre-defined fixed value representing the total number of sampling points in each net load subsequence. The sampling point index identifies the positional order of each sampling point in the net load subsequence, increasing sequentially from the first sampling point. The power value is the net load power value corresponding to the indexed sampling point in the net load subsequence, in kilowatts. The arithmetic mean is the center value obtained by summing the power values ​​of all sampling points in the net load subsequence and then dividing by the sliding window length. The high-pass filter coefficients are pre-defined weighted values ​​for each sampling point position within the sliding window. All high-pass filter coefficients form a numerical sequence with a length equal to the sliding window length. The value of each high-pass filter coefficient ranges from zero to one, and the coefficient values ​​corresponding to high-frequency components in the first half of the window are greater than the coefficient values ​​corresponding to low-frequency components in the second half.

[0030] For each net load subsequence, the edge node first calculates the arithmetic mean of the power values ​​of all sampling points in that subsequence. Then, it subtracts this arithmetic mean from the power value of each sampling point to obtain the centering deviation value for each sampling point. The edge node then squares the centering deviation value of each sampling point to obtain the squared deviation value. This squared deviation value is then multiplied by the high-pass filter coefficient corresponding to the sampling point's location to obtain the weighted squared deviation value for each sampling point. The edge node sums up all the weighted squared deviation values ​​to obtain a total value, and then divides this total value by the sliding window length to obtain an average weighted squared deviation value. Finally, the edge node performs a square root operation on this average weighted squared deviation value, and the result is the high-frequency component amplitude.

[0031] The edge node takes the absolute value of each difference value in the second difference sequence of each net load subsequence, and then sums up all the absolute values ​​to obtain the cumulative sum of the absolute values ​​of the second difference.

[0032] The edge node takes the absolute value of each difference value in the first-order difference sequence of each net load subsequence, and then sums up all the absolute values ​​to obtain the cumulative sum of the absolute values ​​of the first-order differences. The edge node divides the cumulative sum of the absolute values ​​of the second-order differences by the cumulative sum of the absolute values ​​of the first-order differences. The ratio obtained is the waveform curvature of the net load subsequence. The larger the ratio, the more severe the curvature of the net load waveform; the smaller the ratio, the smoother the waveform.

[0033] The edge node concatenates the rate of change features corresponding to each net load subsequence, namely the first-order difference sequence and the second-order difference sequence, the high-frequency component amplitude, and the waveform curvature, into a long sequence in chronological order. The first value of this long sequence comes from the first-order difference value of the first sampling point, and the subsequent values ​​are arranged in sequence to finally obtain the net load fluctuation feature vector of the edge node at the current sliding window time.

[0034] M2: The net load fluctuation feature vector is used as a node identifier and periodically exchanged with adjacent nodes within the communication range of the township micro-balance zone; and the results of the periodic exchange are dynamically time-normalized to obtain the similarity of fluctuation features between the current node and adjacent nodes. In this embodiment of the invention, the net load fluctuation feature vector is used as a node identifier and periodically exchanged with neighboring nodes within the communication range of the township micro-balance zone; and the results of the periodic exchange are dynamically time-normalized to obtain the similarity of fluctuation features between the current node and its neighboring nodes. The process is as follows: The net load fluctuation feature vector is encapsulated into a data packet, and the data packet is broadcast to all neighboring nodes within the communication range according to a preset communication period to generate a local neighborhood feature vector set. The optimal curved path is obtained by performing point-by-point matching on the net load fluctuation feature vector in the local neighborhood feature vector set. The Euclidean distance between each pair of matching points of two adjacent feature vectors in the optimal curved path is used as the original cumulative distance value. The original cumulative distance value after length normalization is compared with the preset similarity conversion threshold, and the similarity of fluctuation features is determined based on the comparison result.

[0035] The net load fluctuation feature vector is encapsulated according to a preset data packet format. During encapsulation, a packet header is added to the beginning of the feature vector data. The packet header contains the unique identifier of this edge node and the timestamp that generated the feature vector. The timestamp is accurate to the millisecond level.

[0036] Each edge node maintains a timer. The duration of this timer is set by the system administrator based on the typical rate of change of net load fluctuations within the micro-balance zone of the township. Every time the timer reaches its duration, it triggers a communication action. The communication action is that the edge node broadcasts the pre-encapsulated data packet to all neighboring nodes within its communication range through its onboard wireless communication module.

[0037] While broadcasting data packets, each edge node also continuously listens to the wireless communication channel, receiving data packets broadcast from neighboring nodes. When a data packet is received, the edge node parses out the node identifier and timestamp in the packet header, extracts the net load fluctuation feature vector in the data packet, and stores the parsed feature vector with the corresponding node identifier.

[0038] Each edge node combines its current net load fluctuation feature vector with the net load fluctuation feature vectors of all its neighboring nodes to form a local neighborhood feature vector set. Each feature vector in this set is accompanied by the identifier of its source node.

