A flexible power distribution network intelligent management and control method based on cloud edge-end cooperation

Abstract

The application provides a flexible power distribution network intelligent management and control method based on cloud edge end cooperation, and belongs to the technical field of smart grids. An intelligent decision model is constructed through a cloud end and distributed to an edge server to monitor power grid states and formulate management and control strategies. Strategy information is exchanged between edge servers to work cooperatively. A sending end edge server updates intelligent regulation and control judgment and decision model parameters based on abnormal operation index probability if a strategy is rejected. If the strategy is accepted and the strategy utility exceeds a utility threshold, the edge server updates the model according to the operation conditions of regional power grids. If the strategy is rejected for multiple times or does not meet multiple threshold stability requirements, the cloud end updates the model by using a ternary loss function containing multiple bias factors, judges multiple domain stability requirements and introduces an adaptive learning rate and the ternary loss function. The application improves the real-time performance and accuracy of decision making.

Description

TECHNICAL FIELD

[0001] The application provides a flexible power distribution network intelligent management and control method based on cloud edge end cooperation, and belongs to the technical field of smart grids. BACKGROUND

[0002] With the continuous development of smart grid technology, the intelligent management and control of power distribution networks has become a key to improving grid operation efficiency, ensuring power supply reliability, and optimizing energy utilization. Traditional power distribution network management and control methods often rely on fixed and centralized control centers, making it difficult to achieve flexible and efficient resource scheduling and fault response. Especially in the face of large-scale access of distributed renewable energy, diversification and rapid changes of power load, traditional methods are difficult to quickly adapt to these dynamic changes, leading to decreased grid operation efficiency and threatened power supply reliability. Cloud edge end cooperation technology integrates the powerful data processing capabilities of cloud computing and the real-time response capabilities of edge computing, as well as the flexible access of terminal devices, providing a new solution for the intelligent management and control of power distribution networks.

[0003] The cloud edge end flexible power distribution network intelligent management and control method can intelligently adjust the operation strategy of the power grid according to changes in power load and the access of distributed renewable energy, achieving optimized energy utilization and energy saving and emission reduction. However, the current cloud edge end flexible power distribution network intelligent management and control still has the following problems: On the one hand, traditional methods fail to fully utilize the cooperation between edge servers, resulting in information isolation and the inability to integrate data and strategies. Lack of close cooperation between the cloud and the edge end leads to slow response in grid state monitoring and strategy formulation. The lack of multi-edge cooperation in calculating strategy utility results in inaccurate and timely decision-making, making it difficult to cope with the complexity of the grid environment. On the other hand, traditional methods fail to fully combine the advantages of cloud and edge computing to optimize the model cooperatively, and do not consider multi-domain stability, affecting the long-term safe operation of the power grid. At the same time, traditional methods do not update the model based on multiple factors improving the loss function and learning rate, resulting in inaccurate model updates. Therefore, it is urgent to propose a flexible power distribution network intelligent management and control method based on cloud edge end cooperation to improve grid operation efficiency, ensure power supply reliability, and optimize energy utilization.

[0004] CN118659894A provides a power distribution network security management and control method and system based on device fingerprints, which monitors and warns abnormal and unknown devices by real-time scanning of power distribution network devices, thereby preventing power distribution network intrusion. However, this technology fails to fully utilize the cooperation between edge servers, resulting in information isolation and the inability to integrate data and strategies. Lack of close cooperation between the cloud and the edge end, and the lack of multi-edge cooperation in calculating strategy utility result in inaccurate and timely decision-making, making it difficult to cope with the complexity of the grid environment.

[0005] The application number 202410335908 provides a power distribution network management and control method, system, terminal and storage medium based on Internet of Things, obtains power grid monitoring data from the Internet of Things terminal by establishing a communication connection with the Internet of Things terminal, and investigates the abnormality of the power grid monitoring data. However, this technology fails to fully combine the advantages of cloud and edge computing to optimize the model, does not consider multi-domain stability, and affects the long-term safe operation of the power grid. Moreover, it does not update the model based on a three-element loss function and an adaptive learning rate constructed from multiple factors, resulting in inaccurate model updates. SUMMARY

[0006] The present application provides a flexible power distribution network intelligent management and control method based on cloud-edge-end collaboration, which solves the following two technical problems:

[0007] 1) The traditional method fails to fully utilize the cooperation between edge servers and does not use multi-edge collaborative strategy utility calculation, resulting in inaccurate and timely decision-making and difficulty in coping with the complexity of the power grid environment.

