Reliability Evaluation For Grinding Process Of Rolling Bearing Based On Poor Information

With the development of mechanical manufacturing technology,the grinding process of rolling bearing is more and more complex and difficult to control,making bearing parts meet the higher technology requirements.There appear some problems with unknown probability distribution and small sample on reliability analysis.If the traditional statistical method still is applied to analyze the reliability on grinding process of rolling bearing,the result is not ideal.Therefore,in view of the reliability problem on grinding process of rolling bearing,based on poor information system theory,the quality parameters of rolling bearing are analyzed using grey bootstrap theory,maximum entropy method,fuzzy set theory,Poisson process and non-sorting grey relation in the paper,to realize the reliability evaluation for grinding process of rolling bearing.Main contents include the following 4 parts:1 Fusing membership function method,maximum membership degree method,rolling mean method,arithmetic mean value method and bootstrap method,a poor information fusion technology is proposed to adjust machining errors of machine.Fusing the output small sample data in machine trial processing using the fusion technology,the estimated true value of parts is obtained,so machining errors of machine are adjusted referring to the estimated true value.According to fuzzy set theory,the estimated interval of parts is predicted with small sample reliable data under the confidence level P=95%,to determine the reliability of the machine adjusted.Case studies show that the poor information fusion technology can be utilized to realize to adjust machining errors of machine,and the reliability of the machine adjusted is at least 95%.2 Aiming at the reliability on running state of grinding system of rolling bearing,a poor information assessment method is proposed to solve reliability assessment problems under the condition of small sample and unknown probability distribution.Based on small sample data in workpiece quality testing,extensive simulation data are generated utilizing grey bootstrap method,which can be processed to construct the probability density function of running state of grinding system using maximum entropy method.Setting the confidence level P,the confidence interval and the expanded uncertainty of running state of grinding system are acquired.Through Poisson counting process,the variation intensity of running state of grinding system iscomputed with the output quality data in grinding process.The reliability function on running state of grinding system is obtained using Poisson process with zero-failure probability.Case studies indicate that the confidence level P is 95%,which is the best choice for running state of grinding system to achieve a good running.The results can provide a theoretical basis to reasonably adjust machine.3 Based on reliability research on running state of grinding system,the reliability on running state with evolution of grinding system is conducted to evaluate in real time using the poor information assessment method.Getting multiple groups of test data sequences,the reliability curve graphs of running state with evolution of grinding system are drawn,which can predict the reliability of running state with evolution of grinding system in real time under the best confidence level P=95%.Case studies indicate that,if the quality achieving reliability degree of inspection data is r>0.65,the area intersection of the probability density function of reliability on inspection data and intrinsic data is A>0.3,and the variation probability of inspection data is PB?0.7,the running state with evolution of grinding system is reliable,which has no mutation;otherwise,it is not reliable,which has occurred mutations.The results can provide a theoretical basis to determine whether grinding process is reliable or not.The sensitivity analysis shows that the size of samples does not affect the evaluation results.4 The variation degree on grinding process of rolling bearing is implemented to evaluate based on non-sorting grey relation and fuzzy set theory.For the time data sequences in order,setting the grey confidence level P=95%,the grey attribute weights of them are computed using non-sorting grey relation,to construct the similarity matrix of grey attribute weights of them.The equivalence matrix of grey attribute weights of the time data sequences is solved using fuzzy transitive closure method.Through hypothetical test and piecewise average equivalence coefficient,the variation process of systemic error of grinder is described to realize evaluation for the variation degree on grinding process.The results are in accordance with the actual situations.