Adaptive cooperative charging method and system of power quality regulation device

Abstract

This invention discloses an adaptive collaborative charging method and system for power quality control equipment. The method includes: collecting power quality control data from a distribution substation and obtaining standardized feature vectors; extracting features to generate a spatiotemporal feature matrix of electrical correlations of the power quality control equipment; constructing a state prediction model to predict the state transition trend of power quality control in the distribution substation and obtain the optimal charging stage strategy; constructing a multi-objective optimization model and combining it with a distributed collaborative control algorithm to obtain the charging control parameters of the power quality control equipment; and designing an adaptive law model to adjust the charging control parameters in real time and perform dynamic evaluation. This invention, through intelligent collaborative design across the entire chain, achieves deep adaptation between the charging of the power quality control equipment and the operating state of the distribution substation. While ensuring safe and stable charging and extending battery life, it significantly improves the renewable energy absorption rate of the substation and reduces line losses.

Description

Technical Field

[0001] This invention relates to the field of energy storage management technology for power distribution areas, and in particular to an adaptive collaborative charging method and system for power quality control equipment. Background Technology

[0002] In the process of power distribution network intelligence and energy transformation, distribution transformer substations, as the "last mile" connecting the power grid and user terminals, are crucial to power supply quality and renewable energy consumption due to their operational stability. As key power regulation facilities, transformer substation power quality control equipment faces challenges such as battery life degradation, thermal runaway risks, and insufficient adaptability to source-load fluctuations due to complex operating conditions and frequent charging and discharging. Furthermore, some older substations have inadequate communication systems, limiting equipment operation to fixed local strategies and making it impossible to meet the coordinated control requirements of the "source-grid-load-storage" system.

[0003] Existing power quality control equipment charging control methods have obvious defects: manual strategy setting relies on experience and has a lag in response; centralized optimization scheduling has high requirements for prediction and communication, and the actual control effect is easily affected; fixed threshold stage switching method lacks coordination between battery status and transformer area operation objectives, making it difficult to balance safety and economy. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides an adaptive collaborative charging method and system for power quality control equipment. Through intelligent collaborative design of the entire link, it realizes deep adaptation between the charging of power quality control equipment and the operating status of distribution substations. While ensuring safe and stable charging and extending battery life, it significantly improves the renewable energy consumption rate of the substations and reduces line losses.

[0005] This invention provides an adaptive coordinated charging method for power quality control equipment, the method comprising: Power quality control data of distribution substations are collected based on sensor networks, and the collected data is standardized to obtain standardized feature vectors. Feature extraction is performed on the standardized feature vector to generate a spatiotemporal feature matrix of electrical correlations between power quality control equipment in the distribution substation. A state prediction model is constructed, and the state transition trend of power quality regulation in the distribution substation is predicted based on the state prediction model, and the optimal charging stage strategy is obtained. A multi-objective optimization model is constructed, and the charging control parameters of the power quality regulation equipment are obtained based on the multi-objective optimization model and a distributed cooperative control algorithm. An adaptive law model is designed to adjust the charging control parameters in real time and to dynamically evaluate the adjustment effect.

[0006] Furthermore, the power quality control data includes: Define key indicators for power quality control and collect data corresponding to these key indicators. These key indicators include photovoltaic array output power, total active load of the distribution area, voltage amplitude of key nodes, effective value of feeder current, DC side voltage of power quality control equipment, charging / discharging current of power quality control equipment, average temperature of battery pack, state of charge of battery, and state of health of battery.

[0007] Furthermore, the standardization process for the collected data to obtain standardized feature vectors includes: Obtain the historical data mean vector, and then perform a standardized comparison process between the collected raw data and the historical data mean vector to obtain a dimensionless standardized feature vector.

[0008] Furthermore, the step of extracting features from the standardized feature vector to generate a spatiotemporal feature matrix of electrical correlations of power quality control equipment in the distribution substation includes: Define the power quality control equipment in the distribution substation as nodes; Define the electrical connections between devices as connection edges between nodes; A spatiotemporal feature matrix is ​​generated based on the defined nodes and connecting edges, and an association graph of power quality control equipment in the distribution substation is constructed.

[0009] Furthermore, the generation of the spatiotemporal feature matrix based on the defined nodes and connecting edges includes: Based on the graph attention network feature extraction algorithm, attention coefficients are assigned to the neighboring nodes of each defined node through multi-layer network iteration; We perform weighted aggregation of the features of neighboring nodes and the features of the node itself to obtain the node feature matrix of each layer; After fusing the spatiotemporal correlation of the feature matrices of each layer of nodes, a high-order spatiotemporal feature matrix is ​​output.

[0010] Furthermore, the construction of the state prediction model, the prediction of the state transition trend of power quality regulation in the distribution substation based on the state prediction model, and the acquisition of the optimal charging stage strategy include: Extract the state vector that reflects the state transition trend of power quality regulation in the distribution substation area; Based on the extracted state vectors, a multidimensional Markov state transition model integrating the transformer area and battery is constructed, and the state transition probabilities under different charging stage strategy actions are obtained. The optimal intelligent decision result for the charging phase strategy is generated based on the state transition probability.

[0011] Furthermore, the optimal intelligent decision result for generating the charging stage strategy based on the state transition probability includes: Based on the Bayesian posterior probability method, the likelihood probability of the state and spatiotemporal characteristics of power quality regulation in the distribution area under different charging stage strategies is calculated. Combined with the prior probability of each charging stage strategy, the posterior probability of each charging stage strategy is generated, and the charging stage strategy with the highest posterior probability is selected as the optimal intelligent decision result.

