Optimization design of jigs used in biomass mould pressing based on BP network and NSGA

One important way to increase biomass utility value is biomass mould pressing, however there are problems such as deformation or strength failure of the plate used in biomass mould pressing jigs. Firstly, L₁₆ (4⁵) orthogonal table and SolidWorks/simulation are applied to analyze the plate model and responses to loading and structure parameters of the plate which includes the maximum stress, the maximum deformation and the mass of the plate are got. Secondly, a three-layer BP neural network model with 4 input nodes, 12 hidden nodes and 3 output node is set up. Test input data is used to simulate the BP network model and regression analysis shows that items such as m, b and r corresponding to three targets have maximum relative error 0.3% compared to ideal values. At last, a three-target fitness function which is to seek the maximum stress, the maximum deformation and the minimum mass of the plate is built based on the BP network. Fast-nondominated-sorting genetic algorithm (NSGA-II) is used to optimize design variables. 19 groups of Pareto solutions are got, which provides choices for rational plans of the plate´s loading and structure parameters. The study indicates that the method applied in this paper is able to optimize the plate and therefore a feasible method is provided for optimization design of mechanical structure.