Self-learning power grid current limiting supervision system based on cloud platform

Abstract

The invention relates to the technical field of intelligent power grids, in particular to a self-learning power grid current limiting supervision system based on a cloud platform, and aims to solve the problems that the load change trend cannot be accurately predicted, preventive maintenance cannot be supported and the stability of a power system is reduced in the prior art. Historical operation parameters are collected through the load change prediction module, and a machine learning model is utilized to realize high-precision prediction of load change in the next day, real-time monitoring and early warning of potential overload risks, so that the prediction accuracy is improved, preventive maintenance is supported, the reliability and stability of a power system are enhanced, manual intervention is reduced, and the working efficiency is improved. The management efficiency is improved, and the requirements of different power distribution networks can be flexibly met.

Description

Technical Field

[0001] The present invention relates to the technical field of smart grids, and more specifically, to a self-learning grid current limiting supervision system based on a cloud platform. Background Art

[0002] With the rapid development of the power system and the continuous growth of electricity demand, the safe operation of the power grid has become an important issue in the power industry. Traditional power grid supervision methods have problems such as low monitoring efficiency and delayed response, which can hardly meet the complex and changing needs of the current power grid.

[0003] The patent application with reference publication number CN114363079A discloses a distributed intelligent data supervision system for a cloud platform, and the supervision system includes: a physical layer, an operating system layer, a security detection and reinforcement layer, a security control layer, a cloud native layer and a tenant-side cloud management platform layer; the physical layer is a distributed physical resource; the operating system layer is connected to the physical layer for managing physical computer resources; the security detection and reinforcement layer is used to perform security scans on asset management; the security control layer is connected to the security detection and reinforcement layer; the cloud native layer is connected to the security control layer; the tenant-side cloud management platform layer is used to provide an independent resource access platform for multiple tenants; based on the endogenous layered resource supervision mode, the monitoring data of each layer is collected, and the abnormal characteristics of the data are analyzed by self-learning to enrich the corresponding supervision knowledge base; However, the above-mentioned reference patent uses a multi-layer architecture to securely manage and monitor cloud platform resources, provides all-round protection from physical resources to tenants, and uses a self-learning mechanism to analyze anomalies to continuously optimize safety measures. However, it cannot quickly warn of potential short-circuit risks, which reduces operation and maintenance efficiency, cannot accurately predict load change trends, cannot support preventive maintenance, and reduces the stability of the power system. At the same time, it cannot formulate corresponding current limiting measures for short-circuit faults and overload conditions, and cannot respond to short-circuit or overload conditions in real time, which reduces operation and maintenance efficiency and overall reliability.

[0004] To this end, we propose a self-learning power grid current limiting supervision system based on cloud platform to address the above problems. Summary of the invention

[0005] The purpose of the present invention is to provide a self-learning power grid current limiting supervision system based on a cloud platform, which solves the problems that the prior art cannot quickly warn of potential short-circuit risks, reduces operation and maintenance efficiency, cannot accurately predict load change trends, cannot support preventive maintenance, and reduces the stability of the power system. At the same time, it cannot formulate corresponding current limiting measures for short-circuit faults and overload conditions, cannot respond to short-circuit or overload conditions in real time, and reduces operation and maintenance efficiency and overall reliability.

[0006] The purpose of the present invention is achieved through the following technical solutions: A self-learning power grid current limiting supervision system based on a cloud platform, comprising a cloud management platform, a data acquisition module, a short circuit risk monitoring module, a load change prediction module and a current limiting strategy generation module; The data acquisition module is used to collect the operating parameters of the distribution network in real time and perform preprocessing operations on the collected operating parameters; The short-circuit risk monitoring module is used to collect the short-circuit evaluation parameters of the distribution network in real time and monitor and evaluate the short-circuit faults of the distribution network; The load change prediction module is used to collect historical operating parameters of the distribution network, build a load change prediction model, predict the load change trend in the next day through the model, and identify potential overload risks based on the prediction results; The current limiting strategy generation module is used to generate the optimal current limiting strategy according to the short-circuit fault evaluation results and load change prediction results of the distribution network.

