Predictive Scheduling System For Ball Mill Maintenance In Lead Paste Production Line Using Fuzzy Time Series Method

Lead Acid Battery consists of two main components: the active material, Lead Paste, and the passive material, Grid. The production of Lead Paste involves several continuous processes, starting from the Lead Lump-making machine, Ball Mill, and Paste Mixing. Lead Powder is produced through the processing of Lead Lump in the Ball Mill, resulting in finer material due to collisions caused by hot airflow and Ball Mill rotation. Lead Lump usage data is crucial for analyzing performance and identifying Ball Mill issues when there is a decrease in performance due to abnormalities in the machine. However, currently, performance trend data is obtained by calculating the initial stock of lead ingots with lead oxide stored in a silo. This method is less accurate and leads to data gaps between the Lead Lump-making and pasting processes. This project goal is to provide supporting data for the necessary maintenance scheduling by the maintenance and engineering team at an Indonesian based battery manufacturer. Fuzzy time series method is chosen as the acquired data is time series data, and this method allows the use of real-time data. The project goal is to achieve the most accurate prediction by comparing several fuzzy time series methods and other prediction methods using RMSE and MAPE evaluation, with an emphasis on using the fuzzy time series method known as the Singh model, which has shown the highest prediction accuracy in previous research. The results show that the Fuzzy Time Series Model Singh provides the most accurate predictions compared to other methods, with RMSE and MAE values of 6% and 5% respectively. Therefore, this method is considered suitable as a decision support system for maintenance scheduling and performance improvement on ball mill.