To increase the accuracy of predicting the operation level of agricultural mechanization in China, this study establishes a wavelet-BP neural network prediction model by targeting the nonlinearity and non-stationary features of the data under the fundamental principle of wavelet analysis and BP neural network. First, the major factors that influence the operation level of agricultural mechanization are determined and analyzed, and dimensionality is reduced through a principal component analysis. Second, the time series of the operation level of agricultural mechanization and the principal component series of the influencing factors are decomposed to obtain low-frequency and high-frequency components. A BP neural network prediction model is built for the low- and high-frequency components. Lastly, the obtained low-frequency and high-frequency components are examined through linear superposition, and the final prediction results are obtained. The proposed method is verified by predicting the operation level of agricultural mechanization in China. Results show that the wavelet-BP neural network prediction model can perform accurate prediction. The model evaluation indices, namely, average relative error, root-mean-square error, Theil IC, consistency indicator, effective coefficient, and excellence rate, are 0.44%, 0.293, 0.002 4, 0.90, 0.972 7, and 100%, respectively; these indices are superior to those of conventional and other models. The research findings can serve as a theoretical basis for the formulation of relevant agricultural mechanization policies and laws in China.

为提高我国农业机械化作业水平的预测精度，针对农业机械化作业水平非线性和非平稳性的特点，基于小波分析和BP神经网络的基本原理，建立小波-BP神经网络的预测模型。首先，系统地分析并提取农业机械化作业水平主要影响因素，采用主成分分析的方法进行降维处理；然后，对我国农业机械化作业水平时间序列和影响因素主成分序列进行小波分解获取低频分量和高频分量，进而对低频分量与高频分量分别建立BP神经网络预测模型；最后，将预测得到的低频分量和高频分量通过线性叠加得到最终预测结果。以我国农业机械化作业水平预测为例对该方法进行验证，结果表明：小波-BP神经网络预测模型具有较好的预测效果，模型评价指标平均相对误差、均方根误差、希尔不等系数、一致性指标、有效系数和优秀率分别为0.44%、0.293、0.0024、0.90、0.9727和100%，各评价指标均优于其他模型。