Identification Of Blunting State Of Grinding Wheel And Study On Characteristic Parameters Of Acoustic Emission

Grinding is an important metal processing method in modern industry.The grinding wheel is a tool for grinding processing.The quality of the surface of the grinding wheel affects the processing efficiency and the quality of the workpiece.During the grinding process,the blunting phenomenon of the grinding wheel will occur,and the abrasive grains on the surface will deform,break and fall off,which needs to be detected and repaired in time.Traditional methods need to rely on worker experience to observe the status of the grinding wheel or processing parameters,and it is difficult to make a timely and accurate judgment.Industrial intelligent development has put forward higher requirements for the detection of the blunting status of the grinding wheel,and provide more ideas.The plastic deformation and abrasive wear of the workpiece will produce acoustic emission(AE)signals,and the AE signal can be used as a basis for judging the blunting status of the grinding wheel,and has obvious characteristics and is not easy to be disturbed by noise.In this paper,the physical mechanism of the grinding process is first studied.On this basis,a method for detecting acoustic blunting of grinding wheels based on variational mode decomposition(VMD)and probabilistic neural network(PNN)is proposed.Based on the improvement of the EMD decomposition method,VMD can adaptively decompose the original signal into multiple intrinsic mode function(IMF)components according to the characteristics of the signal,and filter out the components with larger kurtosis to reconstruct the AE signal.The key to AE detection is the selection of characteristic parameters.AE signals have multiple characteristic parameters,which characterize the AE signal from multiple aspects.Based on the related research,this paper proposes the AE ratio of envelope energy and selects a total of 5 characteristic parameters.A five-dimensional characteristic vector dataset is constructed and processed by artificial neural network.The PNN network has a good processing ability for nonlinear and non-stationary signals.The training set after frame splitting is input into PNN for training,then determine network model parameters and modify them.The intelligent identification system of the blunting state of the grinding wheel is obtained.In this paper,combining theory with practice,an AE detection experimental system was built to verify the recognition ability of the proposed method,and a test set independently and equally distributed with the training set as well as a continuous signal data set of other machining workpieces was constructed.The comprehensive test recognition accuracy reached 94.5%,which can clearly distinguish different grinding wheel blunting states.The method proposed in this paper has successfully realized the real-time accurate identification of the blunting state of the grinding wheel and the early warning of the serious blunting state.In addition,this paper also compares the accuracy of the AE signal with different characteristic parameters to identify the blunting state of the grinding wheel,and probes into the corresponding relationship between the characteristic parameters of different types and the blunting causes of the grinding wheel.Characteristic parameters such as the effective level of AE signal and the power or energy ratio between the AE signal and the original signal can best reflect the degree of grinding wheel blunting.The blunting process of grinding wheel is divided into several stages.Due to the difference of the main blunting reasons,the recognition accuracy of the characteristic parameters of specific types also varies in different blunting stages.As a reference for the reasonable selection of AE characteristic parameters,the conclusion has potential application value.