System Research On Tool Condition Recognition Based On Cutting Sound Signals Analysis During Machining

With the widely application of the numerical control machine tool and the rapid development of high speed cutting technology. Tool wear monitoring and diagnosis technology of tool life became one of the hot metal cutting research as the increase of cutting speed and processing efficiency. Tool as cutting objects that interact directly with the work-piece, the degree of wear of directly affecting the machining precision and surface quality of the work-piece. Real-time monitoring of tool wear State for ensuring the work-piece quality, increase productivity and utilization of tools is very important. This paper carried out the following research work:(1) Built a cutting tool condition monitoring system based on sound signal. Firstly discussed the characteristics and application of various cutting tool monitoring methods,then analysis-ed the research status of cutting tool monitoring system and development trend of cutting tool monitoring system, finally putted forward using cutting sound signal of tool condition monitoring, and combined with analysis of the cutting force signal for multi-sensor information fusion for tool condition monitoring.(2) Designed the overall program of tool monitoring system. Firstly chose the sensor and data acquisition card, selected date acquisition platform, and then designed the experiment project, collected the sound signal and the cutting force signal with cutting through the experiments, signal processing in time domain, frequency domain based on LabVIEW, finally analysis leads to the correlation of cutting sound signals and cutting force signal in the time domain, frequency domain with the characteristics of tool wear.(3) Extract the eigenvector associated with tool wear. Call Advanced signal processing toolkit in the LabVIEW for cutting sound signal multi-resolution analysis by means of wavelet frequency interval analysis, the sound signal was divided for 8 a band interval, observation and analysis of signal in time-frequency domain, and then extracted the energy of each frequency, finally used the percentage of signal frequency energy in thetotal energy as characteristic vector of tool condition monitoring.(4) Build tool State distinguishing model of multi-sensor information fusion. The synthesis of tool wear state-dependent cutting force signal eigenvector and cutting sound signal eigenvector by BP neural network based on multiple parameters information fusion,and achieved the implementation of tool condition monitoring.Through theoretical research and experimental analysis, established the correlation between state of cutting tool and cutting force signal and sound signal, realized non-contact measuring tool condition. Solves the problems of measurement of tool wear in complicated machining environment, offers a new way for the application of cutting tool monitoring method, and provides an effective and feasible method for the cutting tool state monitoring of actual production practice.