Date of Completion


Embargo Period



Load forecasting; Neural Network; Unit Commitment; Dynamic Line Rating

Major Advisor

Peter B. Luh

Associate Advisor

Yaakov Bar-Shalom

Associate Advisor

Laurent Michel

Field of Study

Electrical Engineering


Doctor of Philosophy

Open Access

Open Access


Utility companies, Independent System Operators, and Transmission Operators require sophisticated techniques in short-term load forecasting, unit commitment (UC), and dynamic line rating in view of the complicated customer behavior, increasing integration of renewable energy, and characteristics of transmission lines in power systems. In this thesis, a set of novel methodologies are developed to address the challenges posed by each problem.

1. Short-term load forecasting at the distribution level is difficult in view of the complicated load features, the large number of distribution-level nodes, and possible switching operations. A hierarchical forecasting approach is established to capture load characteristics at different levels. Load of a root node at any user-defined subtree is first forecast by a wavelet neural network with appropriate inputs. Child nodes categorized as “regular” and “irregular” based on load pattern similarities are forecast separately. Switching operation detection and follow-up adjustments are performed to capture abnormal changes and improve forecasting accuracy.

2. With a large number of combined cycle (CC) units represented by configuration-based modeling, solving UC problem through the state-of-the-practice branch-and-cut method suffers from poor performance. The recently developed Surrogate Lagrangian Relaxation is significantly enhanced through adding quadratic penalties on constraint violations to accelerate convergence. Quadratic penalty terms are linearized through a novel use of absolute value functions. Therefore, resource-level subproblems can be formulated and solved by branch-and-cut. Complicated constraints within a CC are thus handled within a subproblem. Subproblem solutions are then effectively coordinated.

3. Employing dynamic thermal rating, which adapts the thermal capacity of an overhead transmission line based on dynamic weather, is difficult in view of many weather factors, weather data availability, weather uncertainties, and the topology of transmission lines. In the developed probabilistic forecasting model, major weather factors are selected based on impact analysis and are modeled through a spatio-temporal regression with all available data sources. Spatial topology and weather uncertainties are simultaneously captured by treating the line rating as the minimum of critical span thermal capacities. An approximation is applied to determine the distribution and extract appropriate percentiles with light computational costs.