[0007] As a preferred embodiment of the present invention, the specific process of the short-circuit risk monitoring module monitoring and evaluating the short-circuit fault of the distribution network is as follows: Obtain short-circuit evaluation parameters of the distribution network, which include operating current, operating voltage, insulation resistance, and temperature of key electrical components, generate a monitoring cycle ZQ, and divide the monitoring cycle ZQ into multiple monitoring periods {zq1, zq2, …, zqn}, that is, ZQ={zq1, zq2, …, zqn}; Obtain the operating current imbalance value of the distribution network within the monitoring period. The operating current imbalance value represents the ratio between the portion of the operating current variation difference greater than the preset operating current variation difference threshold value and the operating current variation difference within each monitoring period. The operating current variation difference represents the difference between the maximum value and the minimum value of the operating current. The operating current imbalance value is marked as DLS. The operating voltage imbalance value of the distribution network within the monitoring period is obtained. The operating voltage imbalance value indicates the ratio between the part of the operating voltage variation difference greater than the preset operating voltage variation difference threshold value and the operating voltage variation difference within each monitoring period. The operating voltage variation difference indicates the difference between the maximum and minimum values ​​of the operating voltage. The operating voltage imbalance value is marked as DYS.

[0008] As a preferred embodiment of the present invention, the insulation resistance imbalance value of the distribution network within the monitoring period is obtained, the insulation resistance imbalance value represents the ratio between the portion of the insulation resistance variation difference greater than the preset insulation resistance variation difference threshold value and the insulation resistance variation difference within each monitoring period, the insulation resistance variation difference represents the difference between the maximum value and the minimum value of the insulation resistance, and the insulation resistance imbalance value is marked as JZS; The temperature imbalance value of the key electrical components of the distribution network within the monitoring period is obtained. The temperature imbalance value of the key electrical components indicates the ratio between the portion of the temperature variation difference of the key electrical components in each monitoring period that is greater than the preset temperature variation difference threshold of the key electrical components and the temperature variation difference of the key electrical components. The temperature variation difference of the key electrical components indicates the difference between the maximum and minimum values ​​of the temperature of the key electrical components. The temperature imbalance value of the key electrical components is marked as ZWS.

[0009] As a preferred embodiment of the present invention, the operating current imbalance value DLS, the operating voltage imbalance value DYS, the insulation resistance imbalance value JZS and the key electrical component temperature imbalance value ZWS are obtained, and the short-circuit fault assessment coefficient DGP is calculated by the following formula: ; Where c1, c2, c3 and c4 are all preset proportional factor coefficients, c4>c3>c2>c1>0, and the short-circuit fault assessment coefficient DGP is compared with the preset short-circuit fault assessment coefficient threshold: If the short-circuit fault assessment coefficient DGP is less than the preset short-circuit fault assessment coefficient threshold, it indicates that the distribution network is in normal operation; If the short-circuit fault assessment coefficient DGP is greater than or equal to the preset short-circuit fault assessment coefficient threshold, it indicates that the distribution network is in a short-circuit fault state.

[0010] As a preferred implementation of the present invention, the specific process of the load change prediction module constructing a load change prediction model and predicting the load change trend within the next day through the model is as follows: Obtain the historical operating parameters of the distribution network, including operating current, operating voltage, operating power, operating load, power factor and ambient temperature, generate a collection cycle, the collection cycle is one year long, divide the collection cycle into k collection periods, divide each collection period into m continuous sub-periods, and mark the midpoint of each sub-period to obtain m midpoints; Taking m midpoint moments as the base point, expand forward and backward by equal time lengths, mark s-1 expansion moments, and then summarize the midpoint moments and s-1 expansion moments to obtain s detection moments; The operating current values ​​at s detection moments are measured by a measuring instrument to obtain s detection operating current values, and the s detection operating current values ​​are accumulated and averaged to obtain m sub-operating current values.

[0011] As a preferred embodiment of the present invention, the expression of the sub-operation current value is: ; In the formula, ZDL zm is the sub-operation current value of the mth sub-cycle, ZDLjcmn is the nth detection operation current value of the mth sub-cycle; Remove the maximum and minimum sub-operating current values, and add up the remaining m-2 sub-operating current values ​​and calculate the average to obtain the operating current mean value; The expression of the mean operating current is: ; In the formula, DL yj is the average operating current, ZDL zp is the sub-operation current value of the p-th sub-cycle.

