Research On Health Monitoring And Maintenance Decisionmaking Method For Multi-state Manufacturing System

Manufacturing system health monitoring and maintenance decision-making are important methods to ensure production process stability,product quality consistency,and system reliability.Health monitoring can monitor the state of the system in real-time and detect system abnormalities to prevent the occurrence of a large number of nonconforming products or major equipment failures.Maintenance decision-making improves the performance and reliability of the system through reasonable maintenance methods and strategies to ensure the performance of the production process and meet the product quality requirements.They complement each other and jointly ensure high quality,high efficiency,high stability,and low-cost operation of the production process.In this dissertation,System state monitoring includes two common methods: using SPC to monitor the production process states and using pattern recognition to monitor the states of equipment(including tools).The process states are assessed by the quality of products produced by the system,and the equipment states are monitored by analyzing sensor signals.In the existing health monitoring and maintenance decision-making models,most of them are aiming at single-component manufacturing system that only health state and failure state are considered.However,the manufacturing system is often composed of multiple MSC and each MSC usually has one or more transition stages,namely sub-health state,from health state to complete failure.To solve the health monitoring and maintenance decision-making problems of this multi-state manufacturing system consisting of multicomponent or having multi-level states,two improved health monitoring methods including production process state monitoring and equipment state identification are proposed firstly.And then maintenance decision-making methods for multi-state manufacturing systems from single-station to multi-station are proposed based on health monitoring.The main research contents and results of this dissertation are as follows:Taking the product quality characteristics of the multi-state manufacturing system as the monitoring object,a production process state monitoring method based on improved SPC is proposed.Firstly,a Vp control chart monitoring strategy is designed based on the concept of delayed failure and the generalized renewal process theory.Then,adaptive sampling rules and non-periodic monitoring methods are designed to improve monitoring performance.To sovle this problem,an economic optimization model is designed under the constraint of statistical characteristics and solved by PSO algorithm.Finally,rationality,economy,flexibility,and other characteristics of the method are demonstrated by the illustrative example,comparison analysis,and sensitivity analysis.Take the equipment of the multi-state manufacturing system as the monitoring object,a method of equipment health state recognition based on the HGHMM is studied.Firstly,a simplified feature selection method based on Pearson correlation analysis and redundancy analysis is proposed.Then,the superior performance of HGHMM on sequential data learning is used to model the equipment state monitoring.Meanwhile,to avoid the BW algorithm falling into local optimal,an improved PSO algorithm is introduced to optimize the initial parameters of the BW algorithm.Finally,the effectiveness of the method is verified through the published tool wear and hydraulic unit data sets.The experimental results and comparative analysis show that the proposed model has very high recognition accuracy.Based on the two types of health monitoring method above,a maintenance decisionmaking model based on health monitoring for single-station multi-state manufacturing system is proposed to provide adaptive solutions for multistage manufacturing systems.Firstly,the stability factors that are difficult to monitor in the system health monitoring are considered and the impact of the four maintenance methods on the system reliability and cost are deeply analyzed.Then,an integrated optimization model combining non-periodic monitoring and non-periodic maintenance is designed taking economy as the optimization goal.Finally,the economy,flexibility,and broad applicability of the model are demonstrated by the illustrative example and comparison analysis.A maintenance decision-making model for the multi-station multi-state manufacturing system based on health monitoring is proposed to provide a solution to the maintenance decision problem of serial multi-state manufacturing system.Firstly,the reliability,the nonlinear rate of nonconforming product rate,and the state transition relation under the hybrid failure mechanism are analyzed.Then the maintenance strategy in various situations is analyzed.Finally,an optimization model is designed according to the principle of economy and solved by the PSO algorithm combined with Monte Carlo simulation.Compared with the maintenance model for single-station,the comparison analysis shows that the multi-station optimization model has a better overall economy.Based on the methods researched above,a prototype system of health monitoring and maintenance decision-making for the multi-state manufacturing system is designed and developed.And the methods proposed in this dissertation are validated by taking the expanding process of an air conditioning production workshop as an application case.