Distribution network substation voltage control method, electronic equipment and medium

Abstract

The invention discloses a distribution network substation voltage control method, electronic equipment and a medium, and the method comprises the steps: constructing an IBDBN-DGAM network which comprises an input layer, an IBDBN module, a DGAM module and a decision layer; wherein the IBDBN module is used for carrying out deep feature extraction and probability modeling on input data based on a multilayer Bayesian network structure, and the DGAM module is used for dynamically adjusting the weight of a network topology structure by utilizing a dynamic graph attention mechanism based on feature representation output by the IBDBN; historical multi-modal data of a distribution network substation is used as input data to be input into an IBDBN-DGAM network; an objective function is constructed to optimize the IBDBN-DGAM network; wherein the multi-modal data of the distribution network substation comprises voltage, current and load data from an SCADA (supervisory control and data acquisition) system and a PMU (power management unit); and acquiring real-time multi-modal data of the distribution network substation, inputting the real-time multi-modal data into the optimized IBDBN-DGAM network, and outputting to obtain a voltage control instruction.

Description

Technical Field

[0001] The present invention relates to the field of electrical engineering technology, and in particular to a voltage control method, electronic equipment and medium for a distribution network substation. Background Art

[0002] In the distribution network environment, voltage control faces many challenges. First, the observability of the distribution network is low, resulting in incomplete and inaccurate system status information, which increases the difficulty of voltage control. Secondly, the access of a large number of distributed energy sources and controllable loads makes the topology and operating status of the distribution network dynamically changing, and traditional static control methods are difficult to adapt to this complexity. In addition, there are a large number of uncertain factors in the distribution network, such as load fluctuations, fluctuations in the output of new energy sources, etc. These uncertainties bring huge challenges to accurate voltage prediction and control.

[0003] Existing distribution network voltage control methods mainly include traditional rule-based control methods, optimization algorithm-based control methods, and machine learning-based control methods. However, these methods have obvious shortcomings in dealing with problems such as low observability, dynamic changes, and uncertainty. Traditional rule-based methods are difficult to adapt to the complex and changeable distribution network environment; methods based on optimization algorithms have high computational complexity and are difficult to achieve real-time control; and although existing machine learning methods have certain learning capabilities, they still need to be improved in dealing with incomplete data and dynamic system modeling. Summary of the invention

[0004] In view of this, the present invention provides a distribution network substation voltage control method, electronic equipment, and medium.

[0005] In a first aspect, an embodiment of the present invention provides a distribution network substation voltage control method, the method comprising:

[0006] Constructing an IBDBN-DGAM network, the IBDBN-DGAM network comprising an input layer, an IBDBN module, a DGAM module, and a decision layer; wherein the IBDBN module performs deep feature extraction and probability modeling on the input data based on a multi-layer Bayesian network structure, and the DGAM module dynamically adjusts the weight of the network topology structure based on the feature representation output by the IBDBN using a dynamic graph attention mechanism;

[0007] The historical multimodal data of the distribution substation is used as input data and input into the IBDBN-DGAM network; an objective function is constructed to optimize the IBDBN-DGAM network; wherein the multimodal data of the distribution substation includes voltage, current, and load data from the SCADA system and PMU unit;

[0008] The real-time multimodal data of the distribution substation is obtained and input into the optimized IBDBN-DGAM network, and the voltage control command is output.

[0009] In a second aspect, an embodiment of the present invention provides an electronic device, comprising a memory and a processor, wherein the memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the above-mentioned distribution network substation voltage control method.

[0010] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the above-mentioned distribution network substation voltage control method.

[0011] In a fourth aspect, an embodiment of the present invention provides a computer program product, including a computer program / instruction, which implements the above-mentioned distribution network substation voltage control method when executed by a processor.

[0012] Compared with the prior art, the present invention has the following beneficial effects:

[0013] The present invention provides a distribution network substation voltage control method, and the present invention constructs a distribution network voltage control framework based on an improved Bayesian deep belief network (IBDBN) and a dynamic graph attention mechanism (DGAM). The framework effectively solves the key technical problems in distribution network voltage control by integrating the uncertainty modeling capability of IBDBN and the dynamic graph representation capability of DGAM, and provides a new technical approach to improve the intelligent level of distribution network operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0015] Figure 1 A flow chart of a distribution network substation voltage control method provided by an embodiment of the present invention;

[0016] Figure 2 A comparison diagram of the 24-hour voltage control effect of an actual distribution network provided by an embodiment of the present invention;

[0017] Figure 3 A schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0018] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of methods consistent with some aspects of the present invention as detailed in the appended claims.

