CN-122026374-A - Multi-energy power distribution network intelligent switching method, system, equipment and medium based on self-adaptive control

CN122026374ACN 122026374 ACN122026374 ACN 122026374ACN-122026374-A

Abstract

The invention relates to the technical field of power system automation, and discloses a multi-energy power distribution network intelligent switching method, a system, equipment and a medium based on self-adaptive control, which comprise the steps of acquiring multi-source heterogeneous data such as power grid operation data, meteorological information, load characteristics and the like in real time by constructing a multi-dimensional state sensing network; the method comprises the steps of establishing an intelligent decision model based on a deep Q network, realizing autonomous learning and dynamic optimization of a switching strategy, designing a multi-time scale coordinated control architecture, completing fault detection in microsecond level, switching decision in millisecond level and system stability adjustment in second level, developing a hybrid energy storage buffer technology, effectively inhibiting switching transient impact, establishing a multi-objective optimization function considering equipment service life and economic cost, and realizing comprehensive balance of power supply reliability, economy and equipment health management.

Inventors

LI ZHEN
ZHANG YAN
Luo Zhongke
WAN HUIJIANG
DENG PU
ZHANG ZHAOFENG
YANG JIERUI
LI QINGSHENG
LONG JIAHUAN
CHEN JULONG
ZHANG YU
TAN ZHUKUI
TANG XUEYONG
WU PENG
GAO HUA

Assignees

贵州电网有限责任公司

Dates

Publication Date: 20260512
Application Date: 20251203

Claims (10)

1. The intelligent switching method of the multi-energy power distribution network based on the self-adaptive control is characterized by comprising the following steps of, Performing hardware self-checking, loading pre-training model parameters and initializing a system state, collecting power grid operation data, environment parameters and load characteristics in real time, and predicting renewable energy output and load requirements; Detecting power grid faults through a signal processing technology, starting isolation measures, and generating a switching action sequence according to the current system state by using a deep reinforcement learning model; generating instructions of all control units according to the decision result, realizing multi-time scale coordination control, executing switching actions according to the optimized switching sequence, and monitoring transient processes; evaluating the switched system state, calculating a reward value, and updating the deep reinforcement learning model parameters based on experience playback; and adjusting control parameters according to the operation statistics, and starting an emergency response mechanism when the abnormality is detected.
2. The intelligent switching method of the multi-energy distribution network based on self-adaptive control of claim 1, wherein the predicting renewable energy output and load demand comprises acquiring node voltage, branch current, system frequency, ambient temperature, energy storage charge state and load demand power in real time through a distributed measurement unit to obtain a system state vector; Predicting photovoltaic power generation power and wind power based on meteorological information to obtain a renewable energy output predicted value; taking prediction uncertainty into consideration, and quantizing a prediction error by using a probability distribution model to obtain probability distribution of predicted power; and carrying out state estimation and prediction on the acquired data through a communication delay compensation technology to obtain an accurate current state of the system.
3. The intelligent switching method of the multi-energy distribution network based on self-adaptive control of claim 2, wherein the steps of detecting the power grid fault and starting the isolation measure comprise analyzing voltage and current waveforms through wavelet transformation, extracting transient characteristics and obtaining wavelet coefficients and energy distribution; Calculating energy distribution of different scales based on energy criteria, and judging fault occurrence conditions to obtain a fault judging result; When a fault is detected, an emergency response is immediately initiated, normal switching operations are suspended, and a fault isolation policy is invoked.
4. The intelligent switching method of the multi-energy power distribution network based on self-adaptive control of claim 3, wherein the generating a switching action sequence comprises inputting a current state into a deep Q network, and calculating Q values of all optional actions through forward propagation to obtain an action cost function; Selecting actions according to the exploration strategy, evaluating switching feasibility, and checking equipment states, power grid conditions and energy storage constraints; when a plurality of switch actions are involved, the action time and sequence of each switch are determined through an optimization algorithm, and an optimized switching sequence is obtained.
5. The intelligent switching method of the multi-energy distribution network based on self-adaptive control of claim 4, wherein the deep Q network comprises the steps of calculating a Q value through a state-action cost function by using a multi-layer fully-connected neural network, wherein a loss function adopts a time sequence differential error and updates network parameters through a gradient descent method, and an experience playback buffer zone defines sample priority based on the time sequence differential error and performs sample selection through importance sampling weights.
6. The intelligent switching method of the multi-energy distribution network based on self-adaptive control of claim 5, wherein the achieving of the multi-time scale coordination control comprises the steps of ultra-fast controlling a time scale to be microsecond and stabilizing voltage by adopting a feedback control transfer function; The rapid control time scale is in millisecond level, a model prediction control optimization switching process is adopted, and deviation between prediction output and a reference signal is minimized through a control input sequence; the slow optimization time scale is in the order of minutes, and an economic dispatch optimization objective function is adopted to minimize the power generation cost, the electricity purchasing cost and the energy storage degradation cost.
7. The intelligent switching method of the multi-energy distribution network based on self-adaptive control according to claim 6, wherein the transient monitoring process comprises the steps of pre-charging an energy storage system before switching, calculating energy storage output power through proportional-integral-derivative control, and checking energy storage charge state and power constraint; Monitoring voltage change, predicting transient process by adopting a second-order oscillation model, and compensating voltage deviation by energy storage output; And executing switching action according to the minimum time interval, avoiding action conflict, and monitoring whether the impact current is in a safety range.
8. The intelligent switching system of the multi-energy power distribution network based on the self-adaptive control is applied to the intelligent switching method of the multi-energy power distribution network based on the self-adaptive control as claimed in any one of claims 1 to 7, and is characterized by comprising a state sensing and predicting module, an intelligent decision module, a coordination control and executing module, an online learning and self-adaptive module and a fault processing and abnormality management module; The state sensing and predicting module is used for collecting power grid operation data, environment parameters and load characteristics in real time and predicting renewable energy output and load demands; the intelligent decision module generates an optimized switching action sequence according to the current system state through a deep reinforcement learning model; The coordination control and execution module generates instructions of all control units according to the decision result, realizes coordination control of multiple time scales, and executes switching actions according to the optimized switching sequence; the online learning and self-adapting module is used for updating the parameters of the deep reinforcement learning model based on experience playback in the running process of the system, and dynamically adjusting the control parameters according to running statistical data so that the system can continuously optimize the performance of the system; The fault processing and abnormality management module is used for detecting the power grid fault in real time through a signal processing technology, and immediately starting corresponding emergency response and isolation measures when the fault or abnormality is detected.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the adaptive control-based intelligent switching method of a multi-energy distribution network according to any one of claims 1 to 7.
10. A computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor realizes the steps of the adaptive control based intelligent switching method of a multi-energy distribution network according to any of claims 1 to 7.