[0039] Each edge node extracts its own feature vector from the local neighborhood feature vector set, and then extracts the feature vector of each of its neighboring nodes in turn. A point-by-point matching operation is performed on these two feature vectors. The specific process of the point-by-point matching operation is as follows: the two feature vectors are regarded as two numerical sequences, and each value in the sequence corresponds to a feature value at a sampling time. It is allowed that a certain position in the first sequence is associated with multiple consecutive positions in the second sequence. By repeatedly trying different correspondence methods, a path is found from the starting point to the ending point of the two sequences and passes through all corresponding points. This path minimizes the sum of the feature value differences of each pair of corresponding points on the path among all possible paths. This path is called the optimal curved path.

[0040] Each edge node travels along the optimal curved path. When it reaches a matching point pair, it calculates the difference between the two values ​​in the pair, squares the difference, and then takes the square root to obtain the Euclidean distance. The Euclidean distances of all matching point pairs on the path are summed one by one to obtain the original cumulative distance value.

[0041] For each edge node, the total number of matching point pairs on the optimal curved path is counted. This total number is used as the total number of steps on the path. Then, the original cumulative distance value is divided by the total number of steps on the path to obtain the normalized original cumulative distance value. This step is to eliminate the influence of feature vector sequences of different lengths on the distance value.

[0042] Each edge node pre-stores a similarity conversion threshold. The threshold is determined as follows: In the initial stage of system deployment, the maintenance personnel select ten edge node pairs with known highly similar fluctuation patterns within the micro-equilibrium zone of the township, calculate the normalized original cumulative distance value between them, and take the maximum value among these ten distance values ​​as the first reference value; then select ten edge node pairs with completely dissimilar fluctuation patterns, calculate the normalized original cumulative distance value between them, and take the minimum value among these ten distance values ​​as the second reference value; add the first reference value and the second reference value and divide by two to obtain the similarity conversion threshold, which is burned into the non-volatile memory of each edge node.

[0043] Each edge node compares the normalized original cumulative distance value with the similarity conversion threshold. If the normalized original cumulative distance value is less than or equal to the similarity conversion threshold, the fluctuation feature similarity between the current neighbor node and this node is determined to be the maximum value; if the normalized original cumulative distance value is greater than the similarity conversion threshold, the fluctuation feature similarity is determined to be zero.

[0044] M3: Based on the similarity of fluctuation characteristics, determine the same group nodes of the township micro-equilibrium area and integrate them into a virtual dynamic cluster; In this embodiment of the invention, the process of determining the same-group nodes of the township micro-equilibrium zone based on the similarity of fluctuation characteristics and integrating them into a virtual dynamic cluster is as follows: The similarity of the fluctuation features of each adjacent node is compared with a preset clustering threshold. Adjacent nodes with a similarity greater than the clustering threshold are selected and marked as candidate cluster nodes. The current edge node sends a clustering confirmation request to the candidate co-group node and receives a confirmation response from the candidate co-group node; If the confirmation response includes information that the current edge node is also marked as a candidate peer node, then the current candidate peer node will be finally determined as a peer node of this node. All the finally determined same-group nodes are integrated into a virtual dynamic cluster, and a cluster member list of the virtual dynamic cluster is generated. The cluster member list includes the node identifier of each same-group node and the corresponding fluctuation feature similarity. After each communication cycle, each edge node repeats the virtual dynamic cluster confirmation step based on the similarity of the fluctuation characteristics of newly received neighboring nodes, and dynamically updates the virtual dynamic cluster. If the similarity of fluctuation characteristics of nodes in the same group in the virtual dynamic cluster is continuously lower than the clustering threshold, then the nodes in the same group will be removed from the current virtual dynamic cluster.

[0045] Each edge node stores a clustering threshold. The specific value of the clustering threshold is determined during system initialization in the following way: The maintenance personnel select ten pairs of edge nodes with highly similar fluctuation patterns within the micro-equilibrium zone of the township, calculate the fluctuation feature similarity between each pair of nodes, and take the minimum value among these ten similarity values ​​as the first value; then select ten pairs of edge nodes with completely dissimilar fluctuation patterns, calculate the fluctuation feature similarity between each pair of nodes, and take the maximum value among these ten similarity values ​​as the second value; add the first value and the second value and divide by two to obtain the quotient as the clustering threshold. This threshold is a real number between zero and one and is written into the read-only memory of each edge node.

[0046] Each edge node reads the clustering threshold from its local memory, then retrieves the similarity values ​​of the fluctuation features of each neighboring node one by one, compares each similarity value with the clustering threshold, and when the similarity value of a neighboring node is greater than the clustering threshold, the edge node extracts the identifier of the neighboring node and stores it in a temporary list. All nodes in this temporary list are called candidate co-group nodes.

[0047] For each candidate cluster node in the temporary list, the current edge node generates a clustering confirmation request message. The header of the message contains the identifier of the current edge node, and the payload of the message contains an instruction code to request to join the cluster. Then, the confirmation request message is sent to the address of the candidate cluster node via unicast through the wireless communication module.

[0048] After sending a clustering confirmation request message, the current edge node starts a timeout timer to wait for the confirmation response message returned by the candidate co-group node. The duration of the timeout timer is set to one-third of the communication cycle. If the confirmation response message is received within the timeout period, the content of the message is parsed and it is checked whether the message contains a specific field. The content of this field is the information that marks the current edge node's identifier as a candidate co-group node.

