Exploited the long-term dependencies in the electric load time series in the States of Texas for predicting more accurate electricity usage by using the recurrent neural network and to help ERCOT develop a contingency plan to respond to the high demand electricity usage under extreme weather. On January 1, 2002, the Texas State Legislature decided to deregulate the electricity industry and open up the supply of electricity to competition. [1] The Electric Reliability Council of Texas (ERCOT), which was formed in 1970, is the organization responsible for managing the flow of electricity to the majority of customers living in Texas via an electric grid and helping to manage regulations for Texas utilities. [1] As the independent system operator for the region, ERCOT schedules power on an electric grid that connects more than 46,500 miles of transmission lines and 710+ generation units to supply more than 26 million Texas customers – representing about 90% of the state’s electric load. [2]
In 2021, the State of Texas suffered a major power crisis which came about as a result of three severe winter storms sweeping across the United States on February 10-11, 13-17, and 15-20. [3] The Texas power outage, which has left millions without power, happened due to the increased use of heaters ramped up power use under the extreme weather, and the wells and pipes of the primary energy source, gas, got frozen and blocked. [4]
Nowadays, there is an irresistible trend of the electric power improvement for developing the smart grids, which applies a large number of new technologies in power generation, transmission, distribution and utilization to achieve optimization of the power configuration and energy saving. [5] Electricity load forecasting in smart grid is crucial to prevent similar power crises across the States and to help ERCOT to predict electricity usage under extreme weather in Winter. Short-term electric load forecasting forecasts the load that is several hours to several weeks ahead. [6] However, accurate forecasting is challenging due to the nonlinear, non-stationary, and nonseasonal nature of the electric load time series. [6]