Description

Multi-energy power distribution network intelligent switching method, system, equipment and medium based on self-adaptive control Technical Field The invention relates to the technical field of power system automation, in particular to a multi-energy power distribution network intelligent switching method, system, equipment and medium based on self-adaptive control. Background With the large-scale access of renewable energy sources, modern power distribution networks exhibit complex operation characteristics of multi-energy power supply and multi-path interconnection. The traditional power distribution network switching system mainly depends on preset control logic and fixed switching strategies, and exposes various technical defects when facing a novel power distribution network with multiple energy sources, frequent load fluctuation and complex operation conditions. The traditional switching control adopts a fixed threshold trigger mechanism, so that the dynamic requirements of different operation scenes cannot be met, the switching time is unreasonable, unnecessary frequent switching is possibly caused, equipment abrasion is increased, and the switching is possibly delayed, so that the power supply quality is possibly reduced. Disclosure of Invention In view of the existing problems, the invention provides a multi-energy power distribution network intelligent switching method, a system, equipment and a medium based on self-adaptive control. Therefore, the technical problems of low response speed, large switching impact, difficult multi-energy coordination, weak self-adaptive capacity and the like of the traditional multi-energy power distribution network switching system are solved. The invention provides a multi-energy power distribution network intelligent switching method based on self-adaptive control, which comprises the following steps of performing hardware self-checking, loading pre-training model parameters and initializing system states, collecting power grid operation data, environment parameters and load characteristics in real time, predicting renewable energy output and load demands, detecting power grid faults through a signal processing technology, starting isolation measures, generating a switching action sequence according to the current system states by using a deep reinforcement learning model, generating instructions of each control unit according to decision results, realizing multi-time scale coordination control, executing switching actions according to the optimized switching sequence, monitoring transient processes, evaluating the switched system states, calculating rewarding values, and updating the deep reinforcement learning model parameters based on experience playback; and adjusting control parameters according to the operation statistics, and starting an emergency response mechanism when the abnormality is detected. The method for intelligently switching the multi-energy distribution network based on the self-adaptive control is characterized in that the predicting renewable energy output and load demand comprises the steps of acquiring node voltage, branch current, system frequency, ambient temperature, energy storage charge state and load demand power in real time through a distributed measuring unit to obtain a system state vector; Predicting photovoltaic power generation power and wind power based on meteorological information to obtain a renewable energy output predicted value; taking prediction uncertainty into consideration, and quantizing a prediction error by using a probability distribution model to obtain probability distribution of predicted power; and carrying out state estimation and prediction on the acquired data through a communication delay compensation technology to obtain an accurate current state of the system. The intelligent switching method for the multi-energy distribution network based on the self-adaptive control is characterized in that the steps of detecting the power network fault and starting the isolation measure comprise the steps of analyzing voltage and current waveforms through wavelet transformation, extracting transient characteristics and obtaining wavelet coefficients and energy distribution; Calculating energy distribution of different scales based on energy criteria, and judging fault occurrence conditions to obtain a fault judging result; When a fault is detected, an emergency response is immediately initiated, normal switching operations are suspended, and a fault isolation policy is invoked. The method for intelligently switching the multi-energy distribution network based on the self-adaptive control comprises the steps of inputting a current state into a deep Q network, and calculating Q values of all optional actions through forward propagation to obtain an action cost function; Selecting actions according to the exploration strategy, evaluating switching feasibility, and checking equipment states, power grid conditions and e