[0012] Furthermore, the construction of the multi-objective optimization model, and the acquisition of charging control parameters for the power quality regulation equipment based on the multi-objective optimization model and the distributed cooperative control algorithm, includes: Construct a multi-objective optimization model and define a multi-objective optimization objective function for the model; Define sub-objective functions for each optimization sub-objective in the multi-objective optimization objective function, and assign weight coefficients to each sub-objective function; Define the constraints for the multi-objective optimization model to perform the optimization; Distributed solution based on the alternating direction multiplier method is used to obtain the charging control parameters of each power quality regulation device.

[0013] Furthermore, the design of the adaptive law model, which adjusts the charging control parameters in real time and dynamically evaluates the adjustment effect, includes: Collect the state error vector of the power quality control equipment during operation; An adaptive law model for power quality control equipment is designed based on Lyapunov stability theory, and the real-time adjustment results of the charging control parameters are generated by combining the state error vector. Multiple performance evaluation indicators are designed, and the adjustment and operation effects of power quality control equipment are dynamically evaluated based on these indicators.

[0014] The present invention also provides an adaptive collaborative charging system for power quality control equipment, the adaptive collaborative charging system for power quality control equipment being used to implement the above-mentioned adaptive collaborative charging method for power quality control equipment, the system comprising: The data acquisition and preprocessing module is used to acquire power quality control data of the distribution substation based on the sensor network, and to standardize the acquired data to obtain a standardized feature vector. The data feature extraction module is used to extract features from the standardized feature vector to generate a spatiotemporal feature matrix of electrical correlation between power quality control equipment in the power distribution substation. A charging stage strategy generation module is used to construct a state prediction model, predict the state transition trend of power quality regulation in the distribution substation based on the state prediction model, and obtain the optimal charging stage strategy. A charging control parameter generation module is used to construct a multi-objective optimization model and obtain the charging control parameters of the power quality regulation equipment based on the multi-objective optimization model and a distributed cooperative control algorithm. The adjustment and dynamic evaluation module is used to design an adaptive law model, adjust the charging control parameters in real time, and dynamically evaluate the adjustment effect.

[0015] This invention provides an adaptive collaborative charging method and system for power quality control equipment. Through a full-link design encompassing multi-source data acquisition and processing, spatiotemporal feature extraction, state prediction and strategy decision-making, multi-objective optimization and distributed control, and adaptive parameter tuning, it achieves deep collaboration between the charging of power quality control equipment and the operating status of distribution substations. It constructs an integrated substation-battery sensing system based on graph attention networks, combines a multidimensional Markov model and Bayesian posterior probability to achieve intelligent dynamic decision-making for charging strategies, completes distributed solutions for multi-objective optimization through the alternating direction multiplier method, and establishes a real-time parameter adjustment and closed-loop evaluation mechanism based on Lyapunov stability theory. This effectively solves the problems of slow response, single objective, and poor collaboration inherent in traditional methods, significantly improving the accuracy of substation state perception and the adaptability of charging strategies. While ensuring safe and stable charging of equipment and extending battery life, it also considers economic efficiency and system energy efficiency, enhances control real-time performance and system scalability, and significantly improves the renewable energy absorption rate of distribution substations while reducing line losses, providing an efficient technical solution for the construction of intelligent distribution substations. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of the adaptive collaborative charging method of the power quality control device in Embodiment 1 of the present invention; Figure 2 This is a flowchart of the spatiotemporal feature matrix for generating the electrical correlation of devices in Embodiment 1 of the present invention; Figure 3 This is a flowchart of generating a spatiotemporal feature matrix based on nodes and connecting edges in Embodiment 1 of the present invention; Figure 4 This is a flowchart of the strategy for obtaining the optimal charging stage in Embodiment 1 of the present invention; Figure 5This is a flowchart of the process for obtaining charging control parameters of the power quality regulation device in Embodiment 1 of the present invention; Figure 6 This is a flowchart illustrating the real-time adjustment and dynamic evaluation of the adjustment effect of the charging control parameters in Embodiment 1 of the present invention; Figure 7 This is an adaptive collaborative charging system architecture diagram of the power quality regulation device in Embodiment 2 of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] In this invention, it should be understood that terms such as “comprising” or “having” are intended to indicate the presence of features, figures, steps, behaviors, components, portions or combinations thereof disclosed in this specification, and are not intended to exclude the possibility that one or more other features, figures, steps, behaviors, components, portions or combinations thereof are present or added.

[0020] It should also be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0021] In the process of intelligent distribution network and energy transition, distribution substations, as a key link connecting the power grid and user terminals, have been identified as having technical problems such as battery life degradation, thermal runaway risk, and insufficient adaptability to source-load fluctuations due to the complex operating conditions and frequent charging and discharging of their power quality control equipment. Existing charging control methods suffer from reliance on experience in manual strategy setting, leading to response lags; centralized optimization scheduling's high requirements for data prediction accuracy and communication reliability are difficult for actual control systems to meet, making control effects susceptible to interference; and fixed threshold stage switching mechanisms fail to achieve dynamic coordination between battery status and substation operating targets, making it difficult to balance equipment operating safety and economy, thus adversely affecting power supply quality stability, renewable energy absorption efficiency, and equipment lifespan.

[0022] For example, in a distribution substation containing photovoltaic arrays and variable loads, inadequate communication infrastructure is identified, and power quality control equipment can only perform charging operations based on local fixed thresholds. When the output power of the photovoltaic array changes abruptly due to environmental factors, the source-load fluctuations are not detected by the equipment in a timely manner, and the charging strategy is not dynamically adjusted. This leads to the voltage amplitude at critical nodes exceeding the allowable range, the effective value of the feeder current abnormally increasing, and the thermal accumulation phenomenon of the battery pack being exacerbated. In this scenario, the dynamic changes in the battery's state of charge and health are not incorporated into the strategy decision-making, and the equipment continues to operate in a fixed mode during peak load periods, increasing the risk of thermal runaway and weakening the overall system control capability.

