To determine the "Six Sigma" statistics, it is useful to define two related concepts, specification limits and the normal distribution.
Specifications Limits
The specification limits are the tolerances or the performance range of products or processes. In many cases, the limits of the specification determined by customer requirements. Example of a specification could be the size (this dimension) of a given circular hole, which opens with a drill, a circuit board on a construction site. The aim of the diameter of the hole can be xy mm, but diameters within the range between the lower (Low Specification Limit - LSL) and upper specification limit (Upper Specification Limit - USL) are accepted. This is because the volatility is so ubiquitous and inevitable in the real world, so should be allowed a certain degree of inaccuracy (Breyfogle & al 2001).
The target value of the range of standards-ideal price; typically located at the exact center between LSL and USL. The specification limits are completely independent of the bell curve shape representing the normal distribution. (Breyfogle et al 2001)
The normal distribution
The bell-shaped curve (Figure 3) is called normal distribution, also known as the Gauss curve. Because of its many properties, it is a useful and valuable tool in the world of statistics and quality. The curve is formed symmetrical and extends from minus to plus infinity on the x-axis. This standard curve is independent of the LSL and USL and represents the dispersion of the diameters obtained, for example, by drilling in a circuit board. The shape of the normal curve simply depends on the process, equipment, personnel and so forth, which can affect the drilling of holes. In other words, the graph summarizes the empirical determination of the amount of variation that exists in the hole fabrication process. (Breyfogle et al 2001)
The shaded vertical lines on the graph in Figure 3 represent the number of standard deviations (s) of a given diameter may be spaced from the average, which is shown as m in the x-axis. The table below the figure shows the number of parts per million that would be outside the specification limits "Six Sigma" if the data industry centered within these limits and have different standard deviations. (Breyfogle 1999)
Quality Level sigma
The scenario presented above addresses the situation where a process is centered. A shift ± 1.5s in average added usual to consider "typical" displacements and inclinations through a process centered on a standard price. This shifting of the average is used in calculating a procedure "planar sigma quality" as shown in Figure 4. (Breyfogle et al. 2001) The quality in terms of "Six Sigma" of Table 1 is comparable to Figure 4.
?? normal Distribution Focused
Μεταφράζονται, παρακαλώ περιμένετε..