Life Evaluation Of Mechanical Products Based On Stochastic Nonlinear Degradation And Its Application In Active Remanufacturing

As an important theory and technical support for intelligent recognition and operating maintenance,life prediction theory and methodology is not only the foundation of reliable service or health management of mechanical products,but also one of the major questions of remanufacturing science in green development.In the "Guidelines for the Medium-and Long-Term National Science and Technology Development Program(2006-2020)",life prediction technology for major products and equipment was listed as the state-of-the-art technology that needed to be developed.The "12th five-year plan" indicated health management technology as one of the key development directions.Therefore,it is of great scientific value and practical significance to develop life assessment theory and prognostic method of mechanical product which could guide the research of decision-making in remanufacturing engineering.With the increasing complexity,highly integration,intelligence and efficiency of mechanical equipment,life estimation for sophisticated mechanical products with high reliability and long lifetime is being more and more difficult to accomplish.Under such circumstances,life analyzing method based on performance degradation by modern monitoring technology becomes a new research direction which is worthy of development in the engineering field.To ensure the credibility of assessment and prediction,it is necessary to establish a degradation model with comprehensive consideration of performance degradation characteristics,and study the estimation/prediction algorithm with higher accuracy.Meanwhile,valid life prediction results play an important role in remanufacturing decisions,not only avoiding failures,but also helping to rationalize remanufacturing activities.In this paper,the mathematical model of stochastic nonlinear degradation process is proposed to describe the performance degradation of mechanical products.The real time iterative estimation method and the optimal timing for remanufacturing are studied based on performance degradation.The specific research content mainly includes the following aspects:].A life estimation model based on performance degradation with direct data is studied and established.Considering the fact that the performance degradation data of mechanical products can be directly monitored or captured,a general mathematical model of the performance degradation process is established by using the nonlinear Wiener process,which integrates the stochastic dynamic nonlinear characteristics of the complex operating environment.The real-time iterative estimation method of random parameters based on EM(Expectation Maximization)algorithm is studied,and the PDF of residual life distribution under first passage time is derived.Finally,a case study using crack growing data verifies the validity of the proposed life prediction method.2.A life estimation model based on performance degradation with indirect data is studied and established.According to the performance degradation state transition and its relationship with the monitoring data.a stochastic nonlinear degradation mathelatical model based on state space model is established.And life evaluation method based on stochastic filtering theory is studied.In the Bayesian framework,Sequential Monte Carlo is employed for researching joint iterative estimation method of degradation and parameters with conjugate distribution.Aiming at the existing questions in stochastic filtering method,the improved particle filtering method is explored to realize parameter updating and degradation estimation,and finally satisfy the requirements of higher precision for life prediction.A simulation study concerning the monitored tool acoustic emission data is used to demonstrate the validity of the proposed model and method.3.A life estimation model based on performance degradation with combination of direct and indirect data is investigated.Considering that it has certain limitation in describing the health state of mechanical products relying only on one kind of performance degradation,a combined degradation model is established utilizing two kind of degradation data and in view of the performance degradation characteristics.The Copula function is adopted to describe the correlation between the two kind of performance degradation,and the PDF of the residual life is analyzed and deducted.The parameter estimation method based on MCMC(Markov Chain Monte Carlo)sampling of Bayesian theory is studied,and the life prediction combined with two performance degradation is then approximated.Finally,the feasibility and validity of the proposed method is verified by analyzing the degradation data of motorized spindle.4.An active remanufacturing timing method based on performance degradation is studied.Considering the important scientific issues of mechanical products in green and safe operation and the lifetime assessment as a fundamental requirement for active remanufacturing,the necessity and value of remanufacturing timing based on performance degradation are analyzed.Afterwards,the life cycle environmental impact model and economic impact model based on performance degradation are respectively built using LCA and LCC method.The environmental impact and economic impact indexes are unified through social willingness to pay,and eventually the environmental-economic comprehensive impact evaluation model based on performance degradation is constructed,which realizes the application of performance degradation assessment in timing decision of remanufacturing.To illustrate feasibility and effectiveness of the proposed method,a real example of engine crankshaft is finally presented.