CN-122000898-A - Power distribution network power inversion risk processing method and device

CN122000898ACN 122000898 ACN122000898 ACN 122000898ACN-122000898-A

Abstract

The embodiment of the application provides a power distribution network power inversion risk processing method and device. The method comprises the steps of obtaining multi-source heterogeneous data, carrying out space-time alignment processing through a data alignment algorithm to obtain space-time aligned multi-mode data, inputting the space-time aligned multi-mode data into a hybrid model to carry out feature extraction and dynamic attention distribution, generating an output predicted value and a load demand predicted value of a distributed power supply connected to a power distribution network in a future period, determining a power inverse air supply risk assessment index of the distributed power supply to the power distribution network based on the output predicted value and the load demand predicted value, and triggering a multi-target self-adaptive defense decision model based on dynamic games if the power inverse air supply risk assessment index is larger than a preset risk threshold value to generate a power inverse air supply risk coping strategy of the power distribution network. The method is used for actively coping with the power reversal and improving the operation safety of the power grid.

Inventors

CHEN DEYANG
LIN ZHIHANG
CAO DEFA
REN MENG
YI YANG
WANG YANWEI
LIU YU
MIAO LU
LUO WEI
LAN XIAOBIN

Assignees

广东电网有限责任公司梅州供电局

Dates

Publication Date: 20260508
Application Date: 20260202

Claims (10)

1. The power inversion risk treatment method for the power distribution network is characterized by comprising the following steps of: Acquiring multi-source heterogeneous data, and performing space-time alignment processing through a data alignment algorithm to obtain space-time aligned multi-mode data, wherein the multi-source heterogeneous data comprises power grid internal structured data and power grid external unstructured data; Inputting the multi-mode data after time-space alignment into a hybrid model for feature extraction and dynamic attention distribution, and generating an output predicted value and a load demand predicted value of a distributed power supply connected to a power distribution network in a future period; Determining a power reversal risk assessment index of the distributed power supply to the power distribution network based on the output predicted value and the load demand predicted value; And if the power back-off risk assessment index is larger than a preset risk threshold, triggering a multi-target self-adaptive defense decision model based on dynamic games, and generating a power back-off risk coping strategy of the power distribution network.
2. The method of claim 1, wherein the grid internal structured data comprises historical voltage and real-time voltage data, historical current and real-time current data, historical power and real-time power data, and wherein the grid external unstructured data comprises grid weather forecast data, holiday information, and social event information.
3. The method of claim 1, wherein the hybrid model comprises a long-term memory network and a transducer encoder; The step of inputting the multi-mode data after space-time alignment into the mixed model for feature extraction and dynamic attention distribution comprises the following steps: Extracting time sequence characteristics of the multi-mode data through a long-short-term memory network, obtaining local time sequence characteristics, and performing position coding on the local time sequence characteristics to obtain a coded characteristic sequence; And performing attention weight distribution on the characteristic sequence by using a multi-head attention mechanism through the transducer encoder so as to determine the space-time characteristic with highest correlation with power dumping, and determining the output predicted value and the load demand predicted value based on the space-time characteristic.
4. A method according to claim 3, wherein said determining a power back-off risk assessment indicator of the power distribution network by the distributed power source comprises: Based on the output predicted value and the load demand predicted value, node voltage and branch power flow distribution of the power distribution network in a future time section are obtained through power flow calculation by utilizing power grid topological parameters; Determining the geometric safety distance between the running state of the power distribution network and the safety domain boundary at a future time section based on the node voltage and branch power flow distribution; acquiring a node voltage out-of-limit risk index and a branch power flow out-of-limit risk index, and fusing to obtain a comprehensive risk index; and generating the power inverse air supply risk assessment index based on the geometric safety distance and/or the comprehensive risk index.
5. The method of claim 4, wherein the security domain boundary comprises at least one of a line capacity constraint, a voltage limit constraint, a main transformer capacity constraint, a feeder-to-main transformer capacity matching constraint, a distributed power access constraint, and a power-dump prevention constraint.
6. The method according to claim 1, wherein the method further comprises: Constructing an defender, a defender strategy set, an attacker and an attacker strategy set, wherein the defender represents a power grid regulation and control system of the power distribution network, the defender strategy set comprises at least one group of cooperative control action combinations, the cooperative control action combinations are used for adjusting the running state of the power distribution network to eliminate the power dumping danger, the attacker represents the uncertainty factor of the power dumping, and the attacker strategy set is used for simulating the least adverse uncertainty scene causing the power dumping; constructing an defender income function based on the system network loss cost, the voltage deviation penalty term, the control equipment action cost, the voltage upper limit penalty term and the branch current out-of-limit penalty term; Constructing an attacker benefit function based on the voltage upper limit degree, the branch current limit degree, the distributed power output limit degree and the power dumping degree; Constructing constraint conditions, wherein the constraint conditions comprise power balance constraint, voltage safety constraint, branch capacity constraint, equipment operation constraint, equipment action frequency limit, power dumping constraint and topology connectivity constraint; and constructing the multi-objective self-adaptive defense decision model of the dynamic game based on the defenders, the defender policy set, the defender benefit function, the aggressors, the aggressor policy set, the aggressor benefit function and the constraint conditions.
7. The method of claim 6, wherein the triggering a dynamic gaming-based multi-objective adaptive defense decision model to generate a power back-off risk countermeasure policy for the power distribution network comprises: solving balance points of the multi-target self-adaptive defense decision model by adopting a mixed integer linear programming method, wherein the balance points are used for representing an defender to predict the optimal reaction of an attacker to any strategy in a defender strategy set; and generating a power inverse supply risk coping strategy of the power distribution network based on the balance points.
8. The method of claim 6, wherein the defender policy set comprises an on-load tap position vector, a capacitive reactor switching state vector, a distributed power source active output adjustment vector, and a flexible load adjustment vector, and wherein the aggressor policy set comprises a distributed power source output disturbance vector and a load demand disturbance vector.
9. The method according to claim 1, wherein the method further comprises: Based on the power-down risk coping strategy, a control instruction set is issued to an edge intelligent agent, and the edge intelligent agent is deployed at a transformer substation or a key node of the power distribution network and is used for converting the control instruction set into equipment instructions and issuing the equipment instructions to end-side equipment for execution; Acquiring an actual execution effect of the strategy uploaded by the edge agent; based on the actual execution effect of the strategy, an online incremental learning algorithm is adopted to update the model parameters of the hybrid model and/or the model parameters of the multi-target self-adaptive defense decision model.
10. The utility model provides a power distribution network power falls air to dangerous processing apparatus which characterized in that includes: the system comprises an acquisition module, a data alignment module and a data processing module, wherein the acquisition module is used for acquiring multi-source heterogeneous data, and performing space-time alignment processing through a data alignment algorithm to obtain space-time aligned multi-mode data, wherein the multi-source heterogeneous data comprises power grid internal structured data and power grid external unstructured data; the prediction module is used for inputting the multi-mode data after time-space alignment into the hybrid model for feature extraction and dynamic attention distribution, and generating a predicted output value and a predicted load demand value of a distributed power supply connected to the power distribution network in a future period; The determining module is used for determining a power inverse risk assessment index of the distributed power supply to the power distribution network based on the output predicted value and the load demand predicted value; The generation module is used for triggering a multi-target self-adaptive defense decision model based on dynamic games to generate a power inverse risk coping strategy of the power distribution network if the power inverse risk assessment index is larger than a preset risk threshold.

