Research On Improved DS Evidence Theory Fusion Algorithm In Machine Tool Condition Monitoring

Machine condition monitoring can ensure the stability and safety of the machining process.In traditional machine condition monitoring,a single sensor is easily affected by the complexity of machine equipment and the instability of the operating environment.The acquired information is usually accompanied by uncertainty and imprecision,and even makes the system appear error judgment.Using multi-sensor fusion technology to monitor the status of machine tools can effectively solve this problem,but how to effectively fuse these data has become the key of machine tool status monitoring research.D-S evidence theory in data fusion technology has superior ability in the representation and reasoning of uncertain information.Therefore,based on the analysis of the basic principle of data fusion and typical data fusion algorithm,this paper focuses on the problems of D-S evidence theory in the fusion processing of uncertain data and conflict data,aiming at the problems of data in machine tools.This paper proposes two improved methods for the uncertainty and conflict of evidence theory,and uses Gauss membership function as the basic probability distribution function in machine tool condition monitoring.The first is to improve the uncertainty data.By introducing the concepts of Z-number and fuzzy analytic hierarchy process,using the Z-number model from the perspective of evidence theory,combining information constraints and reliability,using the idea of analytic hierarchy process to transform the model into a classical fuzzy number,then using the Z-number after fusion of fusion rules to get the results,modeling and processing the uncertainty in sensor data fusion reasonably.Simulation results show that the method can get more accurate results,can reasonably express the uncertainty in the evidence,and make the results more effective.The second method is to improve the conflict data.Firstly,in order to reduce the interference in the process of data collection of multiple sensors,the reliability of sensors is obtained by batch estimation,and the obtained evidence is corrected for the first time;secondly,the correction coefficient of evidence is obtained by calculating the deviation degree of evidence body,and the obtained evidence is corrected for the second time;finally,the modified evidence is fused by using evidence theory to complete the decision and In the process of information processing,the simulation of the two kinds of data shows that the proposed algorithm can reasonably fuse the conflict data,reduce the impact of abnormal data on the fusion results,and also avoid the Zero paradox problem,with good robustness.At last,two new algorithms are tested in machine condition monitoring.The results show that the two improved algorithms can get accurate fusion results and get the same machine condition as the actual one,which can improve the accuracy and robustness of condition monitoring,and effectively solve the uncertainty and conflict of evidence.