[0012] As a preferred embodiment of the present invention, the method of calculating the mean value of the operating current can be used to obtain the mean value of the operating voltage DY yj , operating power average GL yj , operating load average FZ yj , power factor mean GY yj And the average ambient temperature HW yj , the average operating current DL yj , average operating voltage DY yj , operating power average GL yj , operating load average FZ yj , power factor mean GY yj And the average ambient temperature HW yj A load change prediction matrix FBY is constructed in combination, and the prediction matrix FBY is used as the input of the machine learning model. The load change matrix in the next day corresponding to each group of prediction matrices FBY is used as the output of the machine learning model. The load change matrix in the next day is used as the prediction target, and the minimum sum of prediction errors of all training data is used as the training target. The machine learning model is trained until the sum of prediction errors converges and the training is stopped to obtain the load change prediction model.

[0013] As a preferred implementation of the present invention, the load change prediction model is expressed as follows: ; Where FZJ represents the load change matrix in the next day, β1, β2, β3, β4, β5 and β6 are regression coefficients, δ is the random error term, and ∆FBn represents the load change in the nth period in the next day; The real-time operating parameters of the distribution network are obtained, converted into the corresponding prediction matrix FBY and input into the load change prediction model. The real-time load change matrix FZJ in the next day is obtained through the load change prediction model, and the load change in the real-time load change matrix FZJ is compared with the preset load change threshold: If ∆FBn is less than the preset load change threshold, it indicates that the distribution network is in normal operation; If ∆FBn is greater than or equal to the preset load change threshold, it indicates that the distribution network is in an overloaded operating state.

[0014] As a preferred implementation of the present invention, the specific process of the current limiting strategy generation module generating the optimal current limiting strategy is as follows: Obtain the short-circuit fault assessment results and load change prediction results of the distribution network. The short-circuit fault assessment results indicate whether the current distribution network is in a short-circuit fault state. The load change prediction results indicate whether the distribution network is in an overloaded operation state at different time periods in the next day. Generate the optimal current limiting strategy based on the short-circuit fault assessment results and load change prediction results of the distribution network. The generated flow limiting strategy is sent to the cloud management platform. After receiving the flow limiting strategy, the cloud management platform immediately sends the corresponding control instructions to the execution agency for execution.

[0015] As a preferred implementation of the present invention, the specific content of the current limiting strategy is: Current limiting strategy under short circuit fault condition: Current limiting threshold setting: According to the short-circuit current size I d and rated current I e , set the current limiting threshold I x , calculate the current limiting threshold I by the following formula x : I x =f·I e , where f is the safety factor, f=0.8; Current limiting period division: The current limiting period is the period during which the short circuit fault occurs until the fault is eliminated; Current limiting action: Use a circuit breaker to quickly disconnect the faulty line, switch to the backup power supply through an automatic switching device, and connect a current limiting reactor in series in the faulty line to limit the short-circuit current; Current limiting strategy in overload operation state: Current limiting threshold setting: according to the predicted load P y And rated capacity P e , set the current limiting threshold P x , calculate the current limiting threshold P by the following formula x : P x =P e -∆P, where ∆P is the safety margin; Current limiting period division: according to the prediction results, the current limiting period is divided. If the predicted overload period is from t1 to t2, the current limiting period is set to t1-∆t to t2+∆t, where ∆t is the buffer time. Current limiting action: transfer part of the load to other lines through load transfer devices, enable energy storage systems to provide additional power during overload periods, and notify users to reduce electricity load through demand response systems.

