Failure Modes Risk Assessment And Reduction Strategy For Multi-Stage Manufacturing Systems

The manufacturing industry is a vital cornerstone of a country,playing a significant role in its economic foundation.As a crucial carrier of China’s intelligent manufacturing strategy,manufacturing systems bear a considerable responsibility closely tied to the nation’s and its people’s well-being.The risk condition(e.g.,severity,occurrence frequency,and detectability of failure modes)of manufacturing systems has become a critical factor affecting product quality.However,in current manufacturing practices,manufacturing systems often exhibit characteristics such as the multi-stage manufacturing process,the complexity of the manufacturing process,the customization of customer demands,a large number of uncertain factors,and an easily disrupted supply chain network.This leads to the manufacturing system displaying various types of data,e.g.,historical and online data for process variables,response variable data,and the limited knowledge and experience of decision-makers.In this context,the challenge lies in effectively assessing the risks of manufacturing systems using various types of data.Subsequently,it is also challenging for decision-makers to formulate risk reduction strategies.To address the aforementioned challenges,decision-makers need to develop a series of risk assessment methods and risk reduction strategies.Through focusing on critical processes of product manufacturing,i.e.,trial production,formal production,product inspection,and supply chain management,this research aims to implement failure modes risk assessment and risk reduction strategy for manufacturing systems.Firstly,it is difficult to obtain the data associated with risk factors.Also,it is impractical to rely solely on the knowledge and experience of decision-makers for assessing the potential risks(such as improper settings of process variables,production technology defects,and raw material deficiencies)in manufacturing systems.Secondly,conventional methods cannot accurately characterize the evolution mechanism of potential risks due to the overlook of the structural forms of multi-stage manufacturing.Thirdly,during the manufacturing process,potential risks inevitably lead to defective products.Meanwhile,due to the numerous uncertainties,such as uncertain customer demands and limited knowledge and experience of decision-makers,traditional product inspection strategies that consider two types of inspection errors are challenging to apply.Finally,it is important to address the management of unforeseen disruptive events or risks(e.g.,major manufacturing equipment failures,significant raw material shortages,and severe product quality defects)in manufacturing systems.This can prompt considerable impacts on customer-oriented supply chain networks.To cope with the aforementioned challenges,it is imperative to conduct research on failure modes risk assessment and reduction strategy for manufacturing systems.To mitigate potential risks and interruptions in manufacturing systems,risk reduction strategies include product inspection and optimal post-disruption restoration strategies.The primary research contributions and innovative outcomes are summarized as follows:(1)To address the issue of limited knowledge and experience of decision-makers in the trial production process,a FMEA method based on the three-way decision framework is proposed.A customized epidemic model is employed to characterize the new product introduction process during the trial production,obtaining the losses(e.g.,the severity)incurred by alternative production actions in the manufacturing system.A finite Gaussian mixture model is utilized to determine whether a new failure mode occurs in the manufacturing system at arbitrary running time.Subsequently,the detectability and occurrence frequency of new failure modes can be calculated.A dynamic Choquet integral is proposed to fuse assessment data of risk factors to seek the best production action.The results demonstrate that the proposed method behaves with stable performance in identifying the best production action.(2)To cope with the issue of overlooking potential risk evolution mechanisms in different structural forms during the risk assessment of multi-stage manufacturing systems,the data envelopment analysis(DEA)-based FMEA framework is proposed.Given the characteristics of multi-stage manufacturing processes and various structural forms in formal production processes,DEA is adopted to characterize the structural forms(e.g.,serial,parallel,and network structures)of the manufacturing system and modeling the structural correlations between potential risks and induced factors(e.g.,cost of losses and recovery time).Additionally,the proposed framework utilizes the Me measure to handle imprecise parameters related to risk factors.The results indicate that the proposed framework possesses two important properties,i.e.,optimism and prominent risk awareness.Furthermore,the proposed framework exhibits stable performance in determining the risk priority of failure modes.(3)To deal with the challenge of applying traditional product inspection strategies that consider two types of inspection errors,a robust possibilistic programming-based three-way decision method is proposed.By utilizing robust possibilistic programming,the product inspection strategy design is regarded as a process that depends on the upstream manufacturing processes and downstream customer demands.Through the three-way decision,an intermediate/uncertain decision is introduced to simultaneously reduce the errors of erroneously rejecting non-defective products and erroneously accepting defective products.Grounded on the three-way decision,a robust possibilistic programming-based three-way decision is proposed.With the aid of the proposed approach,decision-makers can adopt a flexible attitude to handle imprecise parameters and chance-constrained constraints,ensuring the robustness of the product inspection strategy.The results demonstrate that the proposed approach effectively avoids overly conservative attitudes when designing product inspection strategies.Furthermore,the proposed approach not only performs better in meeting customer requirements but also has significant advantages in terms of product inspection costs and computation time.(4)To address the challenge of supply chain networks being susceptible to disruptive events,along with the difficulty in recovery,a generalized three-way decision enhanced two-stage stochastic programming model is proposed.This research explores the correlations among resilience improvement of the supply chain network,risk aversion,and enhanced risk resistance.Under the scenarios relevant to the occurrence and nonoccurrence of disruptive events,the research systematically investigates the design of optimal post-disruption recovery strategies.To ensure the high level of resilience of the supply chain network at the source,an improved ELECTRE-TRI method is introduced.A rapid recovery strategy is designed for the main manufacturer,effectively addressing uncertainties related to customer demand.Furthermore,a generalized three-way decision is proposed to enhance the risk aversion of the traditional two-stage stochastic programming.The results show that the proposed method has significant advantages in resilience enhancement,risk aversion,and enhanced risk resistance.Moreover,compared to traditional risk metric methods(such as Conditional Value-at-Risk),the proposed method has shorter computation times,greater sensitivity to resilience improvement,and better interpretability.