[0008] 2) The traditional method fails to fully combine the advantages of cloud and edge computing with multi-domain stability, and does not update the model based on a three-element loss function constructed from multiple factors, resulting in inaccurate model updates.

[0009] A flexible power distribution network intelligent management and control method based on cloud-edge-end collaboration includes multi-edge collaborative power distribution network intelligent management and control strategy utility calculation and cloud-edge collaborative management and control model update based on multi-domain stability.

[0010] (1) Multi-edge collaborative power distribution network intelligent management and control strategy utility calculation;

[0011] First, the cloud server trains and issues intelligent control and decision-making models to the edge server to support intelligent decision-making and power distribution network operation state monitoring. Second, the sending edge server calculates the power grid operation state and management and control strategy based on the model and interacts with the receiving edge server. Finally, the sending edge server calculates the strategy utility based on multi-edge collaboration.

[0012] (2) Cloud-edge collaborative management and control model update based on multi-domain stability;

[0013] First, when a strategy is rejected by the receiving server, the sending edge server updates the parameters of the intelligent regulation judgment and decision-making model based on the probability of abnormal operation indicators of each receiving edge server. Second, if the strategy is accepted by the receiving server, and the strategy utility calculated by the sending edge server based on multi-edge collaborative computing is greater than the utility threshold, the intelligent regulation judgment and decision-making model is updated based on the power grid operation status of the control area of ​​each receiving server. Finally, if the strategy is rejected multiple times, or the strategy utility does not meet the multi-domain stability requirements, a ternary loss function and adaptive learning rate are constructed on the cloud side to update the intelligent regulation judgment and decision-making model.

[0014] Specifically:

[0015] (1) Calculation of the utility of distribution network intelligent management and control strategy based on multi-edge collaboration, including the following steps:

[0016] Step 1.1: The cloud server trains and distributes the intelligent control judgment and decision-making model;

[0017] The cloud server trains an intelligent control judgment and decision model adapted to the intelligent management and control of flexible distribution networks based on data from the power grid database. The model analyzes the power grid operation status based on real-time data uploaded by the terminal and edge collaboration parameters, and provides corresponding control strategies.

[0018] The cloud server will use the trained intelligent regulation, judgment, and decision-making model M cloud The data is distributed to each edge server, and the set of edge servers is represented as {d1,...,d...}. j ,...,d J}, where dj represents the j-th edge server, and J is the total number of edge servers. Edge server d j First, it updates its intelligent control judgment and decision-making model based on the model parameters issued by the cloud server, i.e., w edge,j =w cloud Then, based on the regional distribution network operation data and models uploaded by terminals within the control area, the distribution network operation status is assessed and corresponding control strategies are generated. (Edge server d) j Intelligent regulation judgment and decision-making model M edge,j Represented as:

[0019] output = M edge,j (w edge,j (input)

[0020] input = {D j ,ζ}

[0021]

[0022] The input parameters consist of two parts, one of which is d.j The regionally distributed operation data uploaded by the terminal in the control area is denoted as D j , and the other part is an edge collaborative input parameter, which is used for edge collaboration between edge servers. The control strategy is input to analyze the effectiveness of the strategy. When there is no edge collaboration, the parameter is empty, denoted as ζ. S1, S2, and S3 represent the normal, good, and abnormal states of the power grid operation, respectively. is d j The probabilities of the three power grid operation states of the control area. S1: is a key-value pair, indicating the current d j The probability of the power grid operation state of the control area being in state S1 is The power grid control strategy.

[0023] The edge server d j According to the model calculation, the related data of the power grid operation state are used to determine whether the current regional power grid needs to be controlled. According to historical experience, the power grid operation safety threshold is set to P th When , the edge server d j determines that the current power grid operation state of the control area is abnormal, and sends the control strategy to the adjacent server as the sending end server to evaluate and optimize the strategy through edge collaboration.

