CN-122026501-A - Intelligent collaborative traversing system and method for unit vibration area based on dynamic game and self-adaptive prediction

CN122026501ACN 122026501 ACN122026501 ACN 122026501ACN-122026501-A

Abstract

The invention discloses an intelligent collaborative traversing system and method for a vibration area of a generator set based on dynamic game and self-adaptive prediction, which belong to the technical field of power system control and comprise the steps of deploying a sensor on each generator set and collecting running state data of the generator set; the method comprises the steps of preprocessing locally collected data by the edge nodes, predicting whether a generator set enters a vibration area within a preset time window by utilizing a locally deployed adaptive prediction model, collecting prediction results of all the edge nodes by a central control unit, constructing a dynamic game model, issuing a step-size limited adjustment instruction to the relevant generator set by the central control unit, adjusting a control strategy and parameters according to simulation deviation, continuously collecting feedback data after executing control operation, updating the adaptive prediction model, and realizing closed-loop optimization of the parameters. The invention can obviously improve the safety and efficiency of the crossing of the vibration area in the actual hydropower plant or intelligent micro-grid environment.

Inventors

LI MENG
XU ZHIWEN
Song teng
Yang Linxiong
ZHU JINHAI
LI XIAOQIANG
PENG SHIFEI
CHEN QINGSONG
HU JINPING
PU YA
ZHANG ZESHENG
PAN LEI
MA YINGSHUAI
ZHANG BO

Assignees

云南华电金沙江中游水电开发有限公司

Dates

Publication Date: 20260512
Application Date: 20250820

Claims (10)

1. The intelligent collaborative traversing system for the unit vibration area based on dynamic game and self-adaptive prediction is characterized by comprising a unit monitoring unit, an edge computing unit, a communication unit, a central control unit, a digital twin simulation unit and a federal learning training unit; Each generator set of the set monitoring unit is provided with a plurality of sensors, acquires set state data in real time, and sends the set state data to the local edge computing unit and the central control unit; the edge computing unit comprises edge computing nodes deployed beside each generator set and used for executing local data preprocessing and prediction computation; the communication unit comprises a bidirectional communication network constructed by adopting an industrial communication protocol, and is connected with each edge module and the central control unit to realize real-time transmission of monitoring data, prediction results and control instructions; The central control unit comprises a step of summarizing data and prediction results from all the groups and executing a dynamic game algorithm; The digital twin simulation unit is used for establishing a simulation model for each generator set and running in real time; The federal learning training unit includes training the predictive model using federal learning.
2. The intelligent collaborative traversing system for the vibration area of the machine set based on the dynamic game and the self-adaptive prediction, which is disclosed in claim 1, is characterized in that the edge computing unit and the machine set monitoring unit are deployed in one-to-one correspondence and are responsible for running the self-adaptive prediction model locally and accurately predicting the missed vibration trend; The communication network is used for connecting the edge computing unit and the central control unit to realize real-time exchange of data and instructions; The central control unit is used for establishing a dynamic game model based on the prediction data of each unit, solving Nash equilibrium and generating a cooperative control instruction of each unit.
3. The intelligent collaborative traversing system for a set vibration area based on dynamic gaming and adaptive prediction according to claim 2, wherein the central control unit comprises a game analysis module, a shape calculation module and an optimization solving module.
4. The intelligent collaborative traversing system for a set vibration area based on dynamic gaming and adaptive prediction as set forth in claim 3, wherein the game analysis module is configured to construct a game utility function with collaborative terms according to predicted forces and vibration indexes of each set; the Shapley calculation module is used for calculating the Shapley value of each participating unit It is indicated that the number of the elements is, Wherein, the For the Shapley value of each participating unit, N is a unit set which participates in traversing cooperation, S is a subset which does not contain the ith unit, i is the ith unit, f (S) is the cooperation benefit which can be achieved by the unit set S, and f (S { i } is the cooperation benefit which can be achieved by all the unit sets which participate in traversing cooperation; The optimization solving module is used for solving Nash equilibrium of the game and determining optimal output adjustment values of all the units.
5. The intelligent collaborative traversing system for the vibration area of the machine set based on dynamic gaming and adaptive prediction according to claim 4, wherein the adaptive prediction model comprises the advantages of integrating a long-short-term memory network with an autoregressive integral moving average model, comprehensively analyzing nonlinear and linear time series characteristics, and the dynamic gaming model adopts Shapley values of each machine set as a benefit distribution mechanism.
6. The intelligent collaborative traversal system of machine group vibration area based on dynamic gaming and adaptive prediction according to claim 5, wherein the autoregressive integrated moving average model linearly models the time series with differential, AR, and MA components, expressed as, Δ d Y t ＝φ 1 Y t-1 +…+φ p Y t-p +ε t +θ 1 ε t-1 +…+θ q ε t-q Wherein Δ d Y t is a stationary sequence after d-time differentiation, φ 1 ...φ p is a coefficient of an autoregressive term, θ 1 ...θ q is a coefficient of a moving average term, ε t ...ε t-q is a random error term, Y t-1 is a value of t-1 time series, p is an order of the autoregressive term, and q is an order of the moving average term.
7. The intelligent collaborative traversing system for a vibration area of a machine set based on dynamic gaming and adaptive prediction according to claim 6, wherein the central control unit adopts an event-triggered small-step control strategy, sets a trigger threshold e i (t) for each machine set i on the basis of real-time monitoring errors, and triggers the ith machine set to execute small-amplitude output pulse adjustment when meeting the level e i (t)||≥γ i ||e i (t k .
8. The intelligent collaborative traversing method for the set vibrating area based on the dynamic game and the self-adaptive prediction is applied to the intelligent collaborative traversing system for the set vibrating area based on the dynamic game and the self-adaptive prediction as claimed in any one of claims 1 to 7, and is characterized by comprising the following steps: A sensor is deployed on each generator set, running state data of the generator sets are collected, and the data are sent to an edge node and a central control unit; the edge node preprocesses locally acquired data, and predicts whether the generator set enters a vibration area within a preset time window by utilizing a locally deployed self-adaptive prediction model; Each edge node locally trains a self-adaptive prediction model, only uploads model parameters to a central control unit, the central control unit fuses the model parameters uploaded by each node, generates a unified model and transmits the unified model to each edge node for iterative training; The central control unit collects the prediction results of all edge nodes and builds a dynamic game model; when the triggering control condition is met, the central control unit issues a step-size-limited adjusting instruction to the related generator set, so that excessively quick adjustment is avoided; simulating the response of each generator set by using a digital twin model running synchronously with the physical system, and adjusting a control strategy and parameters according to simulation deviation; And after the control operation is executed, feedback data are continuously collected, and the adaptive prediction model is updated through the combination of a federal learning mechanism and the digital twin model, so that the closed-loop optimization of parameters is realized.
9. The intelligent collaborative traversing method for the vibration area of the machine set based on the dynamic game and the self-adaptive prediction according to claim 8, wherein the self-adaptive prediction model is independently trained on respective edge nodes and integrated and optimized through a central control unit.
10. The intelligent collaborative traversing method for the vibration area of the generator set based on the dynamic game and the self-adaptive prediction, as set forth in claim 9, is characterized in that the dynamic game model is constructed based on a dynamic game theory, a game relation is constructed by taking vibration stability and power adjustment cost of each generator set as a gain function, and control responsibility of each generator set in the adjustment process is distributed based on a shape value.

