Research On Fault Prediction And Maintenance Of Oil Drilling Rig Microgrid Based On Deep Learning

This paper is based on the major science and technology project of Sichuan Province in 2019 "Intelligent Drill Rig Development and Application".In recent years,with the continuous upgrading of the scale of oil rig platforms,microgrid as its core system,and the scale of the oil rig microgrid has also been continuously expanded.This paper is dedicated to reasonably and effectively mining and analyzing the power data of the oil rig microgrid index surge.Taking the main power equipment of the oil rig microgrid as the research object,the structure and main fault types are analyzed,established a prediction model of power equipment monitoring indicators and a fault classification and identification model,and formulate maintenance strategies that consider the failure prediction results and maintenance risks.Specifically,it mainly includes the following parts:First,this paper analyzes the mechanism of the main power equipment failures of the oil rig microgrid,and establishes a fault tree analysis model for the power equipment.Finally,the core monitoring indicators of the main equipment are given,and consider the impact of external factors on power equipment.The amount of auxiliary information of the monitoring indicators is divided.Secondly,aiming at the problem that the amount of auxiliary information in the input of the predictive model is not quantified,combined with the idea of combination weighting,the amount of auxiliary information is combined weighting and quantifying.Further,a power equipment monitoring index prediction model based on the Gated Recurrent Unit(GRU)network structure is established,and the network structure of the prediction model is introduced.Finally,the reliability of the prediction model is verified through experiments.Then,considering the problem of fewer fault samples in the oil rig microgrid system,the optimized Softmax is used as the output classifier to establish a fault classification and recognition model for power equipment based on the Multi-LSTM(MLSTM).The predicted value of power equipment monitoring indicators is input into the trained fault classification and recognition model,and the corresponding fault type and occurrence probability are output to realize the prediction of power equipment faults.Validation of examples proves the validity of the model.Finally,based on the failure prediction results of the power equipment of the oil rig,it is considered that the maintenance methods required under different states of the power equipment are different,and the maintenance time and cost required for different maintenance methods are also different.This paper takes maintenance time and maintenance methods as decision variables,evaluates the operation risk of the oil rig microgrid's failure risk,sets up constraints such as power equipment status,grid stability,maintenance resources,etc.,and establishes a failure prediction and risk assessment based The oil rig microgrid maintenance strategy is solved by the dual-group learning particle swarm optimization algorithm,which verifies the validity of the theory proposed in this paper.This article provides some basic research ideas for the failure prediction and maintenance of the oil rig platform microgrid,and provides method support for effectively improving the safety and reliability of the oil rig microgrid under complex working conditions.