[0016] Compared with the prior art, the advantages of the present invention are: (1) In the present invention, the key parameters of the distribution network are collected in real time through the short-circuit risk monitoring module, the imbalance value is calculated and the short-circuit fault assessment coefficient is obtained. It has high accuracy and real-time response capabilities, can quickly warn of potential short-circuit risks, and support preventive maintenance. Its quantitative analysis enhances system reliability, flexibly adapts to different scenarios, reduces manual intervention, and improves operation and maintenance efficiency; (2) In the present invention, the load change prediction module collects historical operating parameters and uses machine learning models to achieve high-precision prediction of load changes in the next day, real-time monitoring and early warning of potential overload risks, improves prediction accuracy, supports preventive maintenance, enhances the reliability and stability of the power system, reduces manual intervention, improves management efficiency, and can flexibly adapt to the needs of different distribution networks; (3) In the present invention, specific current limiting measures are formulated for short-circuit faults and overload conditions through the current limiting strategy generation module, thereby ensuring the high safety of the system, being able to respond to short-circuit or overload conditions in real time, and reacting quickly to minimize the impact on the power grid. By reasonably allocating loads and enabling backup power supplies, resource utilization is optimized and system efficiency is improved. The current limiting threshold and buffer time are flexibly adjusted according to actual conditions to adapt to different power grid environments, reducing the need for manual intervention, and improving operation and maintenance efficiency and overall reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a system block diagram of Embodiment 1 of the present invention; Figure 2 This is a system block diagram of Embodiment 2 of the present invention; Figure 3 This is a logical flow diagram of the first embodiment of the present invention. DETAILED DESCRIPTION

[0018] The following will combine the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all the embodiments. All other embodiments obtained by ordinary technicians in this field without creative work based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0019] Embodiment 1: Figure 1 and Figure 3As shown, the present invention proposes a self-learning power grid current limiting supervision system based on a cloud platform, including a cloud management platform, a data acquisition module, a short circuit risk monitoring module, a load change prediction module and a current limiting strategy generation module; The cloud management platform is connected to the data acquisition module, the short circuit risk monitoring module, the load change prediction module and the current limiting strategy generation module in a one-way communication manner. The current limiting strategy generation module is connected to the short circuit risk monitoring module and the load change prediction module in a one-way communication manner. A data acquisition module is used to collect the operating parameters of the distribution network in real time and perform preprocessing operations on the collected operating parameters. The preprocessing operations include but are not limited to data cleaning, filtering and normalization. The data acquisition module improves the reliability and efficiency of the power system by acquiring and preprocessing the operating parameters of the distribution network in real time, including data cleaning, filtering and normalization. These preprocessing steps ensure the accuracy, clarity and consistency of the data, support more efficient analysis and decision-making, optimize resource allocation, and reduce operating costs.

[0020] The short-circuit risk monitoring module is used to collect the short-circuit evaluation parameters of the distribution network in real time and monitor and evaluate the short-circuit faults of the distribution network; The specific process of the short-circuit risk monitoring module to monitor and evaluate the short-circuit faults of the distribution network is as follows: Obtain short-circuit evaluation parameters of the distribution network, which include operating current, operating voltage, insulation resistance, and temperature of key electrical components, generate a monitoring cycle ZQ, and divide the monitoring cycle ZQ into multiple monitoring periods {zq1, zq2, …, zqn}, that is, ZQ={zq1, zq2, …, zqn}; Obtain the operating current imbalance value of the distribution network within the monitoring period. The operating current imbalance value represents the ratio between the portion of the operating current variation difference greater than the preset operating current variation difference threshold value and the operating current variation difference within each monitoring period. The operating current variation difference represents the difference between the maximum value and the minimum value of the operating current. The operating current imbalance value is marked as DLS. Obtain the operating voltage imbalance value of the distribution network within the monitoring period. The operating voltage imbalance value represents the ratio between the portion of the operating voltage variation difference greater than the preset operating voltage variation difference threshold value and the operating voltage variation difference within each monitoring period. The operating voltage variation difference represents the difference between the maximum value and the minimum value of the operating voltage. The operating voltage imbalance value is marked as DYS. Obtain the insulation resistance imbalance value of the distribution network within the monitoring period. The insulation resistance imbalance value represents the ratio between the portion of the insulation resistance variation difference greater than the preset insulation resistance variation difference threshold value and the insulation resistance variation difference within each monitoring period. The insulation resistance variation difference represents the difference between the maximum value and the minimum value of the insulation resistance. The insulation resistance imbalance value is marked as JZS. Obtain the temperature imbalance value of the key electrical components of the distribution network within the monitoring period. The temperature imbalance value of the key electrical components indicates the ratio between the portion of the temperature variation difference of the key electrical components greater than the preset temperature variation difference threshold of the key electrical components within each monitoring period and the temperature variation difference of the key electrical components. The temperature variation difference of the key electrical components indicates the difference between the maximum value and the minimum value of the temperature of the key electrical components. The temperature imbalance value of the key electrical components is marked as ZWS. The operating current imbalance value DLS, operating voltage imbalance value DYS, insulation resistance imbalance value JZS and key electrical component temperature imbalance value ZWS are obtained, and the short-circuit fault assessment coefficient DGP is calculated using the following formula: ; Where c1, c2, c3 and c4 are all preset proportional factor coefficients, c4>c3>c2>c1>0, and the short-circuit fault assessment coefficient DGP is compared with the preset short-circuit fault assessment coefficient threshold: If the short-circuit fault assessment coefficient DGP is less than the preset short-circuit fault assessment coefficient threshold, it indicates that the distribution network is in normal operation; If the short-circuit fault assessment coefficient DGP is greater than or equal to the preset short-circuit fault assessment coefficient threshold, it indicates that the distribution network is in a short-circuit fault state; The short-circuit risk monitoring module collects key parameters of the distribution network in real time, calculates the imbalance value and derives the short-circuit fault assessment coefficient. It has high accuracy and real-time response capabilities, can quickly warn of potential short-circuit risks, and supports preventive maintenance. Its quantitative analysis enhances system reliability, flexibly adapts to different scenarios, reduces manual intervention, and improves operation and maintenance efficiency.

