How to smoothly let the bearings fit in
PRODUCT
The characteristics of the rolling bearing estimation methods
by：Waxing
2020-06-17
The characteristics of the rolling bearing estimation methods: 1.
Median estimate the median estimate is under the rule of tiny maximization criterion and Hampel codes a optimal estimation of performance data.
Very solid, can well reflect the position of the sample data, defect is not reflect total 2.
HuberM estimate Huber M function for J (
t)
:
[
T, sichuan & lt;
K (
4 -
1)
J(
)
= Ksgn(
)
,I>
K type, J (
0 as HuberM function, 1 as the independent variable, K is constant, the SGN (
)
As the operator number function:
1, 1>
0
sgn(
)
={
0, t=0 (
4 -
2)
(
- - - - - -
1 < 1,
0 type, the SGN (
)
As the symbol function, I as the independent variable.
According to the type (
4 -
1)
And type (
4 -
2)
It can be seen that Huber M function is linear in the middle, the tail is constant, is continuous, not diminishing, bounded function.
Huber M estimation under the rule of tiny maximization criterion and Hampel codes is optimal robust estimation, can reflect the characteristics of the data, but it's difficult to apply in practice, basically has the following reasons: (
1)
Requirements for demanding on the experimental data, the data is continuous, not diminishing, bounded function;
(
2)
Threshold limit, but it is not easy to determine;
(
3)
No inspection standard is given.
3.
L estimates have significance level for a, L estimate is to point to the original sample and removed 100 a % of the number of observations is at each end, and the remaining 100 (
1 -
a)
Observed % value on average.
The advantage is that the average value is not affected by individual outlier data and sound, but the drawback is that the sample size, which reduces the amount of information.
Custom message
Related Products