[0023] If the above problems are not effectively resolved, the operating status of power quality control equipment will continue to deteriorate. Degraded battery health may lead to premature equipment failure, the accumulation of thermal runaway risks may trigger safety accidents, and insufficient adaptability to source-load fluctuations will weaken the distribution area's ability to absorb renewable energy. Furthermore, the lack of a coordinated control mechanism makes it difficult for distribution areas to adapt to the integrated operation requirements of "source-grid-load-storage," affecting the overall stability and reliability of the power grid and hindering the intelligent transformation of the distribution network.

[0024] Example 1 Embodiment 1 of the present invention provides an adaptive coordinated charging method for a power quality control device, the method comprising: Power quality control data of distribution substations are collected based on sensor networks, and the collected data is standardized to obtain standardized feature vectors. Feature extraction is performed on the standardized feature vector to generate a spatiotemporal feature matrix of electrical correlations between power quality control equipment in the distribution substation. A state prediction model is constructed, and the state transition trend of power quality regulation in the distribution substation is predicted based on the state prediction model, and the optimal charging stage strategy is obtained. A multi-objective optimization model is constructed, and the charging control parameters of the power quality regulation equipment are obtained based on the multi-objective optimization model and a distributed cooperative control algorithm. An adaptive law model is designed to adjust the charging control parameters in real time and to dynamically evaluate the adjustment effect.

[0025] In one optional implementation of this embodiment, such as Figure 1 As shown, Figure 1 The flowchart of the adaptive cooperative charging method of the power quality control device in Embodiment 1 of the present invention is shown, including the following steps: S101. Collect power quality control data of the distribution substation based on sensor network, and standardize the collected data to obtain standardized feature vectors; In one optional implementation of this embodiment, power quality control data in the distribution substation is collected through a sensor network consisting of several sensors deployed in the distribution substation.

[0026] Specifically, key indicators for power quality regulation are defined, and data corresponding to these key indicators are collected. These key indicators include photovoltaic array output power, total active load of the distribution area, voltage amplitude of key nodes, effective value of feeder current, DC side voltage of power quality regulation equipment, charging / discharging current of power quality regulation equipment, average temperature of battery pack, state of charge of battery, and state of health of battery. The key indicators for power quality regulation defined include:

[0027] In the formula, For power quality control data, For the output power of the photovoltaic array, For the total active power load of the transformer area, For the voltage amplitude at critical nodes, This is the effective value of the feeder current. For the DC side voltage of power quality control equipment, The charging / discharging current for power quality control equipment, This represents the average temperature of the battery pack. The battery is in its state of charge. Estimate the battery's health status.

[0028] In one optional implementation of this embodiment, after collecting the power quality control data of the aforementioned distribution substation, the collected data is standardized to obtain a standardized feature vector.

[0029] Specifically, the historical data mean vector is obtained, and the collected raw data is compared with the historical data mean vector through standardization to obtain a dimensionless standardized feature vector. The historical data mean vector refers to the vector formed by statistically analyzing the power quality control data collected over a period of time and calculating the average value of each key indicator. This vector serves as a benchmark for data standardization and can reflect the typical numerical level of various power quality indicators of the distribution substation under normal operating conditions. It can be obtained by batch processing and calculating the historical operating data stored for a long time. After standardization and comparison, dimensionless standardized feature vectors are obtained. These vectors are derived from the standardized dataset, characterized by the removal of physical units from the original data, resulting in consistent numerical ranges or similar statistical properties for each feature. This vector serves as the basis for subsequent feature extraction and model input, ensuring that the algorithm is not affected by dimensions or numerical ranges when processing multi-source heterogeneous data, thereby improving model training efficiency and prediction accuracy.

[0030] Furthermore, the calculation formula for the standardized control treatment includes:

[0031] In the formula, This is a vector of historical data means. This is the standard deviation vector of historical data. It is the numerical stability constant. This is the feature vector after standardization and control processing.

[0032] The aforementioned technical solution clarifies the specific key indicators contained in the power quality control data, making the data acquisition process highly targeted and comprehensive. This refined data acquisition avoids blind spots, ensuring that the acquired data fully reflects the complexity of the distribution substation grid operation and the internal state of the power quality control equipment. When standardizing the power quality control data, using the historical data mean vector as a stable reference benchmark effectively avoids interference from real-time data fluctuations or outliers, thus ensuring the stability and consistency of the standardized feature vector. This processing method unifies power quality control data with different dimensions and numerical ranges to a comparable scale, greatly improving the accuracy of subsequent feature extraction (such as the generation of spatiotemporal feature matrices) and avoiding convergence difficulties or biases caused by inconsistent data scales during model training. Furthermore, it provides high-quality, highly stable input data for the state prediction model and multi-objective optimization model, significantly improving the reliability of model prediction of state transition trends and acquisition of optimal charging stage strategies, as well as the accuracy of optimizing charging control parameters. Ultimately, this enables the adaptive collaborative charging method of the power quality control equipment to operate more robustly and efficiently.

[0033] S102. Extract features from the standardized feature vector to generate a spatiotemporal feature matrix of electrical correlation between power quality control equipment in the power distribution substation. In one optional implementation of this embodiment, such as Figure 2 As shown, Figure 2 The flowchart shown in Embodiment 1 of the present invention illustrates the generation of a spatiotemporal feature matrix of electrical relationships between devices, including the following steps: S201. Define the power quality control equipment in the distribution substation area as a node; In one optional implementation of this embodiment, defining power quality control equipment in a distribution substation as nodes means abstracting all equipment involved in power quality control within the distribution substation, such as energy storage systems, reactive power compensation devices, active power filters, and smart loads, into basic units in a graph structure. These nodes can be single, independent physical devices, or logical units aggregated based on function, geographical location, or electrical characteristics.

