Page 36 - 6653

of the component is influenced by the joint action of these
factors, and the distribution curve of the dimensions of the parts
is characterized by a triangle according to the Simpson law.
In the absence of prevailing factors in the processing of
parts, the most probable is the size distribution by the Gauss
law. Graphically, they meet the curve symmetrical shape with
rounded top.
When the symmetric distribution of the composing
errors і and any distribution law are the resultant error:
2
2
2
  K   K   K   ... K  n n 2 ,
3
1
1
3
2
2

where K 1, K 2, K 3, ..., K n – coefficients of relative scattering,
which depend on the type of the distribution curve of the
composite errors  1,  2,  3, ..., n.
The distribution of most random errors when working on
pre-adjusting automatic machines corresponds to the law of
normal distribution. Of course K=1,2.
The main characteristics of the distribution of random
errors are the center of size grouping (arithmetic mean) and the
mean square deviation.
Size grouping center:

A ср=A im i/N, (8.1)
where N – total number of parts in the party; A i – the size of
one detail; m i – the number of hit size A i in the second interval.
Average squared scattering  a number of parts sizes:

  ( x 2 m i / ) N , (8.2)
where x=A i- A m - deviation from the mean value.
With sufficient accuracy for the practice it can be
considered that the deviation of the true size of the entire batch

33