Estimating vibration limit value based on statistics for engine testing data

Through analyzing twenty seven vibration data in ground fire testing of a liquid propellant rocket engine, and twenty eight vibration data of a turbofan engine, we find that on one hand the characteristic parameters are submitted to Gaussian distribution in a time series of ground fire testing, and on the other hand the characteristic parameters of different time series are more likely submitted to F distribution rather than Gaussian distribution. Based on the above work, we study two methods to estimate the vibration limit value of the engine respectively. One method is based on the probability equality principle, and the other adopts 3σ principle. We compare the above two methods with the same vibration data. Through analyzing the differences of the processing results, the advantages and disadvantages of the two algorithms are pointed out. In addition, the processing results are further compared with vibration limit value used in engineering practice, and the results demonstrate the effectiveness and correctness of the proposed methods.