deep learning based average current signal prediction using lstm network

One of the challenges faced by power distribution companies is the prediction of average current in order to enable proper planning for sudden increases and decreases that occur in the sinusoidal current signal. this planning can involve reducing production or strengthening electrical transformers and other equipment before reaching their limits, resulting in cost savings in terms of repairs, minimizing industrial equipment failures, and ultimately benefiting the company. recently, in line with the smart grid initiative, data loggers have been installed in city-level power substations to transmit information such as voltage and current. with this data, which spans one month, we have developed a deep learning model using long short-term memory (lstm) networks to predict the average current for the upcoming week. through a comparative analysis, we have demonstrated the superior performance of our lstm model in comparison to other neural networks, including mlp and gru.