Process capability =6*σ
Where "σ" is the standard deviation of the process under statistical control. According to the definition of standard deviation, we know that the standard deviation of a group of data is the average value of each single data in the group deviating from the average value, so the essence of standard deviation is the width data of the distance between two points on the same number axis, which describes the overall dispersion degree of the group of data (that is, the consistency between the data). The greater the standard deviation, the greater the overall deviation. Generally speaking, the smaller the standard deviation, the better. Therefore, the value of process capability (6σ) can also be regarded as the width of an interval on the number axis, and the smaller the better. But how small is better?
Second, or should I ask, how small is the value of process capability acceptable? It is not difficult to solve this problem, we just need to find a reference for it, and this reference should be a constant width value. The specification tolerance of the object we study just meets this condition, so we usually compare the process capability (6σ) with its corresponding specification tolerance (USL, LSL).
USL-LSL
6σ
Thirdly, its ratio is called process capability index CP, and the value of CP represents the multiple of the specification tolerance width of the research object and the process capability width. The greater the multiple, the stronger the process capability.