Volume 39, Issue 6 (November 2011)
Fuzzy Hypothesis Testing and Time Series Analysis of Rolling Bearing Quality
Poor information means incomplete and insufficient information, such as unknown probability distributions and trends. Evaluation for the evolvement of the rolling bearing quality as a time series belongs to the category of information poor process. Statistics relied on known probability distributions and trends could become ineffective. For this end, a fuzzy hypothesis testing model is proposed to make variability analysis of a time series with poor information. By introducing the weight into the rejection region, the relationship of the improved equivalence relation and the empirical confidence level is established, laying the new foundation for a fuzzy decision-making for a time series with poor information. The model is characterized by permitting the probability distribution and the trend of a stationary or nonstationary time-series to be unknown. The experimental investigation on the friction torque of a rolling bearing shows that the model is correct and effective.