Research On Grinding Monitoring Of Difficult-To-machine Alloy Based On Acoustic Emission And Multi-information Fusion

Difficult-to-machine alloy such as GH4169,TC4 and SiCp/Al have many excellent properties,such as high strength,high hardness,high wear resistance,high specific strength and high temperature resistance.However,these excellent characteristics increase the difficulty of machining,and the grinding process is an extremely complicated process.There are always some unpredictable and uncontrollable influence factors,so that the surface quality of the workpiece during the grinding process is not easy to controll.Therefore,it is urgent to study a method which can accurately predict the surface roughness of workpieces,and provide a reliable prediction model for on-line monitoring of surface roughness.In addition,the high hardness and high wear resistance of the difficult-to-machine alloy make the wear of the grinding wheel become faster during the grinding process,which is likely to lead to the untimely dressing or over dressing.Therefore,it is very important to accurately identify the wear degree of grinding wheel in the grinding process of difficult-to-machine alloy.The main research contents of this paper include:(1)Through the research on the grinding process of difficult-to-machine alloy based on acoustic emission monitoring,the effects of grinding parameters,workpiece material and grinding wheel type on the characteristic parameters and spectral characteristics of acoustic emission signals are obtained.And the reasons for this effect are analyzed.In addition,the corresponding relationship between the surface roughness of the workpiece and the parameters of the acoustic emission signal is obtained.By analyzing the principle of BP neural network algorithm and the specific steps of network training process,a multi-information fusion prediction model based on BP neural network is established.(2)Through the dressing experiment of the grinding wheel and the grinding wheel wear experiment based on acoustic emission monitoring,it is analyzed whether the dressed grinding wheel meets the requirements of the wear experiment.At the same time,the waveform and characteristic parameters in time domain and the signal characteristics in frequency domain of the corresponding acoustic emission signals during grinding of different wear states are analyzed.Besides,the grinding force and grinding force ratio during grinding of different wear states are also analyzed.The wear state of the grinding wheel during the experiment is comprehensively judged.The wavelet packet decomposition of AE signal is used to extract the characteristic frequency band information that has great correlation with the wear state of the grinding wheel,and the non-linear correspondence between AE signal and the wear state of grinding wheel is obtained.Finally,a multi-information fusion identification model of grinding wheel wear state based on BP neural network is established.(3)A monitoring system for grinding wheel wear status was developed by using Lab VIEW and MATLAB,including the development of front panel and back panel of online monitoring interface and offline analysis interface.The function modules of the monitoring system mainly include the signal acquisition module,the signal analysis and processing module,the grinding wheel wear status recognition module and the graph display and data storage module,etc.The usability of the software system is verified by the actual monitoring experiment.