摘要
通过分析收割机传感器输出信号,发现输出信号中掺杂着大量无用噪声信号,使测产的精度产生了很大误差。一般的噪声信号处理使用一些简单的数字滤波方式可以起到滤波的作用,但收割机在田间行驶过程中车辆振动及田间沟壑等对传感器产生的不规则和随机的振动会造成称量结果的误差大和不稳定。为此,提出了一种小波分析结合神经网络的算法对传感器输出信号进行降噪处理的方法。通过实验仿真得出,采用小波神经网络算法处理后的测量信号比传统滤波去噪方法更接近实际测量结果,测量相对平均误差可减小到2.14%。
By analyzing the sensor output signal of the harvester,it is found that the output signal is mixed with a large number of unwanted noise signals,which makes a great error to the accuracy of the measurement. General noise signal processing using some simple digital filtering method can play the role of filtering. However,the irregularities and random vibrations caused by the vibration of the harvesters during the field movement,such as vehicle vibration and field gully,are large and unstable. Aiming at this problem,this paper proposes a wavelet analysis combined with neural network algorithm to denoise the sensor output signal. The experimental results show that the measured signal processed by wavelet neural network algorithm is closer to the actual measurement result than the traditional filtering denoising method. The relative mean error of the measurement can be reduced to 2. 14%.
出处
《农机化研究》
北大核心
2018年第6期43-46,52,共5页
Journal of Agricultural Mechanization Research
基金
河北省自然科学基金项目(F2015209242)
关键词
收割机
测产
降噪
小波神经网络
harvester
yield measurement
noise reduction
wavelet neural network