[0021] The load change prediction module is used to collect historical operating parameters of the distribution network, build a load change prediction model, predict the load change trend in the next day through the model, and identify potential overload risks based on the prediction results; The specific process of the load change prediction module constructing a load change prediction model and using the model to predict the load change trend within the next day is as follows: Obtain the historical operating parameters of the distribution network, including operating current, operating voltage, operating power, operating load, power factor and ambient temperature, generate a collection cycle, the collection cycle is one year long, divide the collection cycle into k collection periods, divide each collection period into m continuous sub-periods, and mark the midpoint of each sub-period to obtain m midpoints; Taking m midpoint moments as the base point, expand forward and backward by equal time lengths, mark s-1 expansion moments, and then summarize the midpoint moments and s-1 expansion moments to obtain s detection moments; The operating current values ​​at s detection moments are measured by a measuring instrument to obtain s detection operating current values, and the s detection operating current values ​​are accumulated and averaged to obtain m sub-operating current values; The expression of the sub-operating current value is: ; In the formula, ZDL zm is the sub-operation current value of the mth sub-cycle, ZDL jcmn is the nth detection operation current value of the mth sub-cycle; Remove the maximum and minimum sub-operating current values, and add up the remaining m-2 sub-operating current values ​​and calculate the average to obtain the operating current mean value; The expression of the mean operating current is: ; In the formula, DL yj is the average operating current, ZDL zp is the sub-operation current value of the p-th sub-cycle; The mean value of the operating voltage DY can be obtained by using the method of calculating the mean value of the operating current. yj , operating power average GL yj , operating load average FZ yj , power factor mean GY yj And the average ambient temperature HW yj , the average operating current DL yj , average operating voltage DY yj , operating power average GL yj , operating load average FZ yj , power factor mean GY yj And the average ambient temperature HW yj The load change prediction matrix FBY is constructed by combination, and the prediction matrix FBY is used as the input of the machine learning model. The load change matrix in the next day corresponding to each group of prediction matrices FBY is used as the output of the machine learning model. The load change matrix in the next day is used as the prediction target, and the sum of the prediction errors of all training data is minimized as the training target. The machine learning model is trained until the sum of the prediction errors reaches convergence, and the training is stopped to obtain the load change prediction model. The load change prediction model is expressed as follows: ; Where FZJ represents the load change matrix in the next day, β1, β2, β3, β4, β5 and β6 are regression coefficients, δ is the random error term, and ∆FBn represents the load change in the nth period in the next day; The real-time operating parameters of the distribution network are obtained, converted into the corresponding prediction matrix FBY and input into the load change prediction model. The real-time load change matrix FZJ in the next day is obtained through the load change prediction model, and the load change in the real-time load change matrix FZJ is compared with the preset load change threshold: If ∆FBn is less than the preset load change threshold, it indicates that the distribution network is in normal operation; If ∆FBn is greater than or equal to the preset load change threshold, it indicates that the distribution network is in an overloaded operation state; By collecting historical operating parameters and using machine learning models through the load change prediction module, high-precision predictions of load changes in the next day can be achieved, and potential overload risks can be monitored and warned in real time. This improves prediction accuracy, supports preventive maintenance, enhances the reliability and stability of the power system, reduces manual intervention, improves management efficiency, and can flexibly adapt to the needs of different distribution networks.

