Published: June 2026 Technical Level: Advanced Category: Artificial Intelligence
Power systems engineering education at the undergraduate and graduate levels was designed around a curriculum that has changed incrementally since the mid-twentieth century: circuit theory, machine theory, power electronics, protection, power systems analysis, and design. The emergence of AI tools as standard professional practice, the rapid growth of distributed energy resources, and the evolving role of the engineer from calculator to integrator create pressure for curriculum redesign that is more substantial than incremental course additions can address. This paper examines the educational requirements that current and emerging power engineering practice creates, identifies the gaps between those requirements and the structure of conventional curricula, and proposes a curriculum framework that develops both the foundational technical knowledge and the AI tool fluency that new engineers need to function effectively in the transformed professional environment. The central argument is that AI tool fluency must be built on a foundation of deep technical knowledge — not substituted for it — and that the most important curriculum adaptation is ensuring that students develop the engineering judgment to evaluate AI outputs critically before they are professionally dependent on tools they do not understand.
Engineering education has always faced a tension between the breadth of knowledge the profession requires and the depth that distinguishes engineering analysis from technical literacy. In power systems, the core curriculum must cover electromagnetic theory, circuit analysis, machine modeling, power electronics, power systems analysis (load flow, short circuit, stability), protection and control, and design integration — a scope that fills four years with little room for elective depth. The AI transition adds new requirements: data science foundations, machine learning methodology, programming competence (Python, numerical methods), and the judgment to evaluate probabilistic AI outputs against deterministic engineering requirements.
The typical response to new curriculum requirements is to add courses at the margin — an elective on machine learning for power systems, a senior seminar on AI tools — without restructuring the core. This response is inadequate for the current transition because the skills needed are not a separate module that can be bolted onto the existing curriculum; they are integrated with the core technical content. A student who learns load flow analysis and also learns that AI approximations exist for load flow, but learns them in separate courses with no integration, will not develop the judgment to know when to use the approximation and when to run the full solver. The integration must be in the curriculum, not just in the catalogue.
Most undergraduate electrical engineering curricula include a programming course — typically an introduction to MATLAB or Python — as a requirement but do not integrate computational thinking throughout the power engineering sequence. Students who learn basic programming as a standalone course and then spend the bulk of their power engineering courses using specialized software (ETAP, SKM, PowerWorld) without programming those tools do not develop the computational fluency that AI-augmented engineering practice requires. They can operate the tools but cannot write a script to batch-process results, cannot implement a simple ML model to classify load data, and cannot debug a data pipeline that feeds real-time measurements into a building energy management system.
Closing this gap requires computational exercises to be integrated throughout the power engineering sequence: load flow studies where students implement the Newton-Raphson algorithm before using ETAP, protection coordination problems where students evaluate TCC criteria programmatically before using coordination software, and energy management problems where students build simple forecasting models before studying commercial BEMS platforms. The objective is not to make power engineers into software developers but to ensure that they understand the computational basis of the tools they use and can operate in environments where the standard tools are not available or not adequate.
Power systems generate large volumes of time-series data: smart meter readings, SCADA measurements, power quality monitor records, relay event logs, and weather data. The analysis of this data for load forecasting, anomaly detection, equipment condition monitoring, and planning is a growing component of utility and engineering firm practice. Undergraduate power systems curricula rarely include systematic exposure to time-series analysis, regression, or classification — the statistical and machine learning foundations that this analysis requires. Students who graduate without these foundations are not prepared to participate in data-driven analysis projects without significant on-the-job retraining.
The curriculum integration point is the power systems analysis sequence. Load forecasting — a standard component of power systems planning — can be developed from first principles of time-series analysis (ARIMA, regression with weather variables, neural network approximation) using real load data from publicly available sources (EIA, utility open data portals). Protection course laboratory exercises can use recorded relay event data to develop classification models for fault type identification. These integrations do not require additional credit hours; they reframe existing course content in a data-driven context.
The most important and least systematically taught component of engineering education is professional judgment — the ability to evaluate technical outputs critically, identify when a result is physically implausible or inconsistent with the engineering basis, and recognize the boundary between what the analysis supports and what it does not. This competency is essential for AI tool use because AI tools produce confident-sounding outputs for inputs outside their training distribution, and the engineer who does not recognize physically implausible results will accept them.
Professional judgment is difficult to teach directly because it is domain-specific and develops through experience with a wide range of problems, including problems where the analysis goes wrong and the error must be identified and corrected. The most effective pedagogical approach is structured engineering problem-solving with embedded errors and anomalies: load flow problems where the student must identify why the solution is physically inconsistent with the system topology, protection coordination exercises where a candidate setting set is presented and the student must identify the coordination failure before it is pointed out, and design review exercises where a peer's or AI-generated design must be checked against code requirements and physical constraints. These exercises develop the critical evaluation habit that AI tool use requires.
A curriculum framework that develops both foundational technical depth and AI tool competency for power systems engineers should be organized around three integrated strands running through the four-year undergraduate sequence: the engineering fundamentals strand covering circuit theory, machine theory, electromagnetic theory, and power electronics; the power systems analysis strand covering load flow, short circuit, protection, stability, and design; and the computational and data strand covering programming, numerical methods, statistical analysis, and machine learning applied to power systems problems.
The computational and data strand is not a separate track — it is integrated into the power systems analysis strand from the beginning. When students learn Newton-Raphson load flow, they also implement it. When they learn inverse-time overcurrent relay characteristics, they also write a script to evaluate coordination margins across a fault current range. When they study load forecasting in a power systems planning course, they also build a simple regression model using real data before studying the commercial optimization tools. This integration ensures that the computational competency is developed in the context of engineering content, not in isolation.
At the graduate level, a one-semester course on AI and machine learning for power systems that covers the dominant ML architectures used in power engineering applications — LSTM for load forecasting, CNN for fault classification, reinforcement learning for optimal dispatch, physics-informed neural networks for power flow approximation — and requires students to implement and validate examples from each category is appropriate as a required course for power engineering graduate students rather than an elective. The course should include explicit treatment of the validation requirements for AI tools in licensed engineering practice, the liability implications of AI-generated outputs, and the conditions under which AI approximations are and are not appropriate substitutes for rigorous analysis.
The analysis in this paper connects to several companion studies in this library. Readers concerned with the upstream and downstream engineering will find AI and Power Systems Engineering Employment develops a closely related aspect of the same problem, while The AI Transformation of Power Systems Engineering extends the treatment into an adjacent domain. For the broader methodological context, AI Risk Assessment for Power Systems