Energy Systems Division

Renewables and Grid Analysis

• Advanced Grid Modeling Home

Advanced Grid Modeling

Hierarchical Machine Learning-based Optimal Parameterization Scheme for WECC Composite Load Model under All Disturbances

Status

Started in May 2022

Objectives

The purpose of the study is to develop an optimal parameterization scheme for the WECC CLM based on a simulation approach assisted by a unique combination of state-of-the-art machine learning (ML) techniques including imitation learning (IL) and reinforcement learning (RL). The parameterization scheme will be used for a thorough validation and benchmarking of the WECC CLM model and for developing optimal parameters for the CLM to duplicate dynamic behaviors of the aggregated loads and the entire system under various disturbances and operating conditions of the grid. The same parameterization scheme will be extended for cases with incomplete information about the loads, i.e., some parameters of the loads are unknown while some other information is becoming more and more available based on DOE's efforts. Additionally, we will propose an alternative approach to develop a library characterizing the basic load features that can be used to better build composite load models.

Approach

We will develop an innovative parameterization approach, i.e., a simulation-based approach assisted by a combination of state-of-the-art ML techniques including imitation and reinforcement learning. The proposed approach will optimize the parameters of the CLM such that the CLM equivalent behaves the same as, or very close to the set of representative loads (consisting of a large number of known load devices) under various disturbances in the simulation. Since the high-fidelity model of a known load device is available, a simulation-based approach will be developed together with the unique hierarchical imitation and reinforcement learning (h-IRL) for this task. We will select a load bus in an example system with a known

The proposed approach will be extended to scenarios without complete knowledge of the load devices and feeder structure. Data-driven methods will be developed for the identification of such information for building simulation models. In addition, a novel bottom-up load modeling method based on a mathematical library will be developed.