Model And Application Study For The Reliability Of Tractor Based On Bootstrap And Mixed Weibull Distribution

In recent years, with the rural land circulation policy and agricultural machinery purchase subsidy policy being intensified,large scale,centralized and modernized production management mode promotes the agricultural machinery demand significantly.Because of the large and medium horsepower tractors are the main power source of agricultural machinery,the domestic tractor manufacturing enterprises have increased the production efforts.Reliability is the core of product quality assurance,however the reliability of domestic tractor is much lower than the foreign machinery in developed countries.The research results and the related literature are few,and the theory and methods for the reliability evaluation are not mature.Combined with tractor field test failure data,the tractor reliability evaluation model and application analysis are considered in this paper based on the reliability engineering,stochastic processes,statistical analysis of parameter and non-parameter estimation theory,fuzzy mathematics and failure mode effect and importance analysis theory.The failure process of the mechanical system is different from the electronic products,thus the reliability theory and methods of electronics can not be applied directly to the tractors.The tractor is a complex mechanical system, and its structural complexity and operating environment particularity determine that the failure mechanisms are diverse.The failure rate is the bathtub curve.The parameters of the Weibull distribution can describe the early failure period, accidental failure and fatigue failure of the bathtub curve completely,so it is suitable to evaluate the tractor reliability.Single Weibull distribution which ignores the diversity of the mechanism describing the failure process of complex mechanical system would produce more considerable error.While the applying foundation of the mixed Weibull model is the classification according to different failure mechanism,which makes up the shortage of the single model,so the mixed model is suitable for describing failure process of tractor.As it was difficult to obtain the complete failure information,the relation matrix between the failure-pattern and failure stress was established to avoid the failure mechanism analysis.The failure data was classified by Fuzzy clustering based on the similarity of failure stress,and the parameters of the mixed Weibull model were estimated based on the Least Squares Method.Combining the failure data of the tractor by tracing test,the reliability of mechanical system is analyzed by using the mixed Weibull model.By comprising the reliability curve of mixed Weibull model with the single Weibull model,it is more reasonable in the mixed model for the evaluationof tractor reliability.The result of research can be used as references for the improvement of reliability and maintainability of tractor,and as theoretical basis for study on the interrelation between mechanical system failures as well.The parameter method is based on the assumption,and the the statistical result is quite ideal when the deviation between the theory and the reality is not differ much.On the contrary,if the assumption is not established, that is, the difference is obvious between the assumptions and the reality, the conclusion have no significance.The non-parameter method is not assumed the specific distribution,so it is universal,stable and not collapsing.In order to improve the computational efficiency and extract more sample information,the non-parameter method is often used in combination with the bootstrap.On the basis of the existing sample data, the bootstrap method can enlarge the sample size, which makes the non-parameter estimation more scientific and realistic.The mathematical model between the failure intensity and the kernel function can be established.The failure intensity can be estimated by non-parameter method based on the bootstrap.Combined with the failure data collecting in the tracing experiment,the initial stages failure intensity curve was mapped,revealing the failure rules in the initial stages.Comparing the MSE of cumulative intensity and cumulative number of failures of the two estimations,the curve which was estimated by non-parameter estimation method was more realistic to the actual experiment.According to the tractor function,the tractor system was divided into five subsystems,which were power system, chassis system, hydraulic system, electrical system and other working equipment.Failure modes and effects analysis(FMEA) and system importance analysis of tractor were carried out,and the mathematical model for the system importance analysis was established to ascertain the main failure position, failure mode, failure cause and the key subsystem of the tractor.According to the results revealed by the parameter and non-parameter estimation,and FMEA and system importance analysis,methods and suggestions for improving the reliability of tractor were proposed in the aspects of design, manufacture,use, maintenance and so on.The estimation methods of tractor failure regulation which kept the failure mechanism and the failure data consistent were discussed in this paper,and the main mode and key subsystem of failure were confirmed from quantitative view.The research results perfect and enrich the theory and methods of evaluation for tractor reliability,improve the effectiveness and scientificalness of the evaluation results, make up the defects of the traditional evaluation model in the reliability analysis of tractor, promote the evaluation results in the application of reliability evaluation, and provide references for the tractor production, use and maintenance decision.