[0034] S202. Define the electrical connection relationships between devices as connection edges between nodes; In one optional implementation of this embodiment, defining the electrical connections between devices as connection edges between nodes refers to clarifying the physical or logical interaction paths between these devices, which are defined as nodes. These connection edges represent the flow of electrical energy or information between devices and are key to understanding the overall behavior of the system.

[0035] S203. Generate a spatiotemporal feature matrix based on the defined nodes and connecting edges, and construct an association diagram of power quality control equipment in the distribution substation area.

[0036] In an optional implementation of this embodiment, a structured data representation is formed by combining the nodes and connecting edges defined in steps S201 and S202 with the running data of each node changing over time.

[0037] Specifically, the expression for the power quality control equipment association diagram of the distribution substation is as follows:

[0038] Furthermore, such as Figure 3 As shown, Figure 3 The flowchart illustrating the generation of a spatiotemporal feature matrix based on nodes and connecting edges in Embodiment 1 of the present invention is shown, including the following steps: S301, A graph attention network-based feature extraction algorithm that iteratively assigns attention coefficients to the neighboring nodes of each defined node through a multi-layer network; In one optional implementation of this embodiment, a graph attention network based on a neural network model capable of processing graph structure data with an attention mechanism dynamically assigns different weights (i.e., attention coefficients) according to the feature similarity or importance between nodes. The implementation method is based on multi-layer attention stacking, where each layer further refines the attention coefficients and node features based on the features output by the previous layer, thereby capturing higher-order dependencies.

[0039] S302. Perform weighted aggregation of the features of neighboring nodes and the features of the node itself to obtain the node feature matrix of each layer; In one optional implementation of this embodiment, the weighted sum of neighbor features is concatenated with the features of the node itself, and then processed through a linear transformation layer and a nonlinear activation function; or the weighted sum of neighbor features is directly summed with the features of the node itself, and then transformed.

[0040] S303. After fusing the spatiotemporal correlation of the feature matrices of each layer of nodes, output a high-order spatiotemporal feature matrix.

[0041] In one alternative implementation of this embodiment, the spatial relationships captured by the graph attention network are combined with the dynamic characteristics of the features that change over time to form a comprehensive representation that reflects the “location” and “time” dimensions of the electrical association.

[0042] The structural expression of the association diagram is as follows:

[0043]

[0044]

[0045] In the formula, For the relationship graph expression, For a set of nodes, Let be the set of edges. For the 0th The set of layer nodes corresponds to the standardized feature vectors after comparison processing. For the first The feature matrix of all nodes in the layer, For the first The feature matrix of all nodes in the layer, For the first Layer learnable weight matrix, In the first In the layer, nodes When aggregating information, it is assigned to its neighboring nodes. Attention coefficient For nodes The set of neighboring nodes, For activation function, Given a learnable parameter vector, This represents a vector concatenation operation. This is the activation function used to calculate the attention score.

[0046] The above technical solution dynamically and adaptively captures the non-uniform electrical correlation strength between power quality control devices in a distribution substation, avoiding the limitation of treating all neighboring nodes equally in traditional methods. Weighted aggregation of neighboring node features and the node's own features ensures that each node's feature representation fully integrates its own information and the contextual information of its important neighboring nodes, generating more representative and discriminative node features. By fusing the spatiotemporal correlations of the node feature matrices at each level, a high-order spatiotemporal feature matrix is ​​output, ensuring that the extracted features not only include the spatial topological relationships between devices but also incorporate the dynamic characteristics of these relationships over time, providing a more comprehensive and accurate input for subsequent state prediction and optimization. This refined feature extraction method significantly improves the depth and accuracy of understanding the electrical correlations of power quality control devices in a distribution substation, providing high-quality input for subsequent state prediction models, thereby improving the prediction accuracy and control performance of the entire adaptive cooperative charging method.

[0047] S103. Construct a state prediction model, predict the state transition trend of power quality regulation in the distribution substation based on the state prediction model, and obtain the optimal charging stage strategy. In one optional implementation of this embodiment, such as Figure 4 As shown, Figure 4 A flowchart illustrating the optimal charging stage strategy in Embodiment 1 of the present invention is shown, including the following steps: S401. Extract the state vector that reflects the state transition trend of power quality regulation in the distribution substation area; In an optional implementation of this embodiment, the state vector reflecting the state transition trend of power quality regulation in the distribution substation includes battery state of charge, net power of the substation, grid voltage deviation index, and average battery pack temperature, expressed as follows:

[0048] In the formula, for The state vector at time t, The battery is in its state of charge. Net power of the transformer area ( ), This is an indicator of grid voltage deviation. This represents the average temperature of the battery pack.

[0049] S402. Construct a multidimensional Markov state transition model for the fusion of the transformer area and battery based on the extracted state vector, and obtain the state transition probabilities under different charging stage strategy actions; In one optional implementation of this embodiment, the multidimensional Markov state transition model of the distribution substation-battery fusion is a probabilistic model used to describe the probability of the power quality control system of the distribution substation transitioning from one state to another under the action of strategies at different charging stages. The "Markov" property means that the future state of the system depends only on the current state and is independent of past states. "Multidimensional" indicates that the model considers multiple dimensions of information in the state vector. "Distribution substation-battery fusion" emphasizes that the model simultaneously considers the overall operating status of the distribution substation and the internal state of the power quality control equipment (especially the battery).

