Thermal error modeling of machine tool spindle based on particle swarm optimization and neural network

To overcome the lower accuracy, poor convergence and worse generality of the thermal error modeling method based on BP neural network, the fuzzy cluster theory was used to group temperature variables and the correlation analysis was conducted to select thermal senstive points to mine the correlation between temperature variables and thermal errors and to reduce the coupling among temperature variables. Besides, the reciprocal of the square sum of the residual errors was regarded as the individual fitness function in the PSO algorithm. Moreover, the performance code of the head of the individual was regarded as the number of the nodes in the hidden layer and the body of the individual was mapped to the weights and thresholds of the BP network. The extreme value and global extreme value of the individual were tracked to update the speed and position of the individuals. Then, the five-point method was utilized to measure the spindle thermal errors of a jig-boring. After that, the elongation and thermal tilt angle models were established based on the BP and PSO-BP models. The results show that the thermal error model based on PSO-BP model can predict thermal errors of the machine tool spindle under different cutting conditions and the results validate the effectiveness of the measuring and modeling methods.