[0022] Embodiment 2: The technical solution of the embodiment of the present invention is different from that of embodiment 1 in that: like Figure 2 As shown, the current limiting strategy generation module is used to generate the optimal current limiting strategy according to the short-circuit fault evaluation result and the load change prediction result of the distribution network; The specific process of the current limiting strategy generation module generating the optimal current limiting strategy is as follows: Obtain the short-circuit fault assessment results and load change prediction results of the distribution network. The short-circuit fault assessment results indicate whether the current distribution network is in a short-circuit fault state. The load change prediction results indicate whether the distribution network is in an overloaded operation state at different time periods in the next day. Generate the optimal current limiting strategy based on the short-circuit fault assessment results and load change prediction results of the distribution network. The generated flow limiting strategy is sent to the cloud management platform. After receiving the flow limiting strategy, the cloud management platform immediately sends the corresponding control instructions to the execution agency for execution; The specific content of the current limiting strategy is: Current limiting strategy under short circuit fault condition: Current limiting threshold setting: According to the short-circuit current size I d and rated current I e , set the current limiting threshold I x , calculate the current limiting threshold I by the following formula x : I x =f·I e , where f is the safety factor, f=0.8, and the specific value of f is assigned according to the actual situation; Current limiting period division: The current limiting period is the period during which the short circuit fault occurs until the fault is eliminated; Current limiting action: Use a circuit breaker (such as a vacuum circuit breaker, SF6 circuit breaker) to quickly disconnect the faulty line, switch to the backup power source (such as a diesel generator, energy storage system) through an automatic switching device, and connect a current limiting reactor in series in the faulty line to limit the short-circuit current; Current limiting strategy in overload operation state: Current limiting threshold setting: according to the predicted load P y And rated capacity P e , set the current limiting threshold P x , calculate the current limiting threshold P by the following formula x : P x =P e -∆P, where ∆P is the safety margin, and the specific value of ∆P is assigned according to the actual situation; Current limiting period division: according to the prediction results, the current limiting period is divided. If the predicted overload period is from t1 to t2, the current limiting period is set to t1-∆t to t2+∆t, where ∆t is the buffer time and the specific value of ∆t is assigned according to the actual situation. Current limiting action: transfer part of the load to other lines through load transfer devices, enable energy storage systems (such as lithium battery energy storage systems) to provide additional power during overload periods, and notify users to reduce power loads through demand response systems; The current limiting strategy generation module formulates specific current limiting measures for short-circuit faults and overload conditions, ensuring the high safety of the system. It can respond to short-circuit or overload conditions in real time and react quickly to minimize the impact on the power grid. By reasonably distributing loads and enabling backup power supplies, resource utilization is optimized and system efficiency is improved. The current limiting threshold and buffer time are flexibly adjusted according to actual conditions to adapt to different power grid environments, reducing the need for manual intervention and improving operation and maintenance efficiency and overall reliability.

[0023] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solutions and improved concepts of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A self-learning power grid current limiting supervision system based on a cloud platform, characterized in that: It includes cloud management platform, data acquisition module, short circuit risk monitoring module, load change prediction module and current limiting strategy generation module; The data acquisition module is used to collect the operating parameters of the distribution network in real time and perform preprocessing operations on the collected operating parameters; The short-circuit risk monitoring module is used to collect the short-circuit evaluation parameters of the distribution network in real time and monitor and evaluate the short-circuit faults of the distribution network; The load change prediction module is used to collect historical operating parameters of the distribution network, build a load change prediction model, predict the load change trend in the next day through the model, and identify potential overload risks based on the prediction results; The current limiting strategy generation module is used to generate the optimal current limiting strategy according to the short-circuit fault evaluation results and load change prediction results of the distribution network.