[0050] Specifically, the expression for the multidimensional Markov state transition model is:

[0051]

[0052] In the formula, for The state vector at time t, The action space represents different charging phase strategies. Indicates the action Next from state Transferred to The probability, Let be the state transition probability tensor.

[0053] S403. Generate the optimal intelligent decision result of the charging stage strategy based on the state transition probability.

[0054] In one optional implementation of this embodiment, after obtaining the state transition probabilities of different charging stage strategies, based on the Bayesian posterior probability method, the likelihood probabilities of the state and spatiotemporal characteristics of power quality control in the distribution area under different charging stage strategies are calculated. Combined with the prior probabilities of each charging stage strategy, the posterior probabilities of each charging stage strategy are generated, and the charging stage strategy with the highest posterior probability is selected as the optimal intelligent decision result.

[0055] Specifically, the expression is:

[0056] In the formula, Indicates the first Each charging stage Representation phase The prior probability, Representing state In the stage The likelihood probability is as follows: Representation of graph features In the stage The likelihood probability is as follows: Representation phase The posterior probability.

[0057] After the above calculations, the charging stage strategy with the highest posterior probability is selected as the optimal intelligent decision result.

[0058] The above technical solution effectively addresses the problems of inaccurate state prediction and poor adaptability of charging strategies in power quality control systems for distribution substations. By extracting refined state vectors, the operating characteristics of distribution substations and power quality control equipment can be captured comprehensively and accurately. Furthermore, a multidimensional Markov state transition model integrating substations and batteries is constructed, enabling the system to understand its dynamic evolution in a probabilistic manner, overcoming the limitations of traditional prediction methods in handling multi-factor coupling and uncertainty. The optimal intelligent decision-making results generated based on this model ensure that power quality control equipment can always adopt the most appropriate charging stage strategy in complex and ever-changing operating environments, thereby significantly improving the intelligence, foresight, and optimization level of the charging strategy, effectively guaranteeing the power quality of the distribution substation, extending equipment lifespan, and improving energy utilization efficiency.

[0059] S104. Construct a multi-objective optimization model, and obtain the charging control parameters of the power quality regulation equipment based on the multi-objective optimization model and the distributed cooperative control algorithm. In one optional implementation of this embodiment, such as Figure 5 As shown, Figure 5 A flowchart illustrating the acquisition of charging control parameters for a power quality regulation device according to Embodiment 1 of the present invention is shown, including the following steps: S501. Construct a multi-objective optimization model and define a multi-objective optimization objective function for the model; In an optional implementation of this embodiment, a multi-objective optimization model is constructed that comprehensively considers economy, safety, equipment lifespan, and energy efficiency. The expression for the multi-objective optimization objective function is defined as follows:

[0060] In the formula, For the economic sub-objective function, For the security sub-objective function, Let the equipment lifespan be the sub-objective function. Let be the energy efficiency sub-objective function. , , , These are the weighting coefficients for each sub-objective function.

[0061] S502. Define the sub-objective functions of each optimization sub-objective in the multi-objective optimization objective function, and assign weight coefficients to each sub-objective function; In an optional implementation of this embodiment, the economic sub-objective function is... Safety sub-objective function Equipment lifespan sub-objective function and energy efficiency sub-objective function Define the objectives and assign weight coefficients to each sub-objective function. , , , The specific value.

[0062] The expressions for each sub-objective function are: Economic sub-objective function :

[0063] In the formula, For time-of-use electricity pricing, For the power exchange between the distribution area and the main network, Battery degradation cost coefficient, For battery degradation increment, To optimize the time domain length.

[0064] Security sub-objective function :

[0065] In the formula, This is the voltage deviation penalty coefficient. for The actual voltage at the grid node at any given time For reference voltage, This is the current deviation penalty coefficient. for The actual current at the grid node at any given time The rated charging current for the battery.

[0066] Equipment lifespan sub-objective function :

[0067] In the formula, This is the penalty coefficient for the deviation of the state of charge. for The actual state of charge of the battery at all times. This represents the optimal state of charge for the battery. This is the temperature deviation penalty coefficient. for Real-time battery temperature This is the optimal operating temperature for the battery.

[0068] Energy efficiency sub-objective function :

[0069] In the formula, This is a penalty factor for charging efficiency. for Real-time charging efficiency This is the power loss penalty factor. for Power loss at any moment.

[0070] S503. Define the constraints for the multi-objective optimization model to perform optimization; In one optional implementation of this embodiment, the constraints include:

[0071]

[0072]

[0073]

[0074] S504. Distributed solution based on alternating direction multiplier method to obtain charging control parameters of each power quality regulation device.

[0075] In one optional implementation of this embodiment, a distributed solution is performed based on the alternating direction multiplier method:

[0076] st .

[0077] The iterative update process includes:

[0078]

[0079]

[0080] In the formula, For the first The local objective function of each energy storage unit. For global consistency constraint functions, , For the first The device in the The optimal local decision for the next iteration is obtained by minimizing the local solution. For the first The unit in the first The dual variable of the next iteration This is the penalty parameter.

[0081] By employing the aforementioned technical solution, based on predicting the state transition trend of power quality regulation in distribution substations and obtaining the optimal charging stage strategy, a multi-objective optimization model is further constructed and a distributed solution method is adopted. This effectively solves the problem that single-objective optimization cannot simultaneously address multiple operational needs. This solution defines a multi-objective optimization objective function and its sub-objective functions and assigns weight coefficients, enabling the determination of charging control parameters for power quality regulation equipment to comprehensively consider multiple conflicting performance indicators such as economy, power quality improvement effect, and equipment lifespan, thus avoiding a situation where one aspect is sacrificed for another. Simultaneously, by defining strict constraints, the safety and feasibility of the obtained charging control parameters in actual operation are ensured. Most importantly, the distributed solution based on the alternating direction multiplier method allows multiple power quality regulation devices to achieve efficient collaborative optimization without relying on a powerful central processing unit, significantly reducing communication burden and computational complexity, and improving the system's flexibility and robustness in responding to complex grid environment changes. This enables power quality regulation equipment to charge more intelligently and efficiently, thereby maximizing system benefits while ensuring stable grid operation.