[0019] The terms used in the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The singular forms "a", "the" and "the" used in the present invention and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0020] like Figure 1 As shown, the present invention provides a distribution network substation voltage control method, the method comprising the following steps:

[0021] Step S1, constructing an IBDBN-DGAM network, wherein the IBDBN-DGAM network includes an input layer, an IBDBN module, a DGAM module, and a decision layer; wherein the IBDBN module performs deep feature extraction and probabilistic modeling on the input data based on a multi-layer Bayesian network structure, and the DGAM module dynamically adjusts the weight of the network topology structure based on the feature representation of the IBDBN output using a dynamic graph attention mechanism.

[0022] Step S2, inputting the historical multimodal data of the distribution substation as input data into the IBDBN-DGAM network; constructing an objective function to optimize the IBDBN-DGAM network; wherein the multimodal data of the distribution substation includes voltage, current, and load data from the SCADA system (Supervisory Control And Data Acquisition) and the PMU unit (Power Management Unit).

[0023] Furthermore, the step S2 further includes: preprocessing the historical multimodal data of the distribution network substation including data cleaning, missing data interpolation, and feature engineering;

[0024] Among them, data cleaning: removing outliers and redundant data; missing data interpolation: estimating missing values ​​through conditional probability distribution based on the probabilistic inference capability of the IBDBN model; feature engineering: extracting and constructing key features that can reflect the dynamic characteristics of the distribution network.

[0025] Set the preprocessing evaluation index to evaluate the effect of data preprocessing. The expression is as follows:

[0026] Q data =α·Completeness+β·Accuracy+γ·Consistency

[0027] In the formula, α, β, and γ are weight coefficients, representing the importance of data completeness, accuracy, and consistency, respectively.

[0028] Furthermore, the expression of the objective function is as follows:

[0029]

[0030] Where J(θ,π) represents the objective function value, which indicates the total loss that the IBDBN-DGAM framework needs to minimize; θ represents the model parameters, which are related to the training of IBDBN (Bayesian Deep Belief Network) and are responsible for feature extraction and uncertainty processing; π represents the strategy parameters, which are related to DGAM (Dynamic Graph Attention Mechanism) and are used to dynamically capture the topology changes of the distribution network; represents the expectation calculated under the data distribution p(x,y), indicating that the main optimization direction of the objective function is based on the data distribution; p(y|x; θ) represents the conditional probability distribution given the input x and parameter θ, which is used to predict the probability of output y; p(y) represents the prior probability of the output variable y; β represents the weight factor of the KL divergence, which is used to control the strength of the similarity constraint between the potential distribution p(z|x; θ) and the prior distribution p(z); KL(·) represents the KL divergence, KL(p(z|x; θ)||p(z)) represents the Kullback-Leibler divergence between the potential variable distribution p(z|x; θ) and the prior distribution p(z), which is used to measure the difference between the distribution generated by the model and the true distribution; λ represents the weight factor of the cumulative reward, which determines the trade-off between short-term and long-term benefits; γ represents the discount factor, which represents the impact of the current decision on future benefits; t∈T, T represents the time step; Cumulative reward, r(s t ,a t ) means in state s t Execute action a t The immediate reward obtained; η represents the weight factor of control and perturbation cost, which is used to weigh the cost optimization in the worst case; represents the worst-case disturbance control problem, C(u) is the control cost, and D(Δ) is the disturbance cost.

[0031] It should be noted that this objective function reflects the multi-objective optimization characteristics of the framework, including maximizing data likelihood, minimizing KL divergence, maximizing cumulative rewards, and minimizing the worst-case control cost. The training process adopts an alternating optimization strategy to update the parameters of IBDBN and DGAM respectively.

[0032] Furthermore, the optimization strategy is the key to ensuring the performance and generalization ability of the model. The optimization process of the IBDBN-DGAM network also includes: optimizing the IBDBN-DGAM network using optimization strategies such as adaptive learning rate adjustment, batch normalization, regularization, and adversarial training.

[0033] Among them, adaptive learning rate adjustment: dynamically adjust the learning rate according to gradient changes; batch normalization: accelerate the training process and improve model stability; regularization: L1 / L2 regularization and Dropout technology prevent overfitting; adversarial training: enhance the model's robustness to perturbations.

[0034] Set the optimization evaluation index to evaluate the optimization effect of the IBDBN-DGAM network. The expression is as follows:

[0035]

[0036] Where N is the number of evaluation samples, and w1, w2, and w3 are the weights of each performance indicator.

[0037] Step S3, obtaining real-time multimodal data of the distribution network substation and inputting it into the optimized IBDBN-DGAM network, and outputting a voltage control instruction.

[0038] Furthermore, in this example, the following key factors need to be considered during the deployment of the IBDBN-DGAM network in the distribution substation:

[0039] Real-time: Stream computing technology is used to ensure that the model can quickly respond to changes in network status.

[0040] Scalability: The modular architecture is designed to facilitate deployment in distribution networks of different sizes.

[0041] Fault-tolerant mechanism: Introduce redundant design and fault detection algorithm to improve system reliability.

[0042] Security: Implement data encryption and access controls to protect sensitive information.

[0043] Furthermore, in this example, a network deployment evaluation index is set to evaluate the effect of IBDBN-DGAM network deployment in the distribution substation, and the expression is as follows:

[0044] ResponseTime = t decision +texecution

[0045]

[0046] Where, t decision represents the decision time, t execution Indicates execution time.

[0047] It should be noted that in this example, in the decision-making stage, the fusion layer integrates the outputs of the IBDBN and DGAM modules and comprehensively utilizes the advantages of the two modules. The decision layer generates the final voltage control strategy based on the output of the fusion layer and provides a specific control solution for the distribution network.

[0048] In summary, the present invention provides a distribution network substation voltage control method. By introducing a probabilistic graph autoencoder structure, IBDBN significantly improves the state estimation capability in the presence of uncertainty and data loss. This improvement enables the model to more accurately reflect the actual operating state of the distribution network, thereby providing reliable data support for voltage control. Secondly, DGAM adaptively captures the time-varying characteristics of the network topology through a multi-head attention mechanism and a time convolutional network, thereby improving the modeling accuracy of the dynamic system. This mechanism ensures that the control strategy can respond to changes in the state of the distribution network in a timely manner and optimize the execution effect of the control action. Specifically, DGAM can dynamically adjust the attention weights, thereby focusing on key nodes and connections at different time steps, and improving the adaptability and flexibility of the model. Finally, the deep fusion of IBDBN and DGAM not only realizes multi-scale probabilistic graph representation learning, adaptive learning of dynamic graph attention, uncertainty-aware voltage control decision-making, low observability adaptation transfer learning, and decision support with enhanced interpretability, but also enhances the robustness and adaptability of the system through uncertainty-aware voltage control decision-making and low observability adaptation transfer learning. IBDBN provides a solid foundation for state estimation, enabling DGAM to more accurately capture the operating characteristics of the distribution network in a dynamic environment. In addition, the interpretability enhancement part of the framework facilitates operation and maintenance personnel to understand and trust the control strategy through transparent decision support, thereby enhancing the practical application value of the system.

[0049] like Figure 2As shown in the main figure, the blue curve represents the proposed method, the orange curve represents the conventional method, and the green dotted line represents the ideal voltage range (0.95-1.05pu). The overall trend shows that the voltage distribution of this method is smoother, with smaller fluctuations, and it remains within the ideal range most of the time. In contrast, the voltage curve of the conventional method fluctuates greatly, especially during the peak and valley periods of load, with obvious overcurrent. During the low load period (about 4-6 hours), the voltage of this method is as low as 0.97pu, while the voltage of the conventional method is as low as 0.94pu, with a difference of 0.03pu. During the low load period (about 4-6 hours), the voltage of this method is as low as 0.97pu. During the peak load period (about 9 to 11 hours), the maximum voltage of this method is 1.03pu, while the maximum voltage of the conventional method is 1.06pu, with a difference of 0.03pu. The results show that this method has better voltage stabilization performance under extreme load conditions. The voltage distribution during the peak and valley load period is further shown in detail. ( Figure 2 (a) and Figure 2 (b) in the figure shows the voltage distribution heat map of the new method and the conventional method during the valley load period. It can be observed that the voltage distribution of the proposed method is more concentrated and the color is more uniform, indicating that the voltage control is more stable. Figure 2 4(c) Figure 2 (d) in Figure 1 is a heat map of voltage distribution during peak load. Again, the proposed technique shows better voltage aggregation, while the heat map of the traditional method shows greater color changes, reflecting greater voltage fluctuations. The heat map analysis shows that the proposed technique maintains good voltage stability under different load conditions, while the traditional method is prone to voltage over-limit problems under extreme load conditions. This result fully demonstrates the superiority of the proposed technique in complex load environments and provides an effective solution for voltage control in low-observation distribution networks.

[0050] Accordingly, the present application also provides an electronic device, comprising: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned distribution network substation voltage control method based on the improved Bayesian deep belief network. Figure 3 As shown, a hardware structure diagram of any device with data processing capability in which the distribution network substation voltage control method based on the improved Bayesian deep belief network provided by the embodiment of the present invention is located, except Figure 3 In addition to the processor, memory and network interface shown, any device with data processing capability in which the apparatus in the embodiment is located may also include other hardware, generally based on the actual functions of the device with data processing capability, which will not be described in detail.

[0051] Correspondingly, the present application also provides a computer-readable storage medium on which computer instructions are stored, and when the instructions are executed by the processor, the distribution network substation voltage control method based on the improved Bayesian deep belief network as described above is implemented. The computer-readable storage medium can be an internal storage unit of any device with data processing capabilities described in any of the aforementioned embodiments, such as a hard disk or memory. The computer-readable storage medium can also be an external storage device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), an SD card, a flash card (FlashCard), etc. equipped on the device. Further, the computer-readable storage medium can also include both an internal storage unit of any device with data processing capabilities and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capabilities, and can also be used to temporarily store data that has been output or is to be output.

[0052] Those skilled in the art will readily appreciate other embodiments of the present application after considering the description and practicing the contents disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art that are not disclosed in the present application. The description and examples are intended to be exemplary only.

[0053] It should be understood that the present application is not limited to the exact construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof.

Claims

1. A voltage control method for a distribution network substation, characterized in that: The method comprises: Constructing an IBDBN-DGAM network, the IBDBN-DGAM network comprising an input layer, an IBDBN module, a DGAM module, and a decision layer; wherein the IBDBN module performs deep feature extraction and probability modeling on the input data based on a multi-layer Bayesian network structure, and the DGAM module dynamically adjusts the weight of the network topology structure based on the feature representation output by the IBDBN using a dynamic graph attention mechanism; The historical multimodal data of the distribution substation is used as input data and input into the IBDBN-DGAM network; an objective function is constructed to optimize the IBDBN-DGAM network; wherein the multimodal data of the distribution substation includes voltage, current, and load data from the SCADA system and PMU unit; The real-time multimodal data of the distribution substation is obtained and input into the optimized IBDBN-DGAM network, and the voltage control command is output.

2. A distribution network substation voltage control method according to claim 1, characterized in that: The method further includes: preprocessing the historical multimodal data of the distribution network substation including data cleaning, missing data interpolation, and feature engineering; Set the preprocessing evaluation index, the expression is as follows: Q data =α·Completeness+β·Accuracy+γ·Consistency Where α, β, and γ are weight coefficients.

3. A distribution network substation voltage control method according to claim 1, characterized in that: The expression of the objective function is as follows: Where J(θ,π) represents the objective function value, which indicates the total loss that the IBDBN-DGAM framework needs to minimize; θ represents the model parameters, which are related to the training of IBDBN (Bayesian Deep Belief Network) and are responsible for feature extraction and uncertainty processing; π represents the strategy parameters, which are related to DGAM (Dynamic Graph Attention Mechanism) and are used to dynamically capture the topology changes of the distribution network; represents the expectation calculated under the data distribution p(x,y), indicating that the main optimization direction of the objective function is based on the data distribution; p(y|x; θ) represents the conditional probability distribution given the input x and parameter θ, which is used to predict the probability of output y; p(y) represents the prior probability of the output variable y; β represents the weight factor of the KL divergence, which is used to control the strength of the similarity constraint between the potential distribution p(z|x; θ) and the prior distribution p(z); KL(·) represents the KL divergence, KL(p(z|x; θ)||p(z)) represents the Kullback-Leibler divergence between the potential variable distribution p(z|x; θ) and the prior distribution p(z), which is used to measure the difference between the distribution generated by the model and the true distribution; λ represents the weight factor of the cumulative reward, which determines the trade-off between short-term and long-term benefits; γ represents the discount factor, which represents the impact of the current decision on future benefits; t∈T, T represents the time step; Cumulative reward, r(s t ,a t ) means in state s t Execute action a t The immediate rewards received; η represents the weight factor of control and disturbance cost, which is used to weigh the cost optimization in the worst case; represents the worst-case disturbance control problem, C(u) is the control cost, and D(Δ) is the disturbance cost.

4. A distribution network substation voltage control method according to claim 1, characterized in that: The method also includes: optimizing the IBDBN-DGAM network using optimization strategies such as adaptive learning rate adjustment, batch normalization, regularization, and adversarial training.

5. A distribution network substation voltage control method according to claim 1, characterized in that: Set the optimization evaluation index to evaluate the optimization effect of the IBDBN-DGAM network. The expression is as follows: Where N is the number of evaluation samples, and w1, w2, and w3 are the weights of each performance indicator.

6. A distribution network substation voltage control method according to claim 1, characterized in that: The network deployment evaluation index is set to evaluate the effect of IBDBN-DGAM network deployment in the distribution substation. The expression is as follows: ResponseTime=t decision +t execution Where, t decision represents the decision time, t execution Indicates execution time.

7. An electronic device comprising a memory and a processor, characterized in that: The memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the distribution network substation voltage control method described in any one of claims 1-6 above.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the distribution network substation voltage control method as described in any one of claims 1 to 6 is implemented.

9. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the voltage control method for a distribution network substation described in any one of claims 1-6 is implemented.

Images