基于BP神经网络的辽宁省农机总动力预测

Prediction on total power of agricultural machinery in Liaoning Province based on BP neural network

为解决农业兼业化、农村空心化和农民老龄化等问题,需要在提高农业机械化水平、农业生产效率、优化农业产业结构和降低农民劳动强度基础上,保证当地农业机械化发展水平符合当地实际需求.因此要求制订农业机械化发展规划所依据的预测数据具有较高的准确性.本文使用遗传算法优化后的BP神经网络以1997～2012年辽宁省农机总动力为时间序列预测模型进行预测.预测结果为：到2014年辽宁农机总动力将达到2.789x107kw,较1994年上涨189.7％,年平均增长5.56％.由预测结果可知,预测值与实际值最大误差2.877％,预测值与历史样本数据之间的绝对值平绝误差为1.124％.预测结果准确性较高,预测精度稳定性较好,为制订农业机械化发展规划提供理论基础和数据依据.

In order to solve the problems of farmer＇s pluralism, rural hollowing and agricultural population aging, the advancing of agricultural mechanization level should be assured to meet the local needs on the basis of the improvement in agricultural mechanization level, production efficiency and its production structure and the lowering of work intensity. Therefore, the predictive data, on the ground of which the plan of agricultural mechanization level would be constructed, must be of high accuracy. By using BP neural network which has been optimized by genetic algorithm and the time series model of total power of agricultural machinery in Lianning province from 1997 to 2012, the prediction was made. The total power of agricultural machinery will reach 2.789xlO7kw in 2014 in Liaoning province, which is189.7% higher than that in 1994, with an annual increase of 5.56%. From the prediction, the maximal error between the predicted and the actual data is 2.877%, and the mean absolute error between the predicted data and historical sample data is 1.124%. The predicting result is of high accuracy with good stability. This research offers theoretical foundation and data basis for the formulation of the development project of agricultural mechanization level.