【摘要】 农业机械总动力是衡量农业机械化水平的一项重要指标。对河南省农业机械总动力进行预测,将为农业机械化部门制定合理的发展规划提供一定的理论支持,同时也为农机企业了解未来市场需求状况提供一定的参考。以1991-2010年间的河南省农业机械总动力统计数据为基础,利用BP神经网络建立了河南省农业机械总动力的预测模型。该模型采用3层BP神经网络,输入层、隐含层和输出层的神经元数目分别为5,13和1。隐含层和输出层的激励函数分别为正切型与对数型Sigmoid函数。采用分步预测的思想,利用自适应学习速率训练方法对该网络进行了训练,获得了该模型中各层之间的连接权值和各层神经元的阈值。利用该模型对现有数据进行了仿真预测,结果表明,该模型具有较高的预测精度。在此基础上,对河南省"十二五"期间的农业机械总动力进行了预测,并给出了预测数据。更多还原

【Abstract】 Total power of agricultural machinery is one of the most important indicators in agricultural mechanization.The prediction of agricultural machinery’s total power in Henan province will provide a theoretical support in the drawing of agricultural mechanization development plan,and it will also provide a reference to agricultural machinery enterprises in understanding future market conditions.The study is based on data of agricultural machinery’s total power in Henan province from 1991 to 2010.The model of prediction was founded using back error propagation neural network.There are three layers in this model;the number of neurons in each layer is 5,13 and 1 respectively.The activation function in hidden layer is tan-sigmoid function;the activation function in output layer is log-sigmoid function.The idea of step prediction and the training method of adaptive learning rate were used in the training.The weight of each connection and the threshold valve of each neuron were gained.The result of simulation prediction with existing data shows that the model has high prediction accuracy.Then,values of agricultural machinery’s total power in Henan province from 2011 to 2015 were predicted,and prediction values were also provided.更多还原