[0024] Step 1.2: Interaction of control strategies between edge servers

[0025] The edge server d j that determines that the current power grid operation state of the control area is abnormal serves as the sending end edge server to send the control strategy to the adjacent edge server d j′ , The control strategy sent by the edge server d is , where d j is the set of adjacent edge servers. d j′ After receiving the control strategy, the receiving end server inputs the control strategy sent by the edge server d j as an edge collaborative input parameter into the model M edge,j′ . The input and output of the model are represented as:

[0026]

[0027] In the formula, the output of the model is the updated power grid operation state based on the edge collaborative control strategy. Let Θ1, Θ2, and Θ3 be the transition thresholds of the three states. If or , it indicates that the power grid operation state has deteriorated after adopting the control strategy, and cannot meet the demand for power grid safety and stability. The receiving end edge server dj′ The rejection control strategy is combined with multiple operating index conditions and corresponding probabilities The updated power grid operating state deviation obtained based on the edge collaborative control strategy is returned to the sending end edge server d j The interaction process is completed. Otherwise, directly combine multiple operating index conditions and corresponding probabilities The updated power grid operating state deviation obtained based on the edge collaborative control strategy is returned to the sending end edge server d j .

[0028] Step 1.3: Utility calculation of distribution network intelligent control strategy based on multi-edge collaboration

[0029] sending end edge server d j Utility ψ of multi-edge collaborative computing strategy j The utility of the strategy considers, on the one hand, the adoption of the control strategy by the receiving end server The power grid is in three operating state changes after adopting the strategy. If the probability of abnormal state increases, it means that the strategy utility for d j′ region is poor. Feedback to the utility of the multi-edge collaborative computing strategy is reflected as follows: when the probability of abnormal state increases or the probability of normal and good is low, the utility will also decrease. On the other hand, it considers the multiple operating index conditions. If the probabilities of voltage limit, line overcurrent and reverse overload exceed the threshold, it means that the strategy has poor collaborative effect among edge servers, and the strategy utility should be reduced, which is specifically expressed as:

[0030]

[0031] In the formula, is the number of receiving end edge servers, Ω1, Ω2, Ω3 are the index factors of the three power grid operating states, respectively, to distinguish the influence difference of different state probability deviations. T v , T r , T b are the probability thresholds of voltage limit, line overcurrent and reverse overload, respectively, ω v , ω r , ω b are the weights of voltage limit, line overcurrent and reverse overload, respectively. is an indicator function, which is 1 when , otherwise it is 0. The formula means that when the probabilities of voltage limit, line overcurrent and reverse overload exceed the threshold, a penalty term is applied to the strategy utility, and the more the threshold is exceeded, the greater the penalty term.

[0032] (2) Cloud-edge collaborative control model update based on multi-domain stability, including the following steps:

[0033] Step 2.1: Intelligent regulation and decision-making model update on the edge server side;

[0034] If the policy is rejected by the receiving end server, the sending edge server updates the power grid operation state and the corresponding probability according to the edge collaborative management policy, and updates the intelligent regulation and decision-making model parameters, denoted as:

[0035]

[0036] In the formula, is the gradient function, and η is the learning rate.

[0037] If the policy is accepted by the receiving end server, and d j The policy utility ψ j based on multi-edge collaborative computing is greater than the utility threshold ψ th , the policy is executed, and the multi-edge collaborative situation is recorded.

[0038] Step 2.2: Multi-domain stability judgment and cloud-side intelligent regulation and decision-making model update;

[0039] If the policy is rejected multiple times, or the policy utility ψ j does not meet the multi-domain stability requirements, the edge server d j uploads the power grid operation and decision to the cloud. Multi-domain stability includes high-frequency fluctuation domain, large-scale fluctuation domain, and confidence domain: high-frequency fluctuation domain can reflect whether the policy utility is repeatedly fluctuating around the threshold; large-scale fluctuation domain can reflect whether the policy utility frequently fluctuates in a large range; and confidence domain reflects the credibility of the management policy effectiveness, which is quantified by the state change of the power grid after the policy is executed on each receiving end edge server region. If any one of the conditions is not met, multi-domain stability cannot be achieved.

