Moving Average Filter for Optimizing Optical Character Recognition (OCR) on Perfume-Filling Machines

Filling perfume on a filling machine which is controlled based on a time limitation system, causes a difference between the set point and the filling results so that the operator must add or subtract to each perfume bottle that has been filled. Based on this, a perfume filling machine was designed with an effort to increase filling accuracy using a weight limitation system, where a load cell sensor is the instrumentation used. Based on error evaluation calculations using Mean Square Error (MSE), the research method using SMA has an average square error value of 0.195 grams. If expressed as a percentage, the error from noise reduction efforts is 0.20%. Meanwhile, WMA has an average square error value of 0.17 grams and an EMA of 0.62 grams. If expressed as a percentage, the error size of the WMA noise reduction effort is 0.20% and the EMA is 0.35%. The system reduced noise from load cell sensor readings using a simple moving average noise filter method, the Simple Moving Average (SMA) method. The application of the SMA method can increase the accuracy of sensor readings by an average of up to 0.13% in other words this method can reduce the average noise to 0.420 grams which is stable at around 1503 RPM - 1509 RPM, and has minimal vibration.