Evaluation of Sustainable Growth in Power Grid Assets Based on GRU Neural Networks

As power grid systems become more complex and as utilities face increasing pressures to optimize investments and reduce costs, accurate life-cycle cost (LCC) prediction models are essential for effective asset management. This paper proposes a GRU-based LCC prediction model for power grid assets, aiming to enhance the forecasting accuracy and efficiency of asset lifecycle costs. The model leverages the capabilities of Gated Recurrent Units (GRU), a type of recurrent neural network, to handle the intricate time-series data associated with power grid operations, including equipment condition, failure rates, and maintenance schedules. A comparative analysis of the GRU, LSTM, and BP models is conducted using performance metrics such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and accuracy. The results demonstrate that the GRU model outperforms both LSTM and BP models. This study highlights the effectiveness of GRU in improving the precision of long-term asset cost predictions, which is crucial for making informed investment decisions, optimizing resource allocation, and ensuring the sustainable growth of power grid infrastructure. The proposed GRU-based model offers a robust solution for enhancing asset management strategies, particularly in the context of integrating renewable energy sources and IoT-based grid monitoring systems.