A novel methodology on the state estimation of the machine tools based on the multi-sensor information fusion in time-frequency space

Abstract

The tendency of the current Computer Numerical Control (CNC) machine tools is high speed, high precision, high reliability, intelligence, integration and openness. However, there exists contradiction between high speed and high precision. The machine tool system is so complicated that its model is difficult to be constructed, simultaneously, in order to keep precision, the model will be very complex, which will lead to the decline of the speed. The paper provides the new methodology integrating the neural network and wavelet in the time-frequency space, which make only use of the signals from systems not of the model. For this new approach, no prior knowledge is required about the statistics of sensor signal, or the behavior of systems. Therefore it can realize the on-line state estimation of machine tool. It can guarantee the accuracy, at the same time, enhance the speed. Moreover, through the simulation integrated navigation system, it was proved that the new method could effectively enhance the positioning accuracy of the integrated system. Consequently, the new methodology could extend the information fusion to CNC machine tools and other sophisticated systems.