Description

Power distribution network power inversion risk processing method and device Technical Field The application relates to the technical field of electric power, in particular to a power distribution network power inverse air supply risk processing method and device. Background As high-proportion distributed energy sources such as photovoltaic and wind power are connected into a power distribution network, power dumping becomes a significant problem threatening the safe operation of the power distribution network under the specific scene that the output peak and the load valley of the distributed power source are overlapped. The power dumping is extremely easy to cause a series of safety risks such as voltage out-of-limit, equipment overload, relay protection misoperation (such as out-of-order tripping) and the like. At present, the existing method for coping with power reversal risk is mainly based on a 'post response' mode, namely, passive intervention is carried out by means of adjusting a transformer tap, switching reactive compensation equipment and the like after power reversal occurs. However, in the conventional countermeasure, there is a control action lag, which reduces the operation safety of the power grid. Disclosure of Invention The power distribution network power reversal risk processing method and device provided by the embodiment of the application are used for actively coping with power reversal and improving the operation safety of the power distribution network. In a first aspect, an embodiment of the present application provides a power distribution network power inverse air supply risk processing method, including: Acquiring multi-source heterogeneous data, and performing space-time alignment processing through a data alignment algorithm to obtain space-time aligned multi-mode data, wherein the multi-source heterogeneous data comprises power grid internal structured data and power grid external unstructured data; Inputting the multi-mode data after time-space alignment into a hybrid model for feature extraction and dynamic attention distribution, and generating an output predicted value and a load demand predicted value of a distributed power supply connected to a power distribution network in a future period; Determining a power reversal risk assessment index of the distributed power supply to the power distribution network based on the output predicted value and the load demand predicted value; And if the power back-off risk assessment index is larger than a preset risk threshold, triggering a multi-target self-adaptive defense decision model based on dynamic games, and generating a power back-off risk coping strategy of the power distribution network. In one possible implementation, the grid internal structured data includes historical voltage and real-time voltage data, historical current and real-time current data, historical power and real-time power data, and the grid external unstructured data includes grid weather forecast data, holiday information and social event information. In one possible implementation, the hybrid model includes a long-short-term memory network and a transducer encoder; The step of inputting the multi-mode data after space-time alignment into the mixed model for feature extraction and dynamic attention distribution comprises the following steps: Extracting time sequence characteristics of the multi-mode data through a long-short-term memory network, obtaining local time sequence characteristics, and performing position coding on the local time sequence characteristics to obtain a coded characteristic sequence; And performing attention weight distribution on the characteristic sequence by using a multi-head attention mechanism through the transducer encoder so as to determine the space-time characteristic with highest correlation with power dumping, and determining the output predicted value and the load demand predicted value based on the space-time characteristic. In one possible implementation manner, the determining the power back-off risk assessment index of the distributed power source to the power distribution network includes: Based on the output predicted value and the load demand predicted value, node voltage and branch power flow distribution of the power distribution network in a future time section are obtained through power flow calculation by utilizing power grid topological parameters; Determining the geometric safety distance between the running state of the power distribution network and the safety domain boundary at a future time section based on the node voltage and branch power flow distribution; acquiring a node voltage out-of-limit risk index and a branch power flow out-of-limit risk index, and fusing to obtain a comprehensive risk index; and generating the power inverse air supply risk assessment index based on the geometric safety distance and/or the comprehensive risk index. In one possible implementation, the security domain bound