Here we present a high-precision method for predicting threshing drum clogging in drum harvesters, with strong applicability for longitudinal axial flow in crawlers. This method can be used to determine the alarm threshold. It entails placing a wireless vibration sensor at the outer end of the rotary shaft of the harvester to collect axial vibration signals from the drum to detect early fault characteristics. The method employs the Hilbert-Huang transform-analysis method to obtain time-frequency spectrum characteristics of the blocking process, thus diagnosing the speed reduction necessary to shut down when the threshing drum is clogged. The traditional method of extracting fault characteristic frequency by Fourier transform is compared with the proposed method. This method can not only diagnose the critical value of the rotating speed when the drum is clogged but also accurately and clearly reflect the trend of the drum clogging fault. The fault is within 3 standard deviations above or below the mean characteristic frequency, and abnormality is within 1 standard deviation above or below the mean characteristic frequency. Based on our experimental analysis, we introduce the concept of the confidence interval, andwe consider crop type and the diversity of working conditions. The proposed method improves the adaptability of forecasting to specific environments and combines agricultural machinery with agronomy. The trend of blocking can be predicted at least 1 s in the early stage of blockage by using a sliding window to process the vibration data collected in real time. This predictive ability makes the method superior to speed-sensor analysis in precision

No abstract

No abstract