EXPLORE PUBLICATIONS BY COUNTRIES

Univariant Linear Local Trent ULLT, Adaptive Linear Neural Network Adaline, Common Information Model CIM, Artificial Intelligence AI, Machine learning ML, Frequency Synchronization.

Power reliability is becoming problematic nationwide. Power outages are becoming increasingly common for various reasons which include climate change, the increase in distributed energy resources (DERs), and old infrastructure. Climate change has promoted a societal push toward carbon neutrality. Energy demands and the correlated load requirements on the grid are increasing and are expected to rise. Additionally, consideration is needed for the predicted largescale use of electric vehicles (EV), along with proliferation of data centers for things such as bitcoin and generative artificial intelligence (AI) types of industries with large computation requirements. Currently, data centers are contributors to the largest growth of power consumption. The capacity requirements to meet future predicted load demand along with anticipated variable energy resources with limited inertia necessitate a quicker method for handling the dynamic impacts to grid stability. Current infrastructure, processes and procedures are not capable of meeting future requirements and a new methodology is necessary. Reliable, stable, resilient forecasting is needed. State estimation is foundational for monitoring real time grid conditions. However, today there are not enough data points, and scalability is needed. AI is critical for monitoring and solving real time issues. There is a growing need for AI integration into the power grid, due to an increase in complexity, demand, and a reduction in overall grid inertia.

In this research work we use a proactive model versus the currently used reactive model in the research community and the available literature where changes are not made until after an event has occurred. Our proposed method utilizes Common Information Model (CIM) connectivity and integration to implement Univariant Linear Local Trend (ULLT) to produce predictive grid state values and Adaptive Linear Neural Network (Adaline) to provide an optimized control signal value. Generation source frequency sensor data is input to ULLT as a time sequence trend and predictive frequency values are generated for each generation source. The predictive frequency values are processed by Adaline to obtain an optimized control value. The system is tuned to utilize predictive future frequency values to correlate to the time the optimized control values signal is implemented.