Improving Performance of Online Monitoring System Based on Control Chart with Adaptive Fuzzy Membership Function

Product quality strongly depend on production process quality. So, performance of production process must be strictly monitored. Statistical Process Control (SPC) is a technique used for evaluate process performance using statistical method to analyze, control, and influence process performance improvement. Combined of two control chart, individual chart and cusum chart could improve performance to analyze a process quality with SPC method In this final project research, made an algorithm which is able to used for resulting SPC decision about process status with fuzzy interference system. In order to detect plant characteristic must be made a fuzzy algorithm which could be applied to variables process (temperature, flow, pressure, and displacement) using 100 data consecutively (confidence level 100%) steady state approaching the aim of historical data records, made a Fuzzy Inference System making decision algorithm using normalized histogram. This research consist of two phase: offline SPC Adaptive Fuzzy membership function (MSPCAF) algorithm test and real time – real plant online SPC Adaptive Fuzzy membership function (MSPCAF) algorithm test. MSPCAF could be used for different plant process variable monitoring, also it more sensitive detecting non random out of control data with 3σ control limit rule than using uniform fuzzy membership function that gets It’s parameter by trial and error. Uniform fuzzy membership function able to detect all random out of control data, but it unable detect non random out of control data, resulting false alarm 5.88% ( 6 data) and missed alarm 1.96% (2 data) on 102 data (711 febuary 2006). MSPCAF able to detect both random and nonrandom out of control data (100% detected), so resulting no false and missed alarm on 102 data (7-11 febuary 2006).