基于卷积神经网络的拖拉机工况识别

农机工况识别在细化农机作业状态和帮助掌握区域污染物排放趋势方面有着重要的研究价值。基于拖拉机不同运行状态下的行驶速度、发动机转速以及实时油耗等时间序列,首次提出将图像识别方法引入到拖拉机工况识别中的思路,并分别应用参数优化的支持向量机与卷积神经网络对实际作业拖拉机工况进行研究。结果表明:(1)基于参数优化的支持向量机可以较好地实现样本点的工况识别且识别准确度达到99.851 9%,但无法实现农机工况的连续性识别,同时无法对农机工况转换阶段进行有效识别。(2)以拖拉机运行速度与发动机转速等信息构建样本图像来描述农机工况变化的数据表达,并在此基础上应用卷积神经网络可以有效实现农机工况的连续性识别,且识别准确率可以达到93.3%。本研究在农机工况识别方面具有一定参考价值,并为后续农机不同工况下区域污染物排放研究提供技术支持。

Recognition of tractor working condition based on convolutional neural network

In general, the identification of working conditions of agricultural machinery had significant research value in refining the working conditions of agricultural machinery and helping to master the trend of regional pollutant discharge. Based on the time series of the tractor running speed, engine speed, and real-time fuel consumption under different running conditions, the research introduced the image recognition method into tractor working condition identification for the first time. At the same time, the research also applied the parameter optimized support vector machine and the convolutional neural network (CNN) to conduct a systematical study related to the tractor working conditions. The related research results indicated that a support vector machine based on parameter optimization could realize the working condition identification of sample points in an ideal way, with the recognition accuracy reaching 99.851 9%. Nevertheless, it cannot realize the continuous identification of agricultural machinery working conditions, nor can it effectively identify the conversion stage of agricultural machinery working conditions. Moreover, in this study, a range of information, including tractor running speed and engine speed, are used to construct the sample image, thereby describing the data expression of agricultural machinery working condition change. The application of the convolutional neural network (CNN) is beneficial to realize the continuous recognition of agricultural machinery working conditions effectively, with the recognition accuracy reaching 93.3%. In short, the research not only provided reference value for the research related to the identification of agricultural machinery working conditions but also provided corresponding technical support for the subsequent research on the regional pollutant emissions produced by agricultural machinery under different working conditions.