Research On Fault Diagnosis And Intelligent Alert Method Of Flywheel Energy Storage System

Power-side energy storage technology is a crucial solution for effectively reducing the challenges of integrating large-scale renewable energy into the grid.Among various energy storage methods,the flywheel energy storage system stands out due to its numerous advantages,such as non-pollution,high energy storage density,high energy conversion efficiency,adaptability to different environments,a long lifespan,and easy maintenance.Therefore,the system has emerged as the most promising energy storage method currently available.However,during long-term operation,flywheel energy storage systems have a relatively high failure rate due to heavy mechanical and therm al loads.These failures may lead to a decline in system stability and reliability,system damage,or even endanger personnel safety.Hence,the study of fault diagnosis and early warning technology for flywheel energy storage is highly important.In this regard,this paper proposes a solution to the problem of flywheel energy storage fault diagnosis.Specifically,the paper employs the information entropy fusion method to extract the fault symptom set and establishes a neural fault Petri net model to achieve precise fault diagnosis.Secondly,aiming at the early warning problem of flywheel energy storage faults,a prediction model for the normal operation state of flywheel energy storage is established using deep learning theory to capture the fault development process,and a reasonable early warning strategy is proposed to achieve early warning of flywheel energy storage operating parameters,which can maintain stable and reliable operation of flywheel energy storage.The main research contents of this paper are as follows:Firstly,this article presents an overview of the flywheel energy storage system,including its working principle and components.Symptom parameters are extracted according to common faults in each component.Furthermore,this article establishes a set of symptoms for detecting faults in flywheel energy storage systems.Additionally,it proposes a complexity evaluation index that uses the information entropy fusion method.Symptom parameters are reduced according to the amount of information,and the optimal selection of the fault set of the flywheel energy storage system is finally determined.From a qualitative and quantitative perspective,a fault diagnosis method for flywheel energy storage is studied,and a fault diagnosis method for flywheel energy storage based on neural Petri nets is proposed.This paper introduces the fault Petri net model and ignition rules,establishes a flywheel energy storage fault diagnosis model based on Petri,and solves the model using the incidence matrix method.Combining the learning ability of neural networks with the advantages of Petri nets,a flywheel energy storage fault diagnosis model based on neural fault Petri nets is established.To confirm the model’s accuracy,the paper conducts a simulation of the actual operation data from a flywheel energy storage power station.The simulation results demonstrate that this method can effectively detect and diagnose faults in flywheel energy storage systems.This paper investigates a method for early warning of single parameter faults in flywheel energy storage systems.The method involves using the operating state parameters from the previous moment as input into a cyclic neural network.Moreover,the paper utilizes the output of the cyclic neural network to predict the operating state parameters of the next moment.This approach establishes a prediction model for the state of flywheel energy storage systems based on a cyclic neural network.The maximum kernel density estimate of the residual sequence between the neural network prediction value and the actual operating data is used as the initial domain value for early warning.The paper also determines a health index that reflects the current operating status of the equipment.This index is determined based on the residual probability distribution histogram within a sliding window.To validate the accuracy of the proposed model,the paper conducts a simulation using actual operation data from a flywheel energy storage power station.The results of this simulation demonstrate that the model can effectively predict the degree of failure in flywheel energy storage systems with high accuracy.