[0049] If the response message does indeed contain that specific field, it indicates that the other node has also included the current edge node in its candidate co-group node list through the same clustering threshold comparison process. In this case, the current edge node will finally determine that the candidate co-group node as its own co-group node and move the node's identifier into its local co-group node master list.

[0050] The current edge node puts all the same-group node identifiers obtained after the above two-way confirmation together with its own node identifier to form a virtual dynamic cluster, and generates a cluster member list of the cluster. The list is stored in the form of a table, and each row contains the node identifier of a same-group node and the calculated fluctuation feature similarity value corresponding to that node.

[0051] After each communication cycle ends, i.e. after the timer completes one broadcast and data reception process, each edge node automatically repeats all the operations mentioned above, including clustering threshold comparison, generating a candidate cluster node list, sending clustering confirmation requests, receiving confirmation responses, and two-way confirmation, based on the fluctuation feature similarity recalculated from the newly received neighboring node broadcast data packets. This dynamically updates the existing virtual dynamic cluster by adding newly confirmed cluster nodes to the cluster member list and updating the latest fluctuation feature similarity value for each node in the list.

[0052] During each dynamic update, each edge node checks whether the fluctuation feature similarity value of each peer node in the cluster member list is continuously lower than the clustering threshold. Continuously lower than means that in the four most recent communication cycles, the fluctuation feature similarity value recalculated for that peer node at the end of each cycle is all less than the clustering threshold. No value greater than or equal to the clustering threshold is allowed in the four cycles. At this time, the edge node removes that peer node from the cluster member list and deletes all associated information of that node in the virtual dynamic cluster.

[0053] M4: Within the virtual dynamic cluster, virtual power allocation is performed on each edge node within the virtual dynamic cluster to obtain the final power adjustment amount for each edge node; In this embodiment of the invention, the process of performing virtual power allocation on each edge node within the virtual dynamic cluster to obtain the final power adjustment amount for each edge node is as follows: Obtain the current state of charge of the energy storage devices and the current reserve capacity of the distributed power units under the jurisdiction of each edge node in the current virtual dynamic cluster; Calculate the virtual potential energy value of the current edge node based on the state of charge and the reserve capacity; in, For edge nodes The virtual potential energy value, The value of the hyperbolic tangent function of the potential energy difference. The weighting coefficient represents the influence of the state of charge on the virtual potential energy value. The weighting coefficient for the impact of reserve capacity on the virtual potential energy value. For nodes The state of charge of the energy storage equipment under its jurisdiction For nodes The reserve capacity of the distributed power units under its jurisdiction; The weighting coefficient for the influence of the state of charge (SBC) on the virtual potential energy value is a preset constant used to control the proportion of the SBC of the energy storage device in the calculation of the virtual potential energy value. The specific value of this coefficient is determined through testing during system initialization, and its range is between 0.01 and 0.1. The weighting coefficient for the influence of reserve capacity on the virtual potential energy value is another preset constant used to control the proportion of the reserve capacity of the distributed power unit in the calculation of the virtual potential energy value. The specific value of this coefficient is also determined through testing during system initialization, and its range is between 0.01 and 0.1. The SBC of the energy storage device refers to the percentage of electricity currently stored in the device relative to its total capacity. This value is collected in real time by the power management terminal and provided to the edge nodes in the form of a percentage. The reserve capacity of the distributed power unit refers to the upper limit of active power output that the unit can currently add, measured in kilowatts. This value is calculated by the edge nodes based on the difference between the current output value and the rated capacity. The virtual potential energy value is a dimensionless value mapped by a hyperbolic tangent function, ranging from -1 to +1.

[0054] The edge node first multiplies the read state of charge (SOC) value of the energy storage device by a weighting factor on the influence of SOC on the virtual potential energy value, obtaining a SOC-weighted value. Simultaneously, it multiplies the reserve capacity value of the distributed power unit by a weighting factor on the influence of reserve capacity on the virtual potential energy value, obtaining a reserve capacity-weighted value. These two weighted values ​​are then added together to obtain a sum. The edge node uses this sum as input to a hyperbolic tangent function. The hyperbolic tangent function performs a non-linear mapping on this input; the output is zero when the input is zero, positive when the input is positive and gradually approaches one, and negative when the input is negative and gradually approaches negative one. The value output by the hyperbolic tangent function after calculation is the virtual potential energy value of the current edge node.

[0055] Each edge node exchanges the virtual potential energy value with its neighboring nodes in the virtual dynamic cluster, and dynamically updates the power adjustment amount of each edge node according to the potential energy difference between the current edge node and each neighboring node. Each edge node adds the updated power adjustment to the current virtual potential energy value to obtain the updated virtual potential energy value, and repeats the process of potential energy value exchange, potential energy difference calculation and power adjustment update. The power adjustment is taken as the final power adjustment of the edge node when the absolute value of the potential energy between all adjacent node pairs is less than the preset convergence threshold.

