Global Sensitivity Analysis and Dynamic Optimization of Multi-motor Driving Transmission System

The multi-motor driving transmission system (MMDTS) is a nonlinear, complex large-scale mechanical system that includes the drive subsystem (DS) and the gear transmission subsystem (GTS); the GTS is one of the most important units that can transmit motion and power from the DS to the working machine. A dynamic optimization approach is proposed to improve the dynamic performance of the MMDTS by improving its vibration displacement root mean square (VDRMS) value in this work. To avoid using the time-consuming dynamic model of the MMDTS in the optimization process, a surrogate model that imitates the behavior of the MMDTS is used as a replacement. Candidate surrogate models are constructed by the kriging, radial basis function, and support vector machine methods by employing sample points obtained by the Max-Min distance Latin hypercube sampling method and the original dynamic model of the MMDTS. The adjusted error-squared (R2 adjusted) is used to validate the accuracy of these surrogate models, and the kriging surrogate model is consequently selected as the final surrogate model for the subsequent Sobol’ global sensitivity analysis and particle swarm optimization (PSO). The most important design parameters are selected using the Sobol’ global sensitivity analysis, and they are set as the optimization variables in order to reduce the problem dimension and improve the computational efficiency. Subsequently, the weighted-combination and non-domination sorting methods are used to obtain the optimal parameter solution, and the VDRMS values of the system that are obtained before and after optimization are compared. The optimization results indicate that the proposed dynamic optimization approach not only can significantly improve the computational efficiency, but also can be successfully applied to the parameter optimization design of the MMDTS in order to improve its dynamic performance.