[0040] Considering that T interaction iterations are performed simultaneously, T interaction policy utilities are obtained, and the set is denoted as ψ j ={ψ j,1 ,...,ψ j,t ,...,ψ j,T}. The condition that does not meet the multi-domain stability requirements is:

[0041]

[0042] In the formula, indicates that the mean of the policy utility deviates from the utility threshold by no more than a constant δ, ​​This indicates that the frequency of policies with utilities less than a certain threshold exceeds Num. th,1 Both conditions simultaneously satisfy the requirement that the strategy utility fluctuates frequently around the threshold, but do not meet the requirement of multi-domain stability. This indicates that the deviation of the policy utility from the threshold in the policy utility set exceeds the maximum range. The frequency is greater than Num th,2 This indicates that the strategy utility frequently fluctuates across a wide range, failing to meet the multi-domain stability requirement. This represents the relative change in the power grid's state after the policy is implemented in each receiving edge server area. If the increase in the probability of the power grid being in a good or normal state is small, and the decrease in the probability of it being in an abnormal state is small, it indicates poor effectiveness. Therefore, when... This indicates poor strategy effectiveness, low confidence, and failure to meet multi-domain stability requirements.

[0043] A ternary loss function is constructed on the cloud side, based on policy utility ψ. j With utility threshold ψ th The deviation, the operational status of each receiving edge server based on various indicators, and the corresponding probability and the control policies adopted by the receiving server Changes in the three operating states of the power grid The intelligent regulation judgment and decision-making model is updated to improve its accuracy, specifically as follows:

[0044]

[0045] In the formula, d(ψ) j ,ψ th ) represents the strategy utility ψ j With utility threshold ψ th The deviation, η cloud The learning rate is adaptive to the cloud side. The worse the effect of the edge interaction strategy, the higher the learning rate, thereby increasing the convergence speed of the model.

[0046] This invention proposes an intelligent management and control method for flexible power distribution networks based on cloud-edge-device collaboration, which has the following technical advantages:

[0047] (1) Calculation of the utility of intelligent control strategies for power distribution networks based on multi-edge collaboration: Intelligent decision-making models are built and distributed from the cloud to edge servers to monitor the power grid status and formulate control strategies. Strategy information is exchanged among the edge servers, and they work collaboratively to achieve intelligent control, improving decision-making speed and accuracy. Finally, the sending edge server calculates the strategy utility based on multi-edge collaboration, further improving the real-time performance and accuracy of decision-making.

[0048] (2) Cloud-edge collaborative management model update based on multi-domain stability; if the strategy is rejected, the sending edge server updates the intelligent regulation and decision model parameters based on the abnormal operation index probability of each receiving edge server. If the strategy is accepted and the strategy utility exceeds the utility threshold, the edge server will further update the model according to the operation of each regional power grid. If the strategy is rejected multiple times or does not meet the multi-threshold stability requirement, the cloud will use a ternary loss function containing multiple bias factors to update the model, which provides a comprehensive and accurate optimization direction for the model by judging the multi-domain stability requirement and introducing an adaptive learning rate and a ternary loss function, to improve the efficiency and accuracy of the update. This method optimizes the model update process by taking advantage of cloud and edge computing. BRIEF DESCRIPTION OF DRAWINGS

[0049] Figure 1 is a flowchart of the present application. DETAILED DESCRIPTION

[0050] The present application proposes a flexible power distribution network intelligent management and control method based on cloud-edge-end collaboration, as shown in Figure 1 The present application proposes a flexible power distribution network intelligent management and control method based on cloud-edge-end collaboration, as shown in

[0051] (1) Strategy utility calculation of intelligent management and control of power distribution network based on multi-edge collaboration

[0052] First, the cloud server trains and issues an intelligent regulation and decision model to the edge server to support intelligent decision-making and power distribution network operation state monitoring. Second, the sending edge server calculates the power grid operation state and management strategy based on the model, and interacts with the receiving edge server. Finally, the sending edge server calculates the strategy utility based on multi-edge collaboration.

[0053] Step 1.1: Cloud server trains and issues intelligent regulation and decision model

[0054] The cloud server is a virtualized server based on cloud computing technology, which provides sufficient computing resources, storage space and application services through the Internet. The cloud server trains an intelligent regulation and decision model suitable for intelligent management and control of flexible power distribution network based on data in the power grid database. The model can analyze the power grid operation state according to the real-time data uploaded by the terminal and the edge collaboration parameters, and give the corresponding management strategy.