[0056] The formula for calculating the final power adjustment is: in, For nodes The final power adjustment amount, The preset convergence step size factor, The value of the hyperbolic tangent function of the potential energy difference. For nodes The potential energy difference between the set of adjacent nodes. For nodes The set of adjacent nodes.

[0057] Each edge node pre-stores a first weighting coefficient and a second weighting coefficient. These two coefficients are determined as follows: In the early stages of system deployment, maintenance personnel select three typical edge nodes in the micro-balance zone of the township and record the power regulation capabilities of their energy storage devices when the state of charge is 20%, 50%, and 80%, as well as the response effects of the distributed power supply when the standby capacity is 0%, 50%, and 100%. Through multiple tests, the coefficient combination that allows the virtual potential energy value to smoothly reflect the power regulation requirements is found. The first weighting coefficient is set to 0.05, and the second weighting coefficient is set to 0.03. These two values ​​are burned into the read-only memory of each edge node.

[0058] Each edge node pre-stores a convergence step size factor. The step size factor is determined as follows: In the early stages of system deployment, the operations and maintenance personnel set up a test environment containing three adjacent edge nodes. They first use an initial value to conduct iterative tests and observe the magnitude of each update of the power adjustment. If the update magnitude is too large, causing the virtual potential energy value to oscillate repeatedly between adjacent nodes and become unstable, the step size factor is reduced by half. If the update magnitude is too small, causing the number of iterations required for convergence to exceed fifty, the step size factor is doubled. After five rounds of testing and adjustment, the value that can guarantee stable convergence within twenty iterations without oscillation is taken as the convergence step size factor, and this value is written into the configuration file of each edge node.

[0059] Each edge node pre-stores a convergence threshold, which the system administrator sets based on the power regulation accuracy requirements of each distributed power unit in the micro-balance zone of the township. Specifically, the administrator checks the minimum adjustable power step size of the energy storage converter in the area, takes half of this step size as the base value of the convergence threshold, checks the power regulation resolution of the photovoltaic inverter, and takes the smaller of the two values. Finally, the convergence threshold is set to 0.1 times this smaller value. For example, if the minimum adjustment step size of the energy storage converter is 1000 watts and the adjustment resolution of the photovoltaic inverter is 2000 watts, then 0.1 times 1000 watts is taken as the convergence threshold. This value is written into the non-volatile memory of each edge node.

[0060] Each edge node reads the power management terminal connected to it through a wired communication interface. The power management terminal collects the current state of charge (SOC) value of the energy storage devices under its jurisdiction every second. The SOC value is expressed as a percentage, ranging from 0% to 100%. At the same time, it reads the current reserve capacity value of the distributed power units under its jurisdiction. The reserve capacity value is in kilowatts, which represents the upper limit of active power output that the distributed power can still increase at the current moment.

[0061] Each edge node multiplies the read state of charge value by a first weighting coefficient and the reserve capacity value by a second weighting coefficient. The two products are then added together to obtain an intermediate sum. This intermediate sum is then used as the input to the hyperbolic tangent function, which maps the input to an output value between -1 and +1. This output value is used as the virtual potential energy value of the current edge node.

[0062] Each edge node generates a broadcast message containing its own virtual potential energy value and sends this message to all neighboring nodes in the virtual dynamic cluster member list via the wireless communication module. Simultaneously, it receives broadcast messages from each neighboring node and parses out the virtual potential energy values ​​of those neighboring nodes. Each edge node subtracts its own virtual potential energy value from the virtual potential energy values ​​of each neighboring node to obtain the potential energy difference between itself and each neighboring node. The potential energy difference can be positive or negative.

[0063] Each edge node inputs each potential energy difference into a nonlinear transformation function. This nonlinear transformation function adopts the form of a hyperbolic tangent function, which maps the potential energy difference to an output value between -1 and +1. Then, the output values ​​corresponding to all adjacent nodes are summed. The sum is multiplied by the convergence step size factor to obtain the power adjustment amount of this node in the current iteration cycle. The power adjustment amount is in kilowatts. A positive value indicates that the power output needs to be increased, and a negative value indicates that the power output needs to be decreased.

[0064] The preset convergence step size factor is a constant pre-stored in the edge nodes, used to control the update magnitude of power adjustment in each iteration. The specific value of this constant is determined through multiple tests in the early stages of system deployment, and its range is usually between 0.01 and 0.2 to ensure stable convergence of the iteration process without oscillations. The potential energy difference refers to the difference between the virtual potential energy value of the current edge node and the virtual potential energy value of each neighboring node in the neighboring node set. This difference can be positive or negative, reflecting the direction and magnitude of the power coordination needs between the two nodes. The hyperbolic tangent function value of the potential energy difference is the value obtained after nonlinearly mapping each potential energy difference. The mapped result ranges from -1 to +1. When the potential energy difference is zero, the mapped value is zero; when the potential energy difference is positive, the mapped value is positive and gradually approaches one; when the potential energy difference is negative, the mapped value is negative and gradually approaches -1. The neighboring node set refers to the set of identifiers of all neighboring edge nodes that the current edge node can directly communicate with within the virtual dynamic cluster. The number of nodes in this set depends on the network topology. The final power adjustment is a converged power adjustment value obtained after multiple iterative calculations. The unit is kilowatt. A positive value indicates that the edge node needs to increase its outward power output, while a negative value indicates that it needs to reduce its power output or absorb power.

