Anda di halaman 1dari 3

Process Behavior Charts and Chaos Theory

Donald J. Wheeler, Ph.D. W. Edwards Deming always described assignable causes as special causesand said that special causes are not part of the system. Thus, he considered a special cause as a temporary causefleeting and ephemeral. Shewhart was less consistent. He sometimes wrote of assignable causes as temporary, but at other times, he seems to have had another model in mind. For example, when he defined a state of statistical control to exist when the chance fluctuations in a phenomenon are produced by a constant system of a large number of chance causes in which no cause produces a predominating effect, Shewhart seemed to imply that when one of the chance causes begins to have a dominating effect, it becomes an assignable cause. Both conceptualizations are valid. Some assignable causes are indeed external to the system while others are internal to the system. This is why the identification of some assignable causes can result in process improvementa tightening of the Natural Process Limits. The examples in Shewharts first book demonstrate this, and the phenomenon has been observed by many who have used his approach. The reason process behavior charts are not sensitive to the conceptual model is that they are holistic in nature. Shewhart set out to characterize the behavior of data from a process, rather than to create a detailed mathematical model of the process itself. Today, a new conceptual model is coming into use. This article will discuss some aspects of the relationship between control charts and Chaos Theory and demonstrate that Shewharts charts work just as well within the new paradigm. Chaos Theory is a relatively new field of study which focuses on characterizations of complex phenomena. A system of several simple cause-and-effect relationships often gives rise to very complex phenomena. A system of several simple cause-and-effect relationships often gives rise to very complex dynamic behaviors. Consider, for example, the horizontal motion of the ball in the mechanism shown in Figure 1. The ball is attached to a spring that gets stiffer as it is stretched or compressed (a nonlinear restoring force). As the board moves cyclically back and forth, the spring is pushed and pulled and causes the ball to move.

Figure 1: A Simple System The movement of the ball is characterized by the horizontal position of the ball and its velocity at any point in time. If the force exerted by the board is weak, then the ball will move in a simple manner. Plotting the position over time would result in a time-series like that shown in the left hand portion of Figure 2, while plotting the velocity versus the position would result in the state space diagram shown on the right of that figure.

When the force from the board is weak, the ball moves in a simple trajectory, which repeats with each cycle of the force from the board. Because the motion is periodic, the state-space path will retrace itself with each board cycle. This simple orbit in the state space is said to be a period-one attractor.

As the driving force on the spring is increased beyond a certain point, the period-one attractor will become unstable and a two-period factor will form. In this state it will take two board cycles before the motion of the ball repeats itself. This period-two attractor is shown in Figure 3.

If the force on the spring increases past some threshold, the ball moves chaotically. No pattern is apparent in the time-series in Figure 4 and the orbits in the state space will never retrace themselves. Nonetheless, the orbits only occupy certain regions of the state space. This latter condition, known as a chaotic attractor, may be considered to be a combination of unstable periodic orbits.

Thus the behavior of a chaotic system is seen to be a collection of many orderly behaviors, none of which dominates under ordinary circumstances. (Figures 2-4 and this quote are reproduced with permission from William L. Ditto and Louis M. Pecora, Mastering Chaos, Scientific American, Aug. 1993, pp.78-82; Copyright 1993 by Scientific American, Inc. All rights reserved.)

The similarity to Shewharts description of a reasonable degree of statistical control is no coincidence. While Shewhart did not have the computational power to solve sets of differential equations, he discovered that we could characterize the overall behavior of the dynamic variables of any system as being predictable within limits or not. Likewise, the similarity between the time-series in Figure 4 and the time-series for a process which displays a reasonable degree of statistical control is strikingboth are characterized by quasi-random variation within bounds, and this variation is the result of many competing cause-and-effect relationships. Any attempt to control or manipulate the process by adjusting one of the cause-and-effect systems will be frustrated by the fact that no one of them is dominant, and the others will mask the effects of the adjustment. Therefore, it is futile to single out any one of these relationships as the magic bullet or steering wheel for this process. (Do not bother trying to find any assignable causes here.) On the other hand, the time-series in Figures 2 and 3 show regular patterns. Such patterns arise when at least one cause-and-effect relationship dominates the others. Here it will be profitable to discover the nature of this dominant cause-and-effect relationship, because such knowledge will provide you with the leverage to steer the process. (What is the assignable cause for the regular patterns?) Thus, over sixty years ago, Shewhart gave us a technique for detecting if a system displays chaotic variation. Without getting lost in the theory, Shewhart examined the behavior of observable variables and came up with a major distinction which is useful as a guide for action: If a process displays chaotic variation, it will be a waste of time to look for assignable causes (because it will be futile to focus on any one of the many competing cause-and-effect relationships.) If a process does not display chaotic variation, it will then be profitable to look for assignable causes (because at least one of the cause-and-effect relationships will dominate the others.) Process Behavior Charts empirically describe time-series as being either predictable or unpredictable. Since they do not impose a model upon the data, they are unaffected by the paradigm we choose to use in explaining the reason for the observed behaviors. This is why the charts have not become, and will never become, obsolete.

This article was first published in SPC INK, 1995 #2. Dr. Donald J. Wheeler is a specialist in the field of Data Analysis and SPC. He has written 15 books and over 120 publications and is internationally known as an author, teacher, and consultant. You may contact him through his company: Statistical Process Controls, Inc. 5908 Toole Drive, Suite C Knoxville, Tennessee 37919 USA Phone: 865-584-5005 Fax: 865-588-9440 Copyright 2002 SPC Press

Web Site: www.spcpress.com

Anda mungkin juga menyukai