[0055] The cloud server issues the trained intelligent regulation and decision model M cloud to each edge server, and the edge server set is represented as {d1,...,d j ,...,d J, where dj represents the jth edge server, and J is the total number of edge servers. The edge server d j First, update the intelligent control judgment and decision model w edge,j cloud according to the model parameters issued by the cloud server, that is, j Then, based on the regional distribution network operation data uploaded by the terminal in the management and control area and the model, the operation state of the distribution network is evaluated and the corresponding management and control strategy is generated. The intelligent control judgment and decision model M edge,j of the edge server d edge,j is represented as:

[0056] output=M edge,j (w j ,input)

[0057] input={D j ,ζ}

[0058]

[0059] Among them, the input parameters include two parts, one part is the regional distribution network operation data uploaded by the terminal in the management and control area, including voltage, current, active power, reactive power, power quality, etc., represented as D j , and the other part is the edge coordination input parameter, which is used in edge coordination between edge servers. The control strategy can be used to analyze the effectiveness of the strategy, and when the edge coordination is not performed, the parameter is empty, represented as ζ. S1, S2, S3 respectively represent that the power grid operation state is normal, good and abnormal. is the probability of the three power grid operation states of d j control area. S1: is a key-value pair, indicating that the probability of the current d j control area power grid operation state in state S1 is is the power grid management and control strategy.

[0060] The edge server d j judges whether the current regional power grid needs to be controlled according to the power grid operation state related data calculated by the model. According to historical experience, the power grid operation safety threshold is set to P th , when , the edge server d j determines that the current management and control area power grid operation state is abnormal, and sends the management and control strategy to the adjacent server as the sending end server, and evaluates and optimizes the strategy through edge coordination.

[0061] Step 1.2: Edge server management and control strategy interaction

[0062] ​Edge server d that determines the abnormal operation status of the power grid in the current controlled area j As the sending edge server, it sends data to the adjacent edge server d. j′ , Sending control policies in, For d j A collection of adjacent edge servers. j′ After receiving the control policy, the receiving server will... j The control strategy sent is input as an edge collaboration input parameter into model M. edge,j′ In the model, the input and output are represented as:

[0063]

[0064] In the formula, the output of the model This represents the updated power grid operating state obtained based on the edge collaborative management and control strategy. Let Θ1, Θ2, and Θ3 be the transition thresholds for the three states. or This indicates that the power grid's operating status deteriorated after the control strategy was adopted, failing to meet the requirements for power grid safety and stability. (Receiver edge server d) j′ The control strategy will be rejected, and various operating indicators such as voltage over-limit and reverse overload will be analyzed along with their corresponding probabilities. Updated power grid operation status deviation obtained based on edge collaborative control strategy Send back to the sending edge server d j Complete the interaction process. Otherwise, directly display various operating indicators such as voltage over-limit and reverse overload, along with their corresponding probabilities. Updated power grid operation status deviation obtained based on edge collaborative control strategy Send back to the sending edge server d j .

[0065] Step 1.3: Utility Calculation of Smart Distribution Network Management and Control Strategy Based on Multi-Edge Collaboration

[0066] Sending edge server d j The utility ψ of multi-edge collaborative computing strategy j The effectiveness of the strategy takes into account, on the one hand, the control strategy adopted by the receiving server. The power grid is in three different operating states. If the probability of abnormal states increases after adopting a certain strategy, it indicates that for d... j′The utility of the strategy in the region is poor. The feedback to the utility of the multi-edge collaborative computing strategy is that when the probability of abnormal state increases or the probabilities of normal and good are low, the utility will also decrease. On the other hand, considering various operation indicators such as voltage out-of-limit, reverse overload indicators, if the probabilities of voltage out-of-limit, line overcurrent and reverse overload exceed the threshold, it means that the strategy is poor in collaborative effect among edge servers, and the utility of the strategy should be appropriately reduced, which is specifically expressed as:

[0067]

[0068] wherein, is the number of receiving end edge servers, Ω1, Ω2, Ω3 are the exponential factors of the three power grid operation states, used to distinguish the influence difference of different state probability deviations. v , T r , T b are the probability thresholds of voltage out-of-limit, line overcurrent and reverse overload, ω v , ω r , ω b are the weights of voltage out-of-limit, line overcurrent and reverse overload. is an indicator function, which is 1 when , otherwise 0. The formula means that when the probabilities of voltage out-of-limit, line overcurrent and reverse overload exceed the threshold, a penalty term is applied to the utility of the strategy, and the more the threshold is exceeded, the greater the penalty term.