[0065] The edge node first obtains the potential energy difference between itself and each of its neighboring nodes, and then calculates the hyperbolic tangent function value for each potential energy difference individually. The hyperbolic tangent function values ​​corresponding to all neighboring nodes are summed to obtain a total value. The edge node multiplies this total value by a preset convergence step size factor, and the product is the power adjustment amount for the current iteration cycle. The edge node adds this power adjustment amount to the current virtual potential energy value, completing one update. The above steps are repeated until the absolute value of the potential energy difference between all neighboring nodes is less than a preset convergence threshold, at which point the iteration stops, and the power adjustment amount calculated in the last iteration is taken as the final power adjustment amount.

[0066] Each edge node adds the calculated power adjustment to the current virtual potential energy value, replacing the original virtual potential energy value with the added value, thus completing one update of the virtual potential energy value. After each update, each edge node checks whether the potential energy differences between itself and all its neighboring nodes satisfy the convergence condition, which is that the absolute value of the potential energy difference is less than the convergence threshold.

[0067] If the absolute value of the potential energy difference between any neighboring node and the current node is greater than or equal to the convergence threshold, the edge node repeats all the operations described above: exchanging virtual potential energy values, calculating potential energy differences, calculating power adjustments, and updating virtual potential energy values, before entering the next iteration cycle. If the absolute value of the potential energy difference between the current node and all neighboring nodes is less than the convergence threshold, the edge node stops iterating, uses the power adjustment calculated in the last iteration cycle as its final power adjustment, and stores this final power adjustment in a local register for use in subsequent steps.

[0068] M5: Based on the final power adjustment of each edge node and the actual local net load value of the township micro-balance zone at the current moment, determine the power setting value of the distributed power unit in the township micro-balance zone at the next moment, and issue an execution command.

[0069] In this embodiment of the invention, the process of determining the next-time power setting value of the distributed power unit in the township micro-balance zone based on the final power adjustment amount of each edge node and the actual local net load value of the township micro-balance zone at the current moment, and issuing the execution instruction, is as follows: The actual local net load value of the township micro-balance zone at the current moment is collected. The actual local net load value is the difference between the power consumption of the load at the current moment and the output power of the distributed power unit. The final power adjustment of each edge node is added to the actual value of the local net load to obtain the preliminary power setting value of the distributed power unit under the jurisdiction of this node at the next moment. Each edge node sorts the distributed power units within its jurisdiction according to their response time from shortest to longest, thus obtaining a priority queue of the distributed power units. Each edge node allocates the initial power setting value to each distributed power unit in sequence according to the priority queue. Based on the power setting value finally allocated to each distributed power unit, a corresponding power execution command is generated, and the power execution command is simultaneously sent to all distributed power units according to a preset timing synchronization mechanism.

[0070] Each edge node reads the current load power consumption value and the output power value of the distributed power unit through a wired communication interface. The load power consumption value is read by the electricity meter installed on the user side, and the output power value of the distributed power unit is reported by the photovoltaic inverter or energy storage converter. The edge node subtracts the output power value of the distributed power unit from the load power consumption value, and the difference is called the actual value of the local net load.

[0071] Each edge node retrieves the final power adjustment from its local register, adds the final power adjustment to the recently acquired actual local net load value, and uses the sum as the initial power setting value for the distributed power unit under its jurisdiction at the next moment.

[0072] Each edge node scans all distributed power units within its jurisdiction, including energy storage converters, photovoltaic inverters, and adjustable loads, and obtains the response time parameter for each unit one by one. The response time parameter refers to the length of time required from receiving a command to actually starting to change the power output, in milliseconds. This parameter is pre-stored in the configuration table of the edge node. The edge node arranges these units into a sequence according to the shortest response time, which is called the priority queue of distributed power units. The unit with the shortest response time is placed at the front of the queue.

[0073] Each edge node starts from the first unit in the priority queue, retrieves the maximum adjustable capacity of the current unit (the maximum adjustable capacity refers to the upper limit of power that the unit can increase or decrease without exceeding the safe operating range), compares the currently allocated preliminary power setting with the maximum adjustable capacity. If the preliminary power setting is less than or equal to the maximum adjustable capacity, the entire preliminary power setting is allocated to the current unit; if the preliminary power setting is greater than the maximum adjustable capacity, the entire maximum adjustable capacity is allocated to the current unit. Then, the allocated capacity is subtracted from the preliminary power setting to obtain the remaining unallocated value, which is then moved to the next unit in the queue. This process is repeated until all unallocated values ​​are allocated or all units in the queue have participated in the allocation.

[0074] Each edge node generates a power execution instruction for each distributed power unit. This instruction contains the unit's unique identifier and the power setting value assigned to that unit. The edge node packages all generated power execution instructions into a single instruction set.

