论文部分内容阅读
首先给出了一种通过对样本数据进行惩罚性划分,产生彼此之间具有差异性的GMDH学习器集合,然后利用遗传算法从已产生的GMDH个体集合中选择最优的个体进行选择性集成的算法,并将该方法应用于煤炭价格系统的实践研究.结果表明这种采用遗传算法选择性集成惩罚性GMDH个体的算法,与单个GMDH算法和采用遗传算法选择性集成随机性GMDH个体的算法相比,明显提高了模型的泛化能力和稳定性.该方法很好地给出了煤炭价格系统的模型,能够准确预测煤炭价格的变动趋势.
First, we give a GMDH learner set that has different characteristics by punishing the sample data punctually, and then use genetic algorithm to select the optimal individuals from the set of GMDH individuals that have been generated to selectively integrate Algorithm and the method is applied to the practice of coal price system.The results show that this method of selective integration of punitive GMDH individuals by genetic algorithm is compared with the single GMDH algorithm and the algorithm of selective integration of random GMDH individuals by genetic algorithm Which obviously improves the generalization ability and stability of the model.The method gives a good model of the coal price system and accurately predicts the change trend of the coal price.