Milling cutter wear monitoring based on whale algorithm optimized LSSVM

Abstract:

In order to solve the problem of monitoring the wear status of milling cutters, an improved whale algorithm is proposed to optimize the state recognition method of least squares support vector machine. Firstly, variational modal decomposition is used to process the vibration signal in the milling process, and the characteristics of the inherent modal functions obtained by decomposition is extracted; then, to tackle the problem that whale algorithm is easy to fall into local optimal solution and low convergence accuracy, a hybrid reverse learning algorithm and the nonlinear convergence factor is introduced, and benchmark functions are used to verify the effectiveness of the improved whale algorithm; finally, the improved whale algorithm optimized LSSVM model is applied to the simulation experiment of milling cutter wear status recognition. The experimental results show that, compared with particle swarm algorithm and traditional whale algorithm, the improved whale algorithm optimized LSSVM has higher recognition accuracy.