2. According to a cloud platform-based self-learning power grid current limiting supervision system according to claim 1, it is characterized in that: The specific process of the short-circuit risk monitoring module monitoring and evaluating the short-circuit fault of the distribution network is as follows: Obtain short-circuit evaluation parameters of the distribution network, which include operating current, operating voltage, insulation resistance, and temperature of key electrical components, generate a monitoring cycle ZQ, and divide the monitoring cycle ZQ into multiple monitoring periods {zq1, zq2, …, zqn}, that is, ZQ={zq1, zq2, …, zqn}; Obtain the operating current imbalance value of the distribution network within the monitoring period. The operating current imbalance value represents the ratio between the portion of the operating current variation difference greater than the preset operating current variation difference threshold value and the operating current variation difference within each monitoring period. The operating current variation difference represents the difference between the maximum value and the minimum value of the operating current. The operating current imbalance value is marked as DLS. The operating voltage imbalance value of the distribution network within the monitoring period is obtained. The operating voltage imbalance value indicates the ratio between the part of the operating voltage variation difference greater than the preset operating voltage variation difference threshold value and the operating voltage variation difference within each monitoring period. The operating voltage variation difference indicates the difference between the maximum and minimum values ​​of the operating voltage. The operating voltage imbalance value is marked as DYS.

3. A self-learning power grid current limiting supervision system based on a cloud platform according to claim 2, characterized in that: Obtain the insulation resistance imbalance value of the distribution network within the monitoring period. The insulation resistance imbalance value represents the ratio between the portion of the insulation resistance variation difference greater than the preset insulation resistance variation difference threshold value and the insulation resistance variation difference within each monitoring period. The insulation resistance variation difference represents the difference between the maximum value and the minimum value of the insulation resistance. The insulation resistance imbalance value is marked as JZS. The temperature imbalance value of the key electrical components of the distribution network within the monitoring period is obtained. The temperature imbalance value of the key electrical components indicates the ratio between the portion of the temperature variation difference of the key electrical components in each monitoring period that is greater than the preset temperature variation difference threshold of the key electrical components and the temperature variation difference of the key electrical components. The temperature variation difference of the key electrical components indicates the difference between the maximum and minimum values ​​of the temperature of the key electrical components. The temperature imbalance value of the key electrical components is marked as ZWS.

4. A self-learning power grid current limiting supervision system based on a cloud platform according to claim 3, characterized in that: The operating current imbalance value DLS, operating voltage imbalance value DYS, insulation resistance imbalance value JZS and key electrical component temperature imbalance value ZWS are obtained, and the short-circuit fault assessment coefficient DGP is calculated using the following formula: ； Where c1, c2, c3 and c4 are all preset proportional factor coefficients, c4>c3>c2>c1>0, and the short-circuit fault assessment coefficient DGP is compared with the preset short-circuit fault assessment coefficient threshold: If the short-circuit fault assessment coefficient DGP is less than the preset short-circuit fault assessment coefficient threshold, it indicates that the distribution network is in normal operation; If the short-circuit fault assessment coefficient DGP is greater than or equal to the preset short-circuit fault assessment coefficient threshold, it indicates that the distribution network is in a short-circuit fault state.

5. The self-learning power grid current limiting supervision system based on a cloud platform according to claim 1 is characterized in that: The specific process of the load change prediction module constructing a load change prediction model and predicting the load change trend in the next day through the model is as follows: Obtain the historical operating parameters of the distribution network, including operating current, operating voltage, operating power, operating load, power factor and ambient temperature, generate a collection cycle, the collection cycle is one year long, divide the collection cycle into k collection periods, divide each collection period into m continuous sub-periods, and mark the midpoint of each sub-period to obtain m midpoints; Taking m midpoint moments as the base point, expand forward and backward by equal time lengths, mark s-1 expansion moments, and then summarize the midpoint moments and s-1 expansion moments to obtain s detection moments; The operating current values ​​at s detection moments are measured by a measuring instrument to obtain s detection operating current values, and the s detection operating current values ​​are accumulated and averaged to obtain m sub-operating current values.