Description

Intelligent collaborative traversing system and method for unit vibration area based on dynamic game and self-adaptive prediction Technical Field The invention relates to the technical field of control of electric power systems, in particular to an intelligent collaborative traversing system and method for a vibration area of a unit based on dynamic game and self-adaptive prediction. Background In conventional power generation systems, the units present vibration regions, i.e., unstable regions in certain loads or speed ranges where the units are susceptible to mechanical or electrical oscillations. The unit can cause severe vibration, unstable operation and even shutdown accidents when the unit passes through the vibration area. In order to avoid crossing of the vibration area, smooth passing through by slowly changing the output or alternatively starting and stopping the machine group is generally required, but the method has low efficiency and poor flexibility. There have been studies to propose a dispatching optimization model considering the risk of traversing a vibrating area, aiming at reducing unnecessary traversing times and water consumption, for example, in a hydropower plant system, the unnecessary traversing times of the vibrating area can be effectively reduced by considering an optimal dispatching model of the risk of traversing the vibrating area. However, these methods are usually optimized at the planning level, lack dynamic coordinated control of real-time operation states of the units, and are difficult to cope with the complexity caused by renewable energy fluctuation and multi-unit isomerism. The potential of combining dynamic gaming with adaptive prediction in coordinating multi-unit ride through control is not fully exploited in the prior art, and an intelligent system which is structured and easy to deploy is not constructed. Therefore, a new system architecture and method are needed to cooperatively control multiple units to realize safe and efficient crossing of the vibration area. Disclosure of Invention The present invention has been made in view of the above-described problems. Therefore, the invention solves the technical problem of how to solve the problem that the prior art has not fully explored the potential of combining dynamic gaming with adaptive prediction in the aspect of coordinating multi-group ride-through control. The intelligent collaborative traversing system for the generator set vibration area comprises a generator set monitoring unit, an edge computing unit, a communication unit, a central control unit, a digital twin simulation unit and a federal learning training unit, wherein each generator set of the generator set monitoring unit is provided with various sensors, state data of the generator set are collected in real time and sent to the local edge computing unit and the central control unit, the edge computing unit comprises edge computing nodes arranged beside each generator set and used for executing local data preprocessing and prediction computation, the communication unit comprises a bidirectional communication network constructed by adopting an industrial communication protocol and connected with each edge module and the central control unit, real-time transmission of monitoring data, prediction results and control instructions is achieved, the central control unit comprises a game and execution of dynamic algorithms, the digital twin simulation unit is used for building a simulation model for each generator set and running in real time, and the federal learning unit comprises a federal learning training prediction model. The intelligent collaborative traversing system for the unit vibration area based on dynamic game and self-adaptive prediction is characterized in that the edge computing unit and the unit monitoring unit are arranged in a one-to-one correspondence mode and are responsible for running the self-adaptive prediction model locally to accurately predict the future vibration trend, the communication network is used for connecting the edge computing unit and the central control unit to realize real-time exchange of data and instructions, and the central control unit is used for establishing a dynamic game model based on the predicted data of each unit and solving Nash equilibrium to generate collaborative control instructions of each unit. As a preferable scheme of the intelligent collaborative traversing system for the unit vibration area based on dynamic game and self-adaptive prediction, the central control unit comprises a game analysis module, a Shapley calculation module and an optimization solving module. As a preferable scheme of the intelligent cooperative traversing system of the unit vibration area based on dynamic game and self-adaptive prediction, the invention is characterized in that the game analysis module is used for constructing a game utility function with cooperative items according to the predicted force and vibratio