160 likes | 269 Views
IET 603 IP 1 Spring 2014 Christopher Smith. Chapter 10 Other Univariate Statistical Process-Monitoring and Control Techniques. Learning Introduction. Xbar and R charts for short Production Runs. The DNOM Chart is the most widely used c
E N D
IET 603IP 1 Spring 2014Christopher Smith Chapter 10 Other Univariate Statistical Process-Monitoring and Control Techniques
Xbar and R charts for short Production Runs • The DNOM Chart is the most widely used c • Essentially, the value that is plotted is the difference between the part measurement and the nominal value. (Note: The nominal value refers to the value that is typically specified on an engineering drawing as the desired value and is often halfway between the lower and upper specification limits.) Once the differences are computed, the control limits may be established in the normal way.ontrol chart.
Xbar and R charts for short Production Runs • The following table illustrates the calculations for the Deviations from Nominal Chart. Essentially, each measured data value is transformed by subtracting off the nominal value (xi = mi- Nominal). The deviations are then used in the usual manner to construct the desired control chart (e.g. Xbar/S Charts, Xbar/R Charts). • Xbar and S charts constructed from the “Deviation from Nominal Data” follow
Calculate Modified Limits for Shewart X bar control chart • This is needed when the natural variability or six sigma is much smaller than the spread of the actual engineering specification. • Calculation formula is (Process Capability Ratio) = (USL-LSL)/6 • Formula for Modified Control Limits =
How to set up an acceptance chart • Acceptance Control charts are used are used to take into account both the risk of rejecting a process operating at satisfactory level (Type 1 and the risk of accepting a process operating at satisfactory level (Type 2 error or
Control Charts for Multiple Stream Processes • A multiple stream process (MSP) consists of several identical processstreams. At a sample time, measurements are obtained from each stream (or a subset of the streams). The measurements are in the same units and often they have the same target value and variance. • MSPs are quite common: • thickness measurements across a webor sheet (as seen in paper production, galvanizing steel, and magnetic tape manufacturing); • diameter measurements at different radii or heights; • measurements of identical features on a single part such as the vanes on a compressor or impeller; • measurements from several identical production tools such as filling heads, cavities in injection molds, or different spindles; • measurements from identical test instruments; • and measurements from different locations on a wafer or disk such as in run-to-run process control in semiconductor manufacturing. • Group Control Chart Formulas Xbar chart UCL = Xbarbar = A2Rbar LCL = Xbarbar –A2Rbar • Rbar chart UCL = D4Rbar LCL = D3Rbar
Control Charts for Multiple Stream Processes With control charts for a MSP, it is important to detect and distinguish between assignable causes that affect all streams and assignable causes that affect one or a few streams. Typically, these two types of assignable causes have different root causes, and the distinction facilitates the identification of the problem and the solution.
SPC with Auto Correlated Process Data • Statistical Process Control (SPC) is widely used for monitoring the performance of processes in manufacturing. Traditional SPC methods require trained individuals to read data which results in slow and limited detection. • Much research has been devoted into developing an online automated system for SPC, so that the abnormality can be detected quickly and corrected by the process operation. To build a system as such, artificial neural networks (ANN) are widely used as tools where complex patterns can be difficult to recognize. • Many research projects involve using random data patterns for training and recognition of patterns for ANN/SPC applications. However, many manufacturing processes involve autocorrelated data, to determine the effect of autocorrelated data, green sand data was analyzed and a neural network was built and trained to analyze a number of out of control patterns. Overall, the network performed best for detecting larger mean shifts.
Model Based Approaches • In recent years, statistical process control for autocorrelated processes has received a great deal of attention. This is due in part to the improvements in measurement and data collection that allow processes to be sampled at higher frequency rates and, hence, data autocorrelation. • In general, the observation-based control charts perform very poorly when data are correlated over time. Under the assumption that the model is correct, the residual-based control charts are superior for all cases considered. • This suggests using a residual-based control chart to detect the mean shift. This is recommended particularly for chemical processes where there are often cascade processes with several inputs but only a few outputs, and where many of the variables are highly autocorrelated.
Batch Means Control Charts • Control charts can be applied to batch processes. However, as in multi-stream processes, the traditional X-bar chart cannot be used as you might expect. When you select subgroups for the X-bar chart, subgroup observations collected from within the batch provide an estimate of the within batch variation, but do not provide a good estimate of the between batch (or longer term) variation. As such, subgroups collected as multiple observations from a batch are irrational subgroups, and not useful for defining the control limits on the X-bar chart. • One alternative is to use a batch means chart, which allows you to use the multiple observation subgroup collected from the batch, but corrects the control limits on the X-bar chart to use the variation between batches, such as is done in a moving range chart. One advantage of the batch means X-bar chart is that it controls both the within batch variation (on the range chart) and the between batch variation (on the X-bar chart). The calculated process capability uses process sigma based on the moving range between the batch averages.
CuScore Control Charts • When monitoring a real-world process, the types of out-of-control situations that are likely to occur may be known ahead of time. For example, a pump that begins to fail may introduce an oscillation into the measurements at a specific frequency. In such cases, specialized CuScore Charts may be constructed to watch for that specific type of failure. • CUSCORE control chart is designed based on the thought of CUSCORE statistic to control the non-stationary autocorrelated process IMA (1, 1). The MMSE method is used to remove the auto-correlation among the observations. By comparing the performance of CUSCORE control chart to Shewhart Control Chart, the CUSCORE control is more effective in the case of the process mean drifting
Changpoint Models for Process Monitoring • This newly found approach focuses on finding the point in time where the underlying model generating a series of observations has changed in some fashion. • Most of the research on changepoints has focused detecting a sustained shift in the overall mean of the process.
Adaptive Control Charts • Traditionally, an Xbarchart is used to control the process mean, and an R chart is used to control the variance. However, these charts are not sensitive to the small shifts in the processes. • The adaptive charts might be considered if the aim is to detect process changes quickly. This type of chart is effective in detecting the disturbances that shift the process mean, increase or decrease the process variance, or lead to a combination of both effects.
Other Control Chart Designs • Health Care Outbreaks of Disease • Scanning statistics in Health • Tool Wear • Fill Control for Beverages
References Terms Taken From • http://www.winspc.com/what-is-spc/ask-the-expert/382-how-do-i-implement-spc-for-short-production-runs-part-1 • https://qualityamerica.com/knowledgecenter/statisticalprocesscontrol/multiple_steam_processes.asp • http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=6031726&abstractAccess=no&userType=inst • http://www.researchgate.net/publication/229880970_Modelbased_control_chart_for_autoregressive_and_correlated_data • https://qualityamerica.com/knowledgecenter/statisticalprocesscontrol/analysis_of_batch_processes_using_a_batch_means_chart.asp • http://www.statgraphics.com/control_charts.htm • http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=5576005&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D5576005 • http://link.springer.com/article/10.1007%2Fs00170-011-3662-2