[0075] Each edge node maintains a preset timing synchronization mechanism. The specific content of this timing synchronization mechanism is as follows: after generating a power execution command, each edge node does not send it immediately, but waits for a global synchronization pulse signal. This global synchronization pulse signal is broadcast once every fixed time interval by a reference clock source in the township micro-balance zone. All edge nodes receive the pulse signal at the same time. At the same millisecond moment of receiving the pulse signal, each edge node sends the power execution command to all distributed power units in its jurisdiction through its communication interface.

[0076] like Figure 2 The diagram shown is a functional block diagram of a township micro-balance zone energy optimization system 100 based on edge computing provided in an embodiment of the present invention, including a fluctuation feature extraction module 101, a similarity calculation module 102, a dynamic cluster formation module 103, a virtual power allocation module 104, and an execution control module.

[0077] In this embodiment, the functions of each module are as follows: The fluctuation feature extraction module 101 is used to extract fluctuation features from the local net load time series dataset within the township micro-balance zone to obtain a net load fluctuation feature vector. The similarity calculation module 102 is used to periodically exchange the net load fluctuation feature vector with adjacent nodes within the communication range of the township micro-balance zone, using the net load fluctuation feature vector as a node identifier; and to dynamically time-normalize the results of the periodic exchange to obtain the fluctuation feature similarity between the current node and adjacent nodes. The dynamic cluster forming module 103 is used to determine the same group nodes of the township micro-balance zone based on the similarity of fluctuation characteristics, and integrate them into a virtual dynamic cluster. The virtual power allocation module 104 is used to perform virtual power allocation on each edge node within the virtual dynamic cluster to obtain the final power adjustment amount of each edge node. The execution control module 105 is used to determine the next moment power setting value of the distributed power unit in the township micro-balance zone based on the final power adjustment amount of each edge node and the actual local net load value of the township micro-balance zone at the current moment, and to issue an execution command.

[0078] As can be seen from the above embodiments, the energy optimization system for township micro-balance zones based on edge computing provided by the present invention achieves dynamic linkage between inspection rules and standard library by automatically binding data fields with standard items and generating quality inspection task chains; by running the quality inspection task chains to directly mark outliers and aggregate the types of inspection conditions they violate and the associated standard items, it provides accurate traceability for data correction, shortens the repair cycle and reduces human error; by splitting the corrected data into resource units according to business themes and configuring API interfaces, it realizes data asset governance as a service, avoids redundant processing in multiple systems, and achieves the integrated effect of closed-loop data quality control and efficient sharing.

[0079] like Figure 3 The diagram shown is a schematic representation of the electronic device that implements the energy optimization method for a township micro-balance zone based on edge computing according to an embodiment of the present invention. The electronic device may include a processor 10, a memory 11, a communication bus 12, and a communication interface 13. It may also include a computer program stored in the memory 11 and capable of running on the processor 10, such as a township micro-balance zone energy optimization program based on edge computing.

[0080] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the memory 11 (e.g., executing an edge computing-based energy optimization program for rural micro-balance zones) and calls data stored in the memory 11 to perform various functions of the device and process data.

[0081] The memory 11 includes at least one type of storage medium, including flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of an electronic device, such as a portable hard drive. In other embodiments, the memory 11 can be an external storage device of the electronic device, such as a plug-in portable hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc. Furthermore, the memory 11 can include both internal and external storage units of the electronic device. The memory 11 can be used not only to store application software and various types of data installed on the electronic device, such as the code of a rural micro-balance zone energy optimization program based on edge computing, but also to temporarily store data that has been output or will be output.

[0082] The communication bus 12 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0083] The communication interface 13 is used for communication between the aforementioned electronic device and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, Bluetooth interface, etc.), typically used to establish communication connections between the electronic device and other devices. The user interface may be a display, an input unit (such as a keyboard), or optionally, a standard wired or wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device and to display a visual user interface.

[0084] The figure only shows an electronic device with components. Those skilled in the art will understand that the structure shown in the figure does not constitute a limitation on the electronic device and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0085] For example, although not shown, the electronic device may also include a power supply (such as a battery) to power the various components. Preferably, the power supply can be logically connected to the at least one processor 10 through a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0086] Figure 4 This diagram illustrates the optimal curvature path obtained through dynamic time warping during the calculation of the similarity of fluctuation characteristics between the current edge node and its neighboring nodes. The horizontal and vertical axes represent the index positions of each sampling point in the net load fluctuation feature vectors of the two edge nodes, respectively. The broken line path represents the optimal matching point pairs traversed from the starting point to the ending point of the two feature vectors. Each point on the path represents a pair of mutually matching sampling points; the closer the path is to the diagonal, the more similar the waveforms of the two feature vectors are.

[0087] Figure 5 This figure illustrates the curves showing the variation of the high-frequency component amplitude of net load calculated from four different edge nodes over time during the fluctuation feature extraction process. The horizontal axis represents time in hours, and the vertical axis represents the high-frequency component amplitude in the same units as net load power. Four curves with different line shapes are plotted, corresponding to nodes one, two, three, and four, respectively. Each curve shows the trend of the high-frequency component amplitude collected at multiple consecutive hourly intervals for the corresponding node. The curve shapes differ between nodes, reflecting the dynamic evolution of the strength of the high-frequency component in the net load fluctuations within each node's jurisdiction over time.