6. A self-learning power grid current limiting supervision system based on a cloud platform according to claim 5, characterized in that: The expression of the sub-operating current value is: ; In the formula, ZDL zm is the sub-operation current value of the mth sub-cycle, ZDL jcmn is the nth detection operation current value of the mth sub-cycle; Remove the maximum and minimum sub-operating current values, and add up the remaining m-2 sub-operating current values ​​and calculate the average to obtain the operating current mean value; The expression of the mean operating current is: ; In the formula, DL yj is the average operating current, ZDL zp is the sub-operation current value of the p-th sub-cycle.

7. A cloud platform-based self-learning power grid current limiting supervision system according to claim 6, characterized in that: The mean value of the operating voltage DY can be obtained by using the method of calculating the mean value of the operating current. yj , operating power average GL yj , operating load average FZ yj , power factor mean GY yj And the average ambient temperature HW yj , the average operating current DL yj , average operating voltage DY yj , operating power average GL yj , operating load average FZ yj , power factor mean GY yj And the average ambient temperature HW yj A load change prediction matrix FBY is constructed in combination, and the prediction matrix FBY is used as the input of the machine learning model. The load change matrix in the next day corresponding to each group of prediction matrices FBY is used as the output of the machine learning model. The load change matrix in the next day is used as the prediction target, and the minimum sum of prediction errors of all training data is used as the training target. The machine learning model is trained until the sum of prediction errors converges and the training is stopped to obtain the load change prediction model.

8. A cloud platform-based self-learning power grid current limiting supervision system according to claim 7, characterized in that: The load change prediction model is expressed as follows: ； Where FZJ represents the load change matrix in the next day, β1, β2, β3, β4, β5 and β6 are regression coefficients, δ is the random error term, and ∆FBn represents the load change in the nth period in the next day; The real-time operating parameters of the distribution network are obtained, converted into the corresponding prediction matrix FBY and input into the load change prediction model. The real-time load change matrix FZJ in the next day is obtained through the load change prediction model, and the load change in the real-time load change matrix FZJ is compared with the preset load change threshold: If ∆FBn is less than the preset load change threshold, it indicates that the distribution network is in normal operation; If ∆FBn is greater than or equal to the preset load change threshold, it indicates that the distribution network is in an overloaded operating state.

9. The self-learning power grid current limiting supervision system based on a cloud platform according to claim 1 is characterized in that: The specific process of the current limiting strategy generation module generating the optimal current limiting strategy is as follows: Obtain the short-circuit fault assessment results and load change prediction results of the distribution network. The short-circuit fault assessment results indicate whether the current distribution network is in a short-circuit fault state. The load change prediction results indicate whether the distribution network is in an overloaded operation state at different time periods in the next day. Generate the optimal current limiting strategy based on the short-circuit fault assessment results and load change prediction results of the distribution network. The generated flow limiting strategy is sent to the cloud management platform. After receiving the flow limiting strategy, the cloud management platform immediately sends the corresponding control instructions to the execution agency for execution.

10. A cloud platform-based self-learning power grid current limiting supervision system according to claim 9, characterized in that: The specific content of the current limiting strategy is: Current limiting strategy under short circuit fault condition: Current limiting threshold setting: According to the short-circuit current size I d and rated current I e , set the current limiting threshold I x , calculate the current limiting threshold I by the following formula x : I x =f·I e , where f is the safety factor, f=0.8; Current limiting period division: The current limiting period is the period during which the short circuit fault occurs until the fault is eliminated; Current limiting action: Use a circuit breaker to quickly disconnect the faulty line, switch to the backup power supply through an automatic switching device, and connect a current limiting reactor in series in the faulty line to limit the short-circuit current; Current limiting strategy in overload operation state: Current limiting threshold setting: according to the predicted load P y And rated capacity P e , set the current limiting threshold P x , calculate the current limiting threshold P by the following formula x : P x =P e -∆P, where ∆P is the safety margin; Current limiting period division: according to the prediction results, the current limiting period is divided. If the predicted overload period is from t1 to t2, the current limiting period is set to t1-∆t to t2+∆t, where ∆t is the buffer time. Current limiting action: transfer part of the load to other lines through load transfer devices, enable energy storage systems to provide additional power during overload periods, and notify users to reduce electricity load through demand response systems.

Images