[0069] (2) Cloud-edge collaborative management model update based on multi-domain stability

[0070] Firstly, when the strategy is rejected by the receiving end server, the sending end edge server updates the intelligent control judgment and decision model parameters based on the abnormal operation indicator probability of each receiving edge server. Secondly, if the strategy is received by the receiving end server, and the utility of the strategy based on multi-edge collaborative computing of the sending end edge server is greater than the utility threshold, the intelligent control judgment and decision model is updated based on the operation of the regional power grid of each receiving server. Finally, if the strategy is rejected for multiple times, or the utility of the strategy does not meet the requirements of multi-domain stability, a ternary loss function and an adaptive learning rate are constructed on the cloud side to update the intelligent control judgment and decision model.

[0071] Step 2.1: Intelligent control judgment and decision model update on edge server side

[0072] If the strategy is rejected by the receiving end server, the sending end edge server updates the intelligent control judgment and decision model parameters according to the updated power grid operation state and the corresponding probability obtained by the edge collaborative management strategy, which is expressed as:

[0073]

[0074] wherein, is a gradient function, and η is a learning rate.

[0075] If the policy is received by the receiving end server, and d j The policy utility ψ j of the multi-edge collaborative computing is greater than the utility threshold ψ th , the policy can be executed, and the multi-edge collaborative situation is recorded.

[0076] Step 2.2: Multi-domain stability judgment and cloud-side intelligent regulation judgment and decision model updating

[0077] If the policy is rejected multiple times, or the policy utility ψ j does not meet the multi-domain stability requirement, the edge server d j Uploads the power grid operation and decision to the cloud. The multi-domain stability includes the high-frequency fluctuation domain, the large-range fluctuation domain, and the confidence domain: the high-frequency fluctuation domain can reflect whether the policy utility repeatedly fluctuates around the threshold value; the large-range fluctuation domain can reflect whether the policy utility frequently fluctuates in a large range; and the confidence domain reflects the credibility of the control policy effectiveness, which can be quantified by the state change of the power grid after the policy is executed by each receiving end edge server region. If any one of the domains does not meet the condition, the multi-domain stability cannot be achieved.

[0078] Consider that T interaction iterations are performed together, and T interaction policy utilities are obtained, and the set is denoted as ψ j ={ψ j,1 ,...,ψ j,t ,...,ψ j,T}. The condition that does not meet the multi-domain stability requirement is:

[0079]

[0080] wherein, indicates that the mean value of the policy utility deviates from the utility threshold value by no more than a constant δ, represents that the frequency of the policy utility in the utility set that is less than the utility threshold value exceeds Num th,1 , and both of them can reflect that the policy utility repeatedly fluctuates around the threshold value, which does not meet the multi-domain stability requirement. represents that the frequency of the policy utility in the utility set that deviates from the threshold value by more than the maximum range exceeds Num th,2 , which indicates that the policy utility frequently fluctuates in a large range, which does not meet the multi-domain stability requirement. This represents the relative change in the power grid's state after the policy is implemented in each receiving edge server area. If the increase in the probability of the power grid being in a good or normal state is small, and the decrease in the probability of being in an abnormal state is small, it indicates poor effectiveness. Therefore, when... This indicates poor strategy effectiveness, low confidence, and failure to meet multi-domain stability requirements.

[0081] A ternary loss function is constructed on the cloud side, based on policy utility ψ. j With utility threshold ψ th The deviation, the operating conditions of each receiving edge server based on voltage over-limit, reverse overload indicators, and other factors, and the corresponding probabilities. and the control policies adopted by the receiving server Changes in the three operating states of the power grid The intelligent regulation judgment and decision-making model is updated to improve its accuracy, specifically as follows:

[0082]

[0083] In the formula, d(ψ) j ,ψ th ) represents the strategy utility ψ j With utility threshold ψ th The deviation, η cloud The adaptive learning rate on the cloud side increases as the edge interaction strategy performs worse, thus accelerating model convergence. The adaptive learning rate and ternary loss function more accurately and comprehensively quantify the deviation between the current model and the optimal model, providing a more precise direction for model learning and calibration.

