کنترل کیفیت آماری در تولید میکرو از طریق چارت چند متغیره μ-EWMA

Micro-manufacturing processes are characterized by high process variability and an increased significance of measurement uncertainty in relation to tight tolerance specifications. Therefore, an approach that separates the superposition of measurement and manufacturing variation is demanded. A novel design for a quality control chart that makes it possible to monitor, control and extract measurement variation from manufacturing variation is proposed. Thus, a definite cause diagnosis on the approval or rejection of micro-components due to errors either in the measurement or in the manufacturing process is possible. The proposed multivariate μ-EWMA chart which is based on weighting each measurement data with its current measurement variation is discussed and benchmarked with traditional control charts.

In order to guarantee stable processes, micro-manufacturing techniques must be controlled and continuously improved by an effective quality assurance. Up to now, there are no such production accompanying methods at hand as the statistical process control (SPC) applied in the macro-world is so far only adequate for non-varying series production with data that is based on a capable measurement instrument [1]. In contrast, micro-manufacturing processes are characterized by an uncertainty of geometric measurement results. The assumption valid for macro-dimensions saying that measurement devices are 10 times more precise than the given tolerance intervals does not hold true for micro-production. The measured data is always subject to a superposition of manufacturing process variation and measurement variation. Furthermore, Estler [2] detected that dimensional metrology in particular is always based on incomplete information and that, therefore, the result of a measurement represents a probability distribution. As a consequence, the superposition of manufacturing process variation and measurement variation renders the control of micro-manufacturing processes more difficult and requires new guidelines and tools. The research challenge consists of the separation of manufacturing and measurement variation in order to control manufacturing processes and not a superposition of both processes. Finding a solution for this problem will allow for progress with regard to the effective control of micro-manufacturing processes and thus, the development of robust manufacturing processes.

To summarize, this article discussed a new QCC design especially set up for the statistical process control of micro-manufacturing processes. It was shown that measurement data always displays a superposition of measurement and manufacturing distribution and therefore, traditional QCCs may wrongly indicate an out of control condition of the manufacturing process. In order to continuously control both manufacturing and measurement process, a new type of QCC with a varying smoothing factor λt dependent on the actual measurement standard deviation is proposed and validated via empirical and simulation results. It was demonstrated that the proposed multivariate μ-EWMA QCC shows the best performance regarding the ARL in the specified cases and is the only chart that allows for the separation of manufacturing and measurement processes, facilitating a clear cause and effect diagnosis in case of an out of control situation.