Research On Health Status Assessment Of Chain Tool Magazine Based On Multi-sensor Data Fusion

With the rapid development of intelligent manufacturing technology,the demand for CNC machine tools in the equipment manufacturing industry is rising.China has become the world’s largest consumer market for machine tools,but domestic high-end machine tool products have failed to form a strong competitive advantage in the market due to poor performance stability and low reliability level.Chain tool magazine is one of the key functional components of machining center,and its frequent failure seriously affects the reliability level of the whole machine of machining center.Health status assessment can identify the current operating condition of the equipment and help with subsequent work such as predictive maintenance,which is a key technical means to reduce the equipment failure rate.In order to achieve accurate and situational maintenance and health management,this paper conducts a study on the health status assessment of domestic chain tool magazines.The main contents of the paper are as follows:(1)A failure mode,impact and hazard analysis method considering maintenance cost is proposed for the problem that the structure principle of chain tool magazine is complicated and the typical failure mode is difficult to determine.The structure function and tool change process of chain tool magazine are analyzed in detail,and the subsystem division of chain tool magazine is carried out.The maintenance cost,severity and frequency of occurrence are selected as evaluation indexes,and the risk of failure mode is measured by the number of possible risks,and the FMECA table of the chain tool magazine is established.Seven typical failure modes of the chain tool magazine are determined to provide a basis for the construction of the health status monitoring system.(2)The hardware construction and software development of the health status monitoring system of the chain tool magazine were carried out because of the problems of unclear status monitoring indexes and cumbersome test operation of the chain tool magazine.Typical failure modes of the chain tool magazine were analyzed,and the monitoring indexes and monitoring scheme of the test were determined.Based on the Lab VIEW programming platform,the software for monitoring the health status of the chain magazine was designed and developed,integrating various functions such as test operation control,acquisition signal display and data storage.On this basis,the reliability bench test of the chain tool magazine was carried out.(3)A health state assessment method that considers the influence of load is proposed for the problem that changes in the type of robot exchange tool can have an impact on the vibration signal.The vibration signals of the manipulator are decomposed using a complementary aggregated empirical modal decomposition method,and the Euclidean distance and multi-scale alignment entropy differences between the eigenmodal functions under different loads are compared,and the signal components that are more influenced by the load changes are excluded.A probabilistic neural network method was used to construct the health state assessment model,and the improved gray wolf optimization algorithm was used to determine the main parameters of the model,and the application was verified.By comparing the evaluation results of different methods,it is verified that the proposed method has a better evaluation effect.(4)An improved D-S evidence-theoretic approach considering conflicting and important evidence information is proposed for the problem of different degradation rates of each component of the chain tool magazine and inconsistent multi-sensor evaluation results.Time-domain features and energy features of single-sensor data were extracted,and feature screening was performed using a maximum correlation minimum redundancy algorithm.Evidence information of D-S evidence theory was obtained along the robotic health state assessment model to obtain the single-sensor assessment results and normalize the summation layer output.The Jousselme distance and information entropy were used to measure the conflicting and importance of the evidence information,respectively,and the combined weight coefficients were obtained to correct the evidence information,and then the fusion was performed according to the synthesis rules.A case study was conducted with two cases of consistent and conflicting assessment results,and the proposed method was verified to have a better fusion assessment effect by comparing the assessment results of different fusion algorithms.