Multi-objective optimization of sawblade multi-spot pressure tensioning process based on backpropagation neural network and genetic algorithm

Resources scarcity has created strict requirements for efficient wood processing. Multi-spot pressure tensioning can effectively stabilize the sawblade and reduce cutting losses, but it is difficult to apply in practice due to the complexity of the process. In this study, the elastic-plastic solid model of multi-spot pressure tensioning was established, which was simplified to a thermal expansion shell model, and the mapping relationship between the two models was determined. The feasibility of the mapping method was verified by experiments. The backpropagation neural network (BPNN) was trained on a database composed of 8160 working conditions and the relative error was found to be less than 5%. Indenter displacement, pressing point quantity, and indenter radius were shown to change the degree of tension to varying extent. The pressing-point-distribution radius determines the development direction of sawblade performance. Increasing the number of pressing point circles can expand the adjustment range for sawblade performance. Both sawblade performance and tensioning energy consumption can be optimized by the genetic algorithm (GA). The optimal process parameters for different applications can be obtained. The combination of finite element method, BPNN, and GA can effectively optimize the multi-spot pressure tensioning process to improve the sawblade performance.