[0082] S105. Design an adaptive law model to adjust the charging control parameters in real time and dynamically evaluate the adjustment effect.

[0083] In one optional implementation of this embodiment, such as Figure 6 As shown, Figure 6 The flowchart illustrates the real-time adjustment and dynamic evaluation of the adjustment effect of charging control parameters in Embodiment 1 of the present invention, including the following steps: S601, Collect the state error vector of the power quality control equipment during operation; In one optional implementation of this embodiment, the state error vector of the power quality control device during operation is collected, such as the deviation between the actual output voltage and the target voltage, the deviation between the actual current and the target current, and the deviation between the battery state of charge and the desired state of charge.

[0084] S602. Based on Lyapunov stability theory, design an adaptive law model for power quality control equipment, and combine the state error vector to generate the real-time adjustment result of the charging control parameters; In one optional implementation of this embodiment, an adaptive law model for the power quality control device is designed based on Lyapunov stability theory, and the expression includes:

[0085]

[0086] In the formula, Let be the state error vector. It is a positive definite symmetric matrix. For parameter estimation error, For adaptive gain, For the input matrix, This is the regression vector.

[0087] S603. Design multiple performance evaluation indicators, and dynamically evaluate the adjustment and operation effect of power quality control equipment based on the performance evaluation indicators.

[0088] In an optional implementation of this embodiment, the performance evaluation indicators include three main indicators: overall system energy efficiency, battery capacity degradation rate, and system response rate, and their expressions include:

[0089]

[0090]

[0091] In the formula, For overall system energy efficiency, for Available energy during the period for Energy input during specific time periods Battery capacity degradation rate, For the initial capacity, Current available capacity For system response rate, , These represent the times when the response reaches 90% and 10%, respectively. The time when the instruction was issued.

[0092] The above technical solution effectively solves the problem that charging control parameters of power quality control equipment in distribution substations are difficult to continuously adapt to actual operating conditions under dynamic environments. By real-time acquisition of state error vectors and designing an adaptive law model based on Lyapunov stability theory, the charging control parameters of the power quality control equipment can be adjusted in real-time, stably, and accurately according to changes in actual operating conditions. This ensures that even under complex conditions such as load fluctuations, uncertain renewable energy output, or changes in equipment characteristics, the charging strategy can remain optimal or near-optimal, thus significantly improving the real-time response capability and robustness of power quality control. Simultaneously, by designing multiple performance evaluation indicators and conducting dynamic evaluations, the system can continuously monitor the adjustment effect, providing feedback for further optimization of the adaptive law model, forming a self-learning and self-improving closed-loop control system. This not only guarantees the long-term effectiveness of power quality control but also optimizes the battery charging and discharging process, extends equipment lifespan, and reduces overall operating costs, thereby achieving more efficient and economical energy management while maintaining stable power quality in the distribution substation.

[0093] In summary, Embodiment 1 of this invention provides an adaptive collaborative charging method for power quality control equipment. Through a full-link design encompassing multi-source data acquisition and processing, spatiotemporal feature extraction, state prediction and strategy decision-making, multi-objective optimization and distributed control, and adaptive parameter tuning, it achieves deep collaboration between the charging of power quality control equipment and the operating status of distribution substations. It constructs an integrated perception system for substations and batteries based on graph attention networks, combines a multidimensional Markov model with Bayesian posterior probability to achieve intelligent dynamic decision-making for charging strategies, completes distributed solutions for multi-objective optimization using the alternating direction multiplier method, and establishes a real-time parameter adjustment and closed-loop evaluation mechanism based on Lyapunov stability theory. This effectively solves the problems of slow response, single objective, and poor collaboration inherent in traditional methods, significantly improving the accuracy of substation state perception and the adaptability of charging strategies. While ensuring safe and stable charging of equipment and extending battery life, it also considers economic efficiency and system energy efficiency, enhances control real-time performance and system scalability, and significantly improves the renewable energy absorption rate of distribution substations while reducing line losses, providing an efficient technical solution for the construction of intelligent distribution substations.

[0094] Example 2 Embodiment 2 of the present invention provides an adaptive collaborative charging system for power quality control equipment. The adaptive collaborative charging system for power quality control equipment is used to implement the adaptive collaborative charging method for power quality control equipment in Embodiment 1. The system includes a data acquisition and preprocessing module, a data feature extraction module, a charging stage strategy generation module, a charging control parameter generation module, and an adjustment and dynamic evaluation module.

[0095] In one optional implementation of this embodiment, such as Figure 7As shown, Figure 7 The diagram shows the adaptive collaborative charging system architecture of the power quality control device in Embodiment 2 of the present invention, which includes the following modules: The data acquisition and preprocessing module 10 is used to acquire power quality control data of the distribution substation based on the sensor network, and to perform standardized processing on the acquired data to obtain standardized feature vectors. In an optional implementation of this embodiment, the power quality control data includes: Define key indicators for power quality control and collect data corresponding to these key indicators. These key indicators include photovoltaic array output power, total active load of the distribution area, voltage amplitude of key nodes, effective value of feeder current, DC side voltage of power quality control equipment, charging / discharging current of power quality control equipment, average temperature of battery pack, state of charge of battery, and state of health of battery.

