扫雷犁, 递阶遗传算法, 神经网络, 建模

 A method for designing a radial basis function neural network was proposed based on the hierarchical genetic algorithm, and the technique was employed to settle the modelling problem of the electrohydraulic system of a certain mine sweeping plough with complex nonlinear characteristics. In the proposed hierarchical genetic algorithm, the control genes with binary code were used to identify the structure of neural networks, i.e. the number of hidden units, and the parameter genes with real code were employed to identify the parameters of neural networks, i.e. the center parameters of hidden units. In order to have a better generalization performance and simpler network structure, the Akaike's information criterion was adopted to evaluate the fitness of individual chromosomes. The experimental results and comparisons with other modeling methods indicate that the proposed approach can be more effectively to construct precise models for the mine sweeping plough.