基于EMD-POT模型的拖拉机关键零部件载荷时域外推方法

Time-domain Load Extrapolation Method for Tractor Key Parts Based on EMD-POT Model

针对传统POT外推方法存在的两方面问题:对非平稳载荷适应性不足以及缺少对外推重构环节的讨论,提出基于经验模态分解-峰值超阈值(Empirical mode decomposition-peaks over threshold,EMD-POT)模型的载荷时域外推方法。首先,以拖拉机传动轴为研究对象,对其在犁耕工况下的实测数据进行载荷特性分析,结果表明,拖拉机作业状态的频繁转换是造成载荷非平稳的主要因素。其次,针对第一个问题,结合EMD方法,在不考虑趋势载荷的情况下,实现对主成分载荷的无损提取;针对第二个问题,提出外推负载筛选、阈值限定与极值排序相结合的方法,实现外推载荷的精准重构。最后,将外推后的主成分载荷与趋势载荷进行叠加实现载荷时域外推。将改进的外推方法与传统POT外推方法得到的结果进行对比,结果表明,改进方法能够显著解决传统方法在处理非平稳载荷时外推结果失真的问题,提出的重构方法能够进一步提高外推与实测载荷循环的一致性,均值和标准差的稳定性分别提高了28.5%和31.2%。损伤计算结果表明,改进方法的累计损伤能够合理覆盖实测损伤,相比于随机重构方法,损伤一致性提高了9.4%。外推结果可为仿真分析及可靠性试验提供更准确的负载输入,进一步提高了载荷谱的编制精度。

To address the two problems of the traditional POT extrapolation method:the lack of adaptation to non-stationary loads and the lack of discussion on the extrapolation reconstruction method,a load time-domain extrapolation method based on the empirical mode decomposition-peaks over threshold(EMD-POT)model is proposed.First,the tractor drive shaft is used as the research object,and its load characteristics are analysed by the measured data under ploughing conditions.The results show that the change of tractor operating conditions is the main factor causing the non-stationary load.Second,for the first problem,the EMD method is combined to achieve lossless extraction of principal component loadings without considering trend loadings;for the second problem,a combination of extrapolated load screening,threshold qualification and extreme value ranking is proposed to achieve accurate reconstruction of extrapolated loads.Finally,the extrapolated stable load is superimposed on the trend load,and the load history after extrapolation is obtained.The results show that the improved method can significantly solve the problem of distortion of extrapolation results when dealing with non-stationary loads by the traditional method,and the proposed reconstruction method can further improve the consistency of the extrapolated and measured load cycles,with the stability of the mean and standard deviation improving by 28.5%and 31.2%respectively.The damage calculation results show that the cumulative damage of the improved method can reasonably cover the measured damage,and the damage consistency is improved by 9.4%compared to the random reconstruction method.