An approach to in-process surface texture condition monitoring

Surface textures formed in the machining process have a great influence on parts’ mechanical behaviours. Normally, the surface textures are inspected by using the images of the machined and cleaned parts. In this paper, an in-process surface texture condition monitoring approach is proposed. Based on the grey-level co-occurrence matrices, some surface texture image features are extracted to describe the texture characteristics. On the basis of the empirical model decomposition, some sensitive features are also extracted from the vibration signal. The mapping relationship from texture characteristics to texture image features and vibration signal features is found. A back propagation neural network model is built when the signal features and the texture conditions are respectively inputs and outputs. The particle swarm optimization is used to optimise the weights and thresholds of the neural network. Experimental study verifies the approach's effectiveness in monitoring the surface texture conditions during the machining process. The approach's accuracy and robustness are also verified. Then, the surface texture condition can be monitored efficiently during the machining process.