[0088] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0089] The memory 11 in the electronic device stores a rural micro-balance zone energy optimization program based on edge computing, which is a combination of multiple instructions. When run in the processor 10, it can achieve the following: M1: Extract fluctuation features from the local net load time series dataset within the township micro-balance zone to obtain the net load fluctuation feature vector; M2: The net load fluctuation feature vector is used as a node identifier and periodically exchanged with adjacent nodes within the communication range of the township micro-balance zone; and the results of the periodic exchange are dynamically time-normalized to obtain the similarity of fluctuation features between the current node and adjacent nodes. M3: Based on the similarity of fluctuation characteristics, determine the same group nodes of the township micro-equilibrium area and integrate them into a virtual dynamic cluster; M4: Within the virtual dynamic cluster, virtual power allocation is performed on each edge node within the virtual dynamic cluster to obtain the final power adjustment amount for each edge node; M5: Based on the final power adjustment of each edge node and the actual local net load value of the township micro-balance zone at the current moment, determine the power setting value of the distributed power unit in the township micro-balance zone at the next moment, and issue an execution command.

[0090] Specifically, the specific implementation method of the processor 10 for the above instructions can be referred to the description of the relevant steps in the corresponding embodiment of the accompanying drawings, and will not be repeated here.

[0091] Furthermore, if the modules / units integrated into the device are implemented as software functional units and sold or used as independent products, they can be stored in a storage medium. The storage medium can be volatile or non-volatile. For example, the storage medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0092] In the several embodiments provided by this invention, it should be understood that the disclosed electronic devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0093] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0094] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0095] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0096] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for optimizing energy consumption in township micro-balance zones based on edge computing, characterized in that, The method includes: M1: Extract fluctuation features from the local net load time series dataset within the township micro-balance zone to obtain the net load fluctuation feature vector; M2: The net load fluctuation feature vector is used as a node identifier and periodically exchanged with adjacent nodes within the communication range of the township micro-balance zone; and the results of the periodic exchange are dynamically time-normalized to obtain the similarity of fluctuation features between the current node and adjacent nodes. M3: Based on the similarity of fluctuation characteristics, determine the same group nodes of the township micro-equilibrium area and integrate them into a virtual dynamic cluster; M4: Within the virtual dynamic cluster, virtual power allocation is performed on each edge node within the virtual dynamic cluster to obtain the final power adjustment amount for each edge node; M5: Based on the final power adjustment of each edge node and the actual local net load value of the township micro-balance zone at the current moment, determine the power setting value of the distributed power unit in the township micro-balance zone at the next moment, and issue an execution command.

2. The energy consumption optimization method for township micro-balance zones based on edge computing as described in claim 1, characterized in that, The process of extracting fluctuation features from the local net load time-series dataset within the township micro-balance zone to obtain the net load fluctuation feature vector is as follows: Multiple edge nodes are deployed at each distributed power unit within the township micro-balance zone; Each edge node obtains the load power consumption sequence and distributed power output sequence of continuous historical time periods within its jurisdiction, and subtracts the distributed power output from the load power consumption at the same moment to generate a local net load time series dataset. The local net load time series dataset is divided into multiple net load subsequences with fixed window lengths by a sliding window, and the first-order difference sequence and the second-order difference sequence of each net load subsequence are used as the rate of change features of the net load subsequence. The formula for calculating the amplitude of the high-frequency component in the net load subsequence is as follows: in, The amplitude of the high-frequency component, The length of the sliding window. For the sampling point One index, The first subsequence of the net load Power values ​​at each sampling point The arithmetic mean of the net load subsequences. These are the preset high-pass filter coefficients; The waveform curvature of the net load subsequence is obtained by comparing the cumulative absolute value of the second difference with the cumulative absolute value of the first difference. The rate of change characteristic, the amplitude of the high-frequency component, and the waveform curvature corresponding to each net load subsequence are concatenated in chronological order to generate the net load fluctuation feature vector of the edge node at the current sliding window moment.

3. The energy consumption optimization method for township micro-balance zones based on edge computing as described in claim 2, characterized in that, The net load fluctuation feature vector is used as a node identifier and periodically exchanged with neighboring nodes within the communication range of the township micro-balance zone; the results of the periodic exchange are then dynamically time-warped to obtain the similarity of fluctuation features between the current node and its neighboring nodes. The process is as follows: The net load fluctuation feature vector is encapsulated into a data packet, and the data packet is broadcast to all neighboring nodes within the communication range according to a preset communication period to generate a local neighborhood feature vector set. The optimal curved path is obtained by performing point-by-point matching on the net load fluctuation feature vector in the local neighborhood feature vector set. The Euclidean distance between each pair of matching points of two adjacent feature vectors in the optimal curved path is used as the original cumulative distance value. The original cumulative distance value after length normalization is compared with the preset similarity conversion threshold, and the similarity of fluctuation features is determined based on the comparison result.

