Cement Plant Power Demand Prediction Using 1D Convolutional Neural Network Model: A Case Study

The management and operation of the power system would be significantly enhanced if accurate prior knowledge of power exchanges were provided, which could provide valuable insights into the system and help predict the required energy, enabling appropriate solutions to various challenges and preventing failures. The power consumption, reactive power, and power factor are critical parameters for assessing the performance of power system operations and planning. This paper presents One-Dimensional Convolutional Neural Networks (1DCNN) models for forecasting fluctuations of these parameters within the electrical power demand of a cement factory, thereby enabling effective monitoring of the cement production line. In addition, Long Short-Term Memory (LSTM) and Bidirectional Long Short-Term Memory (Bi-LSTM) models are modified to improve their effectiveness. Subsequently, the outcomes of the models were compared with those of conventional ARIMA models. Meanwhile, Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Bias Error (MBE), and Mean Absolute Percentage Error (MAPE) are used to evaluate the predictive model's accuracy. It is a frequently used metric, especially in regression tasks. The results indicated that the suggested models can effectively predict the behavior of such a power system with an RMSE of less than 3%.