Rolling bearing performance data robustness judge principles and methods

Robustness judgment principle of rolling bearing performance data: 1. Rolling bearing performance data histogram epicenter of data distribution characteristics of rolling bearing. In statistics, when the sample size is the number of more than 800 data can be thought of data distribution for normal distribution or asymptotic normal distribution, which is the data on both sides of the histogram for middle high low symmetric or nearly symmetric graphics. If rolling bearing performance teach according to histogram, conform to the normal points or asymptotic normal distribution, can be thought of rolling bearing performance data do not exist from the micro value, otherwise, can think of rolling bearing performance data of discrete values. Using the data frequency histogram with a normal distribution or asymptotic normal distribution, the characteristics of robustness performance data of the rolling element bearing can preliminary judgment. 2. Rolling bearing performance normality test using rolling bearing performance data histogram with a normal distribution or asymptotic normal distribution similarity, can preliminary judgment rolling bearing performance data are discrete values, and then progress test rolling bearing performance data of normality, deviates from the value of the normal distribution is the discrete values. Using this method can determine scope, the discrete values of rolling bearing performance data for subsequent rolling bearing performance data sound of lay the foundation. The robustness judgment method of rolling bearing performance data: data robustness to determine steps: data is steady is the precondition of data analysis, the robust data can be reasonable and correct estimation results, unsound data will lead to mistakes or even disastrous estimation results. Data is steady is the precondition of data analysis, the robust data in the experimental process, however, can reasonably come with the correct estimate due to the environment, instrument, temperature for rolling bearing performance data, such as the material of solid element shadow can fluctuate, deviating from the true value, pollution or discrete data, the data is the data distribution changes. Therefore, how to judge whether the steady rolling bearing performance experimental data is an important content of rolling bearing performance analysis. Gross error is part of the discrete data, the judgment criterion has lai, principles, and display a nie criteria, romanov, benchmark, Neil criterion, Mrs Grubbs criteria and Dixon criterion, etc. These guidelines are based on the experimental data obey a certain distribution, and the judgment method to calculate trival, gross error is only part of the discrete data at the same time. This section presents a rolling bearing based on statistical performance data sound judgment method, steps as follows: ( 1) Get a lot of rolling bearing performance through the experiment data, due to large rolling bearing performance data, can assume that the data is normal distribution; ( 2) According to the principle of statistics, the data are grouped and make data frequency histogram, and compared with normal distribution, can be roughly tell the differences of the experimental data with normal distribution; ( 3) Establish the probability distribution of rolling bearing performance data, and compared with normal distribution, the probability analysis of experiment data and normal distribution probability conforms to the situation; ( 4) If the rolling bearing performance data and normal distribution probability cumulative frequency on a straight line, the rolling bearing performance data is moderate, otherwise the rolling bearing performance data is not steady; ( 5) According to the data deviate from normal distribution, the analysis of the distribution of the rolling bearing performance of discrete data.