4. The energy consumption optimization method for township micro-balance zones based on edge computing as described in claim 1, characterized in that, The process of determining the same-group nodes in the township micro-equilibrium zone based on the similarity of fluctuation characteristics and integrating them into a virtual dynamic cluster is as follows: The similarity of the fluctuation features of each adjacent node is compared with a preset clustering threshold. Adjacent nodes with a similarity greater than the clustering threshold are selected and marked as candidate cluster nodes. The current edge node sends a clustering confirmation request to the candidate co-group node and receives a confirmation response from the candidate co-group node; If the confirmation response includes information that the current edge node is also marked as a candidate peer node, then the current candidate peer node will be finally determined as a peer node of this node. All the finally determined same-group nodes are integrated into a virtual dynamic cluster, and a cluster member list of the virtual dynamic cluster is generated. The cluster member list includes the node identifier of each same-group node and the corresponding fluctuation feature similarity. After each communication cycle, each edge node repeats the virtual dynamic cluster confirmation step based on the similarity of the fluctuation characteristics of newly received neighboring nodes, and dynamically updates the virtual dynamic cluster. If the similarity of fluctuation characteristics of nodes in the same group in the virtual dynamic cluster is continuously lower than the clustering threshold, then the nodes in the same group will be removed from the current virtual dynamic cluster.

5. The energy consumption optimization method for township micro-balance zones based on edge computing as described in claim 1, characterized in that, Within the virtual dynamic cluster, virtual power allocation is performed on each edge node to obtain the final power adjustment amount for each edge node. The process is as follows: Obtain the current state of charge of the energy storage devices and the current reserve capacity of the distributed power units under the jurisdiction of each edge node in the current virtual dynamic cluster; Calculate the virtual potential energy value of the current edge node based on the state of charge and the reserve capacity; Each edge node exchanges the virtual potential energy value with its neighboring nodes in the virtual dynamic cluster, and dynamically updates the power adjustment amount of each edge node according to the potential energy difference between the current edge node and each neighboring node. Each edge node adds the updated power adjustment to the current virtual potential energy value to obtain the updated virtual potential energy value, and repeats the process of potential energy value exchange, potential energy difference calculation and power adjustment update. The power adjustment is taken as the final power adjustment of the edge node when the absolute value of the potential energy between all adjacent node pairs is less than the preset convergence threshold.

6. The energy consumption optimization method for township micro-balance zones based on edge computing as described in claim 5, characterized in that, The formula for calculating the final power adjustment is: in, For nodes The final power adjustment amount, The preset convergence step size factor, The value of the hyperbolic tangent function of the potential energy difference. For nodes The potential energy difference between the set of adjacent nodes. For nodes The set of adjacent nodes.

7. The energy consumption optimization method for township micro-balance zones based on edge computing as described in claim 6, characterized in that, The process of determining the next-time power setting value of the distributed power unit in the township micro-balance zone based on the final power adjustment amount of each edge node and the actual local net load value of the township micro-balance zone at the current moment, and issuing an execution command, is as follows: The actual local net load value of the township micro-balance zone at the current moment is collected. The actual local net load value is the difference between the power consumption of the load at the current moment and the output power of the distributed power unit. The final power adjustment of each edge node is added to the actual value of the local net load to obtain the preliminary power setting value of the distributed power unit under the jurisdiction of this node at the next moment. Each edge node sorts the distributed power units within its jurisdiction according to their response time from shortest to longest, thus obtaining a priority queue of the distributed power units. Each edge node allocates the initial power setting value to each distributed power unit in sequence according to the priority queue. Based on the power setting value finally allocated to each distributed power unit, a corresponding power execution command is generated, and the power execution command is simultaneously sent to all distributed power units according to a preset timing synchronization mechanism.

8. A rural micro-balance zone energy optimization system based on edge computing, characterized in that, The system is used to implement the energy consumption optimization method for rural micro-balance zones based on edge computing according to any one of claims 1-7, the system comprising: The fluctuation feature extraction module is used to extract fluctuation features from the local net load time series dataset within the township micro-balance zone to obtain the net load fluctuation feature vector. The similarity calculation module is used to periodically exchange the net load fluctuation feature vector with neighboring nodes within the communication range of the township micro-balance zone, using the net load fluctuation feature vector as a node identifier; and to dynamically time-normalize the results of the periodic exchange to obtain the fluctuation feature similarity between the current node and neighboring nodes. The dynamic cluster formation module is used to determine the same-group nodes of the township micro-balance zone based on the similarity of fluctuation characteristics, and integrate them into a virtual dynamic cluster. The virtual power allocation module is used to perform virtual power allocation on each edge node within the virtual dynamic cluster to obtain the final power adjustment amount for each edge node. The execution control module is used to determine the next power setting value of the distributed power unit in the township micro-balance zone based on the final power adjustment amount of each edge node and the actual local net load value of the township micro-balance zone at the current moment, and to issue execution instructions.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the energy optimization method for township micro-balance zones based on edge computing as described in any one of claims 1 to 7.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the energy optimization method for township micro-balance zones based on edge computing as described in any one of claims 1 to 7.

Images