Analysis And Application Of Compound Alarms For Key Components Of Flexible Cable Hoisting Robots

With the advent of the Industry 4.0,high-speed rail industry in China has already entered the intellectualized era,and thus the production efficiency of high-speed rail has also greatly increased.In a complex production environment,the devices on the production line still have faults and alarms,and sometimes multiple alarms occur in a very short time,which is defined as compound alarms.Multiple alarms make it difficult to locate problems and thus seriously harm the production efficiency.Alarm correlation analysis is a common method to solve this problem.This method has made great achievements in network alarms,but there is still many research gaps of alarm correlation analysis in the research on of production lines of high-speed rail.The cable hoisting robot is one of the most important components in the high-speed rail production line.Therefore,this thesis investigates the compound alarms of the cable lifting robot,and uses the findings to help engineers quickly locate problems,which is of great practical significance.This thesis makes deep study on related theories,such as various types of alarm correlation analysis including rule-based alarm correlation analysis,case-based reasoning,and model-based reasoning alarm correlation analysis,and the discretization methods of time series data,including the sliding window discretization and variable sliding window.Based on the existing experimental conditions and characteristic analysis result of alarm data of the key components of cable hoisting robots,this thesis uses data mining to analyze the relevance of the compound alarms of the key components of cable hoisting robots.Firstly,based on the feature of practical alarm data of the critical components of cable hoisting robots-long duration,this thesis modifies the sliding window discretization and proposes a discretization method of variable window with delay parameter to improve the accuracy of data discretization.After the data processing by the discretization method of variable window with delay parameter,the transaction data with sequential constraints is generated.Secondly,since the original FP-Growth algorithm is not suitable for the mining of transaction data with sequential constraints,this thesis modifies the FP-Growth algorithm and proposes OFP-Growth algorithm.Thirdly,because of the subjective and inaccurate problems caused by the artificial setting of the support threshold and the confidence threshold of the FP-Growth algorithm and the OFP-Growth,this thesis optimize the OFP-Growth algorithm by using glowworm swarm optimization algorithm(GSO)and proposes the GSO-OFP-Growth algorithm.Finally,after the verification of effectiveness of these methods,this thesis created a compound alarm correlation analysis system for the key components of cable hoisting robots and achieves the combination of the algorithm and industrial production.