[0096] In an optional implementation of this embodiment, the standardization process of the collected data to obtain a standardized feature vector includes: Obtain the historical data mean vector, and then perform a standardized comparison process between the collected raw data and the historical data mean vector to obtain a dimensionless standardized feature vector.

[0097] The data feature extraction module 20 is used to extract features from the standardized feature vector to generate a spatiotemporal feature matrix of electrical correlation between power quality control equipment in the power distribution substation. In an optional implementation of this embodiment, the step of extracting features from the standardized feature vector to generate a spatiotemporal feature matrix of electrical correlations of power quality control equipment in a distribution substation includes: Define the power quality control equipment in the distribution substation as nodes; Define the electrical connections between devices as connection edges between nodes; A spatiotemporal feature matrix is ​​generated based on the defined nodes and connecting edges, and an association graph of power quality control equipment in the distribution substation is constructed.

[0098] In an optional implementation of this embodiment, generating the spatiotemporal feature matrix based on the defined nodes and connecting edges includes: Based on the graph attention network feature extraction algorithm, attention coefficients are assigned to the neighboring nodes of each defined node through multi-layer network iteration; We perform weighted aggregation of the features of neighboring nodes and the features of the node itself to obtain the node feature matrix of each layer; After fusing the spatiotemporal correlation of the feature matrices of each layer of nodes, a high-order spatiotemporal feature matrix is ​​output.

[0099] The charging stage strategy generation module 30 is used to construct a state prediction model, predict the state transition trend of power quality regulation in the distribution substation based on the state prediction model, and obtain the optimal charging stage strategy. In an optional implementation of this embodiment, the step of constructing a state prediction model, predicting the state transition trend of power quality regulation in the distribution substation based on the state prediction model, and obtaining the optimal charging stage strategy includes: Extract the state vector that reflects the state transition trend of power quality regulation in the distribution substation area; Based on the extracted state vectors, a multidimensional Markov state transition model integrating the transformer area and battery is constructed, and the state transition probabilities under different charging stage strategy actions are obtained. The optimal intelligent decision result for the charging phase strategy is generated based on the state transition probability.

[0100] In an optional implementation of this embodiment, the optimal intelligent decision result for generating the charging stage strategy based on the state transition probability includes: Based on the Bayesian posterior probability method, the likelihood probability of the state and spatiotemporal characteristics of power quality regulation in the distribution area under different charging stage strategies is calculated. Combined with the prior probability of each charging stage strategy, the posterior probability of each charging stage strategy is generated, and the charging stage strategy with the highest posterior probability is selected as the optimal intelligent decision result.

[0101] A charging control parameter generation module 40 is used to construct a multi-objective optimization model and obtain the charging control parameters of the power quality regulation device based on the multi-objective optimization model and a distributed cooperative control algorithm. In an optional implementation of this embodiment, the step of constructing a multi-objective optimization model and obtaining the charging control parameters of the power quality regulation equipment based on the multi-objective optimization model and a distributed cooperative control algorithm includes: Construct a multi-objective optimization model and define a multi-objective optimization objective function for the model; Define sub-objective functions for each optimization sub-objective in the multi-objective optimization objective function, and assign weight coefficients to each sub-objective function; Define the constraints for the multi-objective optimization model to perform the optimization; Distributed solution based on the alternating direction multiplier method is used to obtain the charging control parameters of each power quality regulation device.

[0102] The adjustment and dynamic evaluation module 50 is used to design an adaptive law model, adjust the charging control parameters in real time, and dynamically evaluate the adjustment effect.

[0103] In an optional implementation of this embodiment, the design of the adaptive law model, the real-time adjustment of the charging control parameters, and the dynamic evaluation of the adjustment effect include: Collect the state error vector of the power quality control equipment during operation; An adaptive law model for power quality control equipment is designed based on Lyapunov stability theory, and the real-time adjustment results of the charging control parameters are generated by combining the state error vector. Multiple performance evaluation indicators are designed, and the adjustment and operation effects of power quality control equipment are dynamically evaluated based on these indicators.

[0104] In summary, Embodiment 2 of this invention provides an adaptive collaborative charging system for power quality control equipment, which implements the adaptive collaborative charging method for power quality control equipment in Embodiment 1. Through a full-link design encompassing multi-source data acquisition and processing, spatiotemporal feature extraction, state prediction and strategy decision-making, multi-objective optimization and distributed control, and adaptive parameter tuning, it achieves deep collaboration between the charging of power quality control equipment and the operating status of distribution substations. It constructs an integrated perception system for substations and batteries based on graph attention networks, combines a multidimensional Markov model and Bayesian posterior probability to achieve intelligent dynamic decision-making for charging strategies, completes distributed solutions for multi-objective optimization through the alternating direction multiplier method, and establishes a real-time parameter adjustment and closed-loop evaluation mechanism based on Lyapunov stability theory. This effectively solves the problems of slow response, single objective, and poor collaboration in traditional methods, significantly improving the accuracy of substation state perception and the adaptability of charging strategies. While ensuring safe and stable charging of equipment and extending battery life, it also considers economic efficiency and system energy efficiency, enhances control real-time performance and system scalability, and significantly improves the renewable energy absorption rate of distribution substations while reducing line losses, providing an efficient technical solution for the construction of intelligent distribution substations.

