Clustering Based Oil Production Rate Forecasting Using Dynamic Time Warping With Univariate Time Series Data

forecasting oil production rate from oil well have become focused in many researches. Many papers have presented the successful forecasting for individual well production rate. However single model for each well may not be effective for large oilfield which has hundreds or thousands of wells. In this situation, deploying clustering model prior forecasting might be the solution. Due to different historical data per each well in form of time-series data, dynamic time warping being proposed to deliver searching for minimum distance between each time-series data with different time-point. Reducing number of models definitely might impact model accuracy (56 of datasets have higher error), however it has benefit on less complex of model deployment. © 2021 IEEE.