联合收获机惯性分离室工艺参数的优化——基于改进遗传神经网络

Optimization of Process Parameters in the Inertial Separation Chamber of Combine Harvester——Based on Improved Genetic Neural Network

针对联合收获机惯性分离室工艺参数最优组合问题,提出了一种改进的遗传神经网络优化方法。该方法利用区间自适应遗传算法自适应动态调整区间的特点,彻底取代BP神经网络中'误差逆传播'过程,将区间自适应遗传算法和神经网络的优势有机地结合起来,有效解决BP困境。优化实验结果表明:该方法拟合精度明显优于二次回归的拟合精度,更接近联合收获机惯性分离室参数的实际情况,为农业机械中通过实验设计方式解决的'黑箱'优化问题提供了一种新的方法。

To the optimal combination of process parameters of the inertial separation chamber of the combine harvester,an improved genetic neural network optimization method was proposed,which completely replace the'error back-propagation'process of BP neural network algorithm by using the characteristics of adaptive dynamic adjustment interval about interval adaptive genetic algorithm.The advantages of interval adaptive genetic algorithm was combined with neural network,and the dilemma of BP algorithm was effectively solved.The results show that the fitting accuracy of this method is obviously better than that of quadratic regression,which is closer to the actual parameter of the inertial separation chamber of the combine harvester.We provides a new way for the'black box'being solved by experimental design in agricultural machinery.