Fault Analysis And Autonomous Special Patrol System Of Substation Patrol Robot Based On Data Mining Technology

With the construction of smart grid and the popularization of substation manual substitution technology in recent years,substation automatic inspection robot has been widely used in substations at all levels.However,the current patrol robot system can not automatically find and analyze the equipment with hidden troubles.It can only generate reports on the equipment with defects.Many hidden troubles can only be predicted and checked manually,which increases the production burden of operation and maintenance personnel.In order to enhance the efficiency of robots in substation inspection,make the inspection robots better serve the operation and maintenance work,and reduce the burden of personnel,this paper proposes using data mining technology to analyze the operation data of substations,so that the inspection robot system can independently complete the fault diagnosis,find the relevant inspection points,and formulate the special inspection plan.The main work is as follows:Firstly,the three-party integrated database of substation is constructed.In order to collect equipment operation data needed in data mining process,the production system,monitoring background and robot patrol data are integrated to form a comprehensive database for defective equipment mining after removing noise interference.Taking the background of the robot as the data analysis platform of the comprehensive database,we optimize the sampling database of the robot,construct a more scientific inspection strategy,improve the inspection efficiency of the inspection robot,and achieve targeted data acquisition,so as to improve the quality of the sample of the comprehensive database.At the same time,knowledge base and fault information base are constructed based on production management system,which provide basis for defect classification and defect-related fault points mining.Secondly,equipment fault mining is completed.In-depth study of data mining applications,screening methods suitable for fault analysis,to determine a combination of rough sets and fault tree data mining methods.Rough set method is used to effectively solve the problems of noise and complexity of substation operation data,and fault tree method is used to mine equipment fault types.Taking transformer autonomous state assessment as an example,combining expert evaluation and equipment maintenance professional knowledge base(State Grid Corporation's substation detection and management regulations,substation site regulations),this paper discusses the fault tree construction which can realize automatic fault diagnosis of robot backstage system.Thirdly,completing fault-related inspection point correlation.The inspection period and cycle of each fault are determined by classification of defect grade in knowledge base.The necessity of association between fault information base and account information in production system is introduced,and the algorithm of association degree mining is introduced.Compared with Apriori algorithm,FP-growth algorithm does not generate candidate frequent sets,and the mining efficiency is significantly improved.Finally,the FP-growth algorithm is proposed to mine the fault information base.By mining the correlation information between the fault inspection points,the location of the inspection points should be selected,and all the inspection points should be inspected after the fault occurs should be determined and prioritized.Fourthly,the strategy formulation and route planning of robot special patrol are completed.This paper discusses the path optimization methods for solving multiple special patrol plans,and makes simulation verification and comparison.Finally,it decides to use Dijkstra algorithm and simulated annealing algorithm to optimize the special patrol route of multi-fault equipment,so that the robot can automatically carry out the optimal strategy special patrol of the fault equipment patrol points determined by the background analysis system without affecting the daily routine patrol.