Fault Diagnosis Research Of High Speed Railway Turnout Based On Deep Forest

The railway turnout system is one of the most important equipment in the railway infrastructure,and its status will directly affect the driving safety and operational efficiency of the entire railway system.With the continuous speed increase of high-speed railway in China,the reliability requirement of turnout equipment is becoming more and more stringent,and higher requirement is imposed on the maintenance of turnouts.At present,the electric power sections of major railway bureaus in China generally use the skylight and periodic maintenance to maintain the turnout equipment.Due to the insufficient granularity of the centralized monitoring system at this stage,the cause of the failure is still mainly analyzed manually in the event of a failure,which greatly depends on the professional knowledge and practical experience of the field personnel.This method is difficult to meet the needs of current railways in terms of diagnostic efficiency and cost,so it is necessary to conduct intelligent research on the fault diagnosis algorithm.In order to solve the above problems,this paper conducts on-the-spot investigations and studies a large number of documents to explore the working principle and failure mode of the high-speed railway turnout system.Due to the complex structure of the speed-up turnout system and the diverse field environment,it is difficult to establish a precise model from the mechanism.Therefore,this paper solves the problem through data modeling.Considering the small sample size of the turnouts'failure data and the high dimensionality of the action power curve,this paper proposes a fault diagnosis algorithm based on deep forest for speed-up turnout,and compares it with other algorithms in the literature to verify the superiority of the method.On the basis of this,the diagnosis of the secondary fault of the electrical fault is indicated by the indicating voltage.The main work of the thesis is as follows:(1)Requirement analysis of the turnout system:a large number of documents are investigated to establish the current development status and problems of fault diagnosis methods.The widely-used ZYJ7 turnout system is selected as the research object,and its operating process and principle are analyzed.The real-world demands for fault diagnosis of speed-up turnout system are provided.(2)Failure mode analysis:fault sample of turnouts is obtained by on-site investigations.This paper summarizes the common failure modes by referring to the literature and communicating with experts,and this paper also combines the working principle and control circuit of the turnout to analyze the corresponding fault causes and maintenance strategies in different modes.(3)Research on fault diagnosis algorithm:this paper classifies and statistically analyzes the fault sample as a data basis,and proposes a deep forest-based fault diagnosis algorithm based on its characteristics,and the algorithm is combined with theoretical analysis and grid search.The accuracy is close to 97%,and finally compared with various algorithms.Subdividing fault diagnosis and location for sub-class electrical faults under non-constructive circuit faults are carried out by combining the indicating voltage.(4)Design and validation of the fault diagnosis software:software system requirement is decided by communicating with domain experts from railway signaling sector,and a software platform including the online diagnosis and model training modules is developed by C#.The research results of the paper are high-precision turnout fault diagnosis algorithms and corresponding system software.The software is ideal in the laboratory test phase,which verifies the effectiveness of the proposed method.