Research Of Extreme Load Measurement And Its Application In Engineering Vehicles Driveline

Extreme load in the process of operation of construction vehicle will cause largerfatigue damage, even cause the final fracture. What’s more, the accurate measurementof extreme load has significant effect on accuracy of load spectrum compiling, whichaffects the reliability of the life prediction. Measure the possible extreme load or makereasonable estimation of probability which extreme load appears in the whole lifeperiod from a limited load samples is the important basis of compiling load spectrumprecisely. The extreme theory has been established in the field of economy, weatherand wave research, but there are still significant variances between the predictedextreme load and the measured values in engineering vehicles. This paper introducesthe extreme theory to the study of extreme load measurement combined by the NationalNatural Science Foundation (No:), and discusses the determination of threshold andmodel in this process. The main research contents and works are listed as follows:Firstly, this paper introduces the EMEF method and the Q-Q plot to diagnose thethick tail of the measured load data with its amplitude and mean. The Block-Maximamodel (BMM) and Peak-Over-Threshold model (POT) are adopted to measure theextreme load based on the thick tail characteristics. This paper mainly summarizes theparameter estimation method of generalized extreme value distribution (GEV) andgeneralized pareto distribution (GDP). Monte carlo simulation method is adopted tostudy the estimated accuracy under different parameters and different sample size.Secondly, a new threshold selection method is proposed based on comprehensiveevaluation in view of the uncertainty in graphic method. Monte carlo simulationmethod is adopted to verifiy the effectiveness of the proposed method.Finally, the application of BMM model and POT model is concerned. The tailexact probability model of measured load data is established, extreme load calculatedby which can be used for checking static strength. Mixed probability model ofamplitude is established and distribution characteristics is described, which establishstandard for load spectrum compiling. Analysis results show that sample size requirement of POT model is less obviously compared with BMM model and POTmodel have higher stability and smaller error. So we should adopt the extreme loadmeasured by POT model when load sample is short. However BMM model can be usedwhen load sample is large, because this method don’t need to choose the threshold valuewhich is simple and convenient. In order to get the best estimated result of extremeload, both models can be adopted if possible.