[0105] The above provides a detailed description of the adaptive coordinated charging method and system for power quality control equipment provided by the present invention. Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0106] Furthermore, the embodiments of the present invention have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. An adaptive coordinated charging method for power quality control equipment, characterized in that, The method includes: Power quality control data of distribution substations are collected based on sensor networks, and the collected data is standardized to obtain standardized feature vectors. Feature extraction is performed on the standardized feature vector to generate a spatiotemporal feature matrix of electrical correlations between power quality control equipment in the distribution substation. A state prediction model is constructed, and the state transition trend of power quality regulation in the distribution substation is predicted based on the state prediction model, and the optimal charging stage strategy is obtained. A multi-objective optimization model is constructed, and the charging control parameters of the power quality regulation equipment are obtained based on the multi-objective optimization model and a distributed cooperative control algorithm. An adaptive law model is designed to adjust the charging control parameters in real time and to dynamically evaluate the adjustment effect.

2. The adaptive coordinated charging method for power quality control equipment as described in claim 1, characterized in that, The power quality control data includes: Define key indicators for power quality control and collect data corresponding to these key indicators. These key indicators include photovoltaic array output power, total active load of the distribution area, voltage amplitude of key nodes, effective value of feeder current, DC side voltage of power quality control equipment, charging / discharging current of power quality control equipment, average temperature of battery pack, state of charge of battery, and state of health of battery.

3. The adaptive coordinated charging method for power quality control equipment as described in claim 1, characterized in that, The standardization process for the collected data to obtain standardized feature vectors includes: Obtain the historical data mean vector, and then perform a standardized comparison process between the collected raw data and the historical data mean vector to obtain a dimensionless standardized feature vector.

4. The adaptive coordinated charging method for power quality control equipment as described in claim 1, characterized in that, The step of extracting features from the standardized feature vector to generate a spatiotemporal feature matrix of electrical correlations between power quality control equipment in a distribution substation includes: Define the power quality control equipment in the distribution substation as nodes; Define the electrical connections between devices as connection edges between nodes; A spatiotemporal feature matrix is ​​generated based on the defined nodes and connecting edges, and an association graph of power quality control equipment in the distribution substation is constructed.

5. The adaptive coordinated charging method for power quality control equipment as described in claim 4, characterized in that, The spatiotemporal feature matrix generated based on the defined nodes and connecting edges includes: Based on the graph attention network feature extraction algorithm, attention coefficients are assigned to the neighboring nodes of each defined node through multi-layer network iteration; We perform weighted aggregation of the features of neighboring nodes and the features of the node itself to obtain the node feature matrix of each layer; After fusing the spatiotemporal correlation of the feature matrices of each layer of nodes, a high-order spatiotemporal feature matrix is ​​output.

6. The adaptive coordinated charging method for power quality control equipment as described in claim 1, characterized in that, The construction of the state prediction model, based on which the state prediction model is used to predict the state transition trend of power quality regulation in the distribution substation, and to obtain the optimal charging stage strategy, includes: Extract the state vector that reflects the state transition trend of power quality regulation in the distribution substation area; Based on the extracted state vectors, a multidimensional Markov state transition model integrating the transformer area and battery is constructed, and the state transition probabilities under different charging stage strategy actions are obtained. The optimal intelligent decision result for the charging phase strategy is generated based on the state transition probability.

7. The adaptive cooperative charging method for power quality control equipment as described in claim 6, characterized in that, The optimal intelligent decision result for generating the charging stage strategy based on the state transition probability includes: Based on the Bayesian posterior probability method, the likelihood probability of the state and spatiotemporal characteristics of power quality regulation in the distribution area under different charging stage strategies is calculated. Combined with the prior probability of each charging stage strategy, the posterior probability of each charging stage strategy is generated, and the charging stage strategy with the highest posterior probability is selected as the optimal intelligent decision result.

8. The adaptive coordinated charging method for power quality control equipment as described in claim 1, characterized in that, The construction of a multi-objective optimization model, and the acquisition of charging control parameters for power quality regulation equipment based on the multi-objective optimization model and a distributed cooperative control algorithm, include: Construct a multi-objective optimization model and define a multi-objective optimization objective function for the model; Define sub-objective functions for each optimization sub-objective in the multi-objective optimization objective function, and assign weight coefficients to each sub-objective function; Define the constraints for the multi-objective optimization model to perform the optimization; Distributed solution based on the alternating direction multiplier method is used to obtain the charging control parameters of each power quality regulation device.

9. The adaptive coordinated charging method for power quality control equipment as described in claim 1, characterized in that, The adaptive law model is designed to adjust the charging control parameters in real time and dynamically evaluate the adjustment effect, including: Collect the state error vector of the power quality control equipment during operation; An adaptive law model for power quality control equipment is designed based on Lyapunov stability theory, and the real-time adjustment results of the charging control parameters are generated by combining the state error vector. Multiple performance evaluation indicators are designed, and the adjustment and operation effects of power quality control equipment are dynamically evaluated based on these indicators.

10. An adaptive cooperative charging system for power quality control equipment, characterized in that, The adaptive collaborative charging system of the power quality control equipment is used to implement the adaptive collaborative charging method of the power quality control equipment according to any one of claims 1-9, the system comprising: The data acquisition and preprocessing module is used to acquire power quality control data of the distribution substation based on the sensor network, and to standardize the acquired data to obtain a standardized feature vector. The data feature extraction module is used to extract features from the standardized feature vector to generate a spatiotemporal feature matrix of electrical correlation between power quality control equipment in the power distribution substation. A charging stage strategy generation module is used to construct a state prediction model, predict the state transition trend of power quality regulation in the distribution substation based on the state prediction model, and obtain the optimal charging stage strategy. A charging control parameter generation module is used to construct a multi-objective optimization model and obtain the charging control parameters of the power quality regulation equipment based on the multi-objective optimization model and a distributed cooperative control algorithm. The adjustment and dynamic evaluation module is used to design an adaptive law model, adjust the charging control parameters in real time, and dynamically evaluate the adjustment effect.

Images