Claims

1. A cloud edge-end collaborative based flexible power distribution network intelligent management and control method, characterized in that, The utility model relates to a multi-edge coordination based distribution network intelligent management and control strategy utility calculation and a cloud-edge coordination management and control model update based on multi-domain stability. (1) Multi-edge coordination based distribution network intelligent management and control strategy utility calculation; First, the cloud server trains and issues intelligent control judgment and decision-making models to the edge server to support intelligent decision-making and distribution network operation state monitoring; Second, the sending edge server calculates the power grid operation state and management and control strategy based on the model and interacts with the receiving edge server; and finally, the sending edge server calculates the strategy utility based on multi-edge coordination. The utility model relates to the following steps: Step 1.1: The cloud server trains and issues intelligent control judgment and decision-making models. The cloud server trains intelligent control judgment and decision-making models suitable for flexible distribution network intelligent management and control based on data in the power grid database. The models analyze the power grid operation state based on real-time data uploaded by terminals and edge coordination parameters and give corresponding management and control strategies. The cloud server will use the trained intelligent control, judgment, and decision-making model The data is distributed to each edge server, and the set of edge servers is represented as follows: ,in, Indicates the first One edge server, Total number of edge servers; edge servers First, it updates its intelligent control judgment and decision-making model based on the model parameters issued by the cloud server. Then, based on the regional distribution network operation data and models uploaded by terminals within the control area, the operation status of the distribution network is assessed and corresponding control strategies are generated; edge server Intelligent regulation judgment and decision-making model Represented as: ， ， ， Wherein, the input parameters include two parts, one part is The regional power grid operation data uploaded by the terminal in the control area, denoted as , the other part is the edge coordination input parameter, which is used for edge coordination between edge servers. The input control strategy is used to analyze the effectiveness of the strategy. When there is no edge coordination, the parameter is empty, denoted as ; 、 、 respectively represent that the power grid operation state is normal, good and abnormal; 、 、 is The probability corresponding to the three power grid operation states of the control area; is a key-value pair, indicating the current The probability that the power grid operation state of the control area is in state is ; is the power grid control strategy; Edge server According to the model calculation, the power grid operation state related data is used to determine whether the current regional power grid needs to be controlled; according to historical experience, the power grid operation safety threshold is set to When , the edge server determines that the current control area power grid operation state is abnormal, and sends the control strategy to the adjacent server as a sending end server to evaluate and optimize the strategy through edge coordination. Step 1.2: Management and control strategy interaction between edge servers Edge server for determining abnormal operation state of current management area power grid as a sending end edge server to adjacent edge servers sending management and control strategy wherein, for a set of adjacent edge servers; as a receiving end server after receiving the management and control strategy, the sent management and control strategy is input into the model as an edge coordination input parameter the model input and output are represented as: ， ， In the formula, the output of the model , , is the updated power grid operation state obtained based on the edge collaborative control strategy; set , , is the transition threshold value of the three states, if , or indicates that the power grid operation state becomes worse after the control strategy is adopted and cannot meet the demand of power grid safety and stability, the receiving end edge server will reject the control strategy and send back to the sending end edge server , the multiple operation index conditions and corresponding probabilities , , the deviation of the updated power grid operation state obtained based on the edge collaborative control strategy ; complete the interaction process; otherwise, directly send back to the sending end edge server , the multiple operation index conditions and corresponding probabilities , , the deviation of the updated power grid operation state obtained based on the edge collaborative control strategy ; Step 1.3: Multi-edge coordination based distribution network intelligent management and control strategy utility calculation Sending edge server Utility of Multi-Edge Collaborative Computing Strategy The effectiveness of the strategy takes into account, on the one hand, the control strategy adopted by the receiving server. The power grid is in three different operating states. If the probability of abnormal states increases after adopting a certain strategy, it indicates that... The strategy is ineffective in this region. This is reflected in the effectiveness of the multi-edge collaborative computing strategy as follows: when the probability of abnormal states increases or the probability of normal and good conditions decreases, the effectiveness also decreases. Furthermore, considering various operational indicators, if the probability of voltage exceeding limits, line overcurrent, and reverse overload exceeds the threshold, it indicates that the strategy has poor collaborative effect among edge servers, and the strategy effectiveness should be reduced. Specifically: ， wherein, is the number of receiving edge servers, , , are the index factors of three grid operating states, respectively, to distinguish the influence difference of different state probability deviation; , , are the probability thresholds of voltage excursion, line overcurrent and reverse overload, respectively, , , are the weights of voltage excursion, line overcurrent and reverse overload, respectively; is an indicator function, which is 1 when , otherwise 0; this equation means that when the probabilities of voltage excursion, line overcurrent and reverse overload exceed the thresholds, a penalty term is imposed on the strategy utility, and the more the threshold is exceeded, the greater the penalty term. (2) Cloud-edge coordination management and control model update based on multi-domain stability. First, when the strategy is rejected by the receiving server, the sending edge server updates the intelligent control judgment and decision-making model parameters based on the abnormal operation index probability of each receiving edge server; second, if the strategy is received by the receiving server and the strategy utility calculated by the sending edge server based on multi-edge coordination is greater than the utility threshold, the intelligent control judgment and decision-making model is updated based on the power grid operation state of each receiving server management and control area; finally, if the strategy is rejected multiple times or the strategy utility does not meet the multi-domain stability requirements, a ternary loss function and an adaptive learning rate are constructed on the cloud side to update the intelligent control judgment and decision-making model. 2.The cloud edge-end collaborative based flexible power distribution network intelligent management and control method according to claim 1, characterized in that, (2) Cloud-edge coordination management and control model update based on multi-domain stability, including the following steps: Step 2.1: Intelligent control judgment and decision-making model update on the edge server side If the policy is rejected by the receiving end server, the sending end edge server obtains an updated power grid operation state according to an edge coordination management policy and the corresponding probability Update the intelligent control judgment and decision model parameters, denoted as: ， In the formula, is a gradient function, is a learning rate; If strategy Received by the receiving server, and Strategy utility based on multi-edge collaborative computing Greater than the utility threshold Then execute the strategy. And record the details of this multi-edge collaboration; Step 2.2: Multi-domain stability judgment and intelligent control judgment and decision-making model update on the cloud side If the policy is rejected multiple times, or the policy utility is not satisfied, then the edge server will upload the grid operation and decision to the cloud; the multi-domain stability includes a high-frequency fluctuation domain, a large-scale fluctuation domain, and a confidence domain: the high-frequency fluctuation domain is capable of reflecting whether the policy utility repeatedly fluctuates at a high frequency near a threshold value; the large-scale fluctuation domain is capable of reflecting whether the policy utility frequently appears large-scale fluctuations; the confidence domain reflects the credibility of the effectiveness of the control policy, which is quantified by the state change of the grid after the policy is executed by the edge server in each receiving end area; the multi-domain stability cannot be achieved if the conditions of any one domain are not satisfied; Consider co-conduct The next interaction iteration, resulting in The individual interaction policy utility, collectively represented as The case where the multi-domain stability requirement is not met is represented as: ， In the formula, Indicates the mean of strategy utility The deviation from the utility threshold does not exceed a constant. , This indicates that the frequency of strategies with utilities less than a certain threshold exceeds a certain threshold. Both conditions simultaneously satisfy the requirement that the strategy utility fluctuates frequently and repeatedly around the threshold, but do not satisfy the multi-domain stability requirement. This indicates that the deviation of the policy utility in the policy utility set from the threshold exceeds the maximum range. The frequency is greater than This indicates that the strategy utility frequently fluctuates across a wide range, failing to meet the multi-domain stability requirement; , , This represents the relative change in the power grid's state after the policy is implemented in each receiving edge server area. If the increase in the probability of the power grid being in a good or normal state is small, and the decrease in the probability of it being in an abnormal state is small, it indicates poor effectiveness. Therefore, when... This indicates poor strategy effectiveness, low confidence, and failure to meet multi-domain stability requirements; The cloud side constructs a ternary loss function based on the policy utility and the deviation of the utility threshold , each receiving edge server based on multiple operation index conditions and corresponding probabilities , and the receiving end server adopts a control strategy The power grid is in three operating states The intelligent control judgment and decision model is updated to improve the model accuracy, which is specifically represented as: ， ， In the formula, The policy utility is represented The deviation from the utility threshold The deviation from the utility threshold The cloud-side adaptive learning rate, the worse the effect of the edge interaction strategy, the higher the learning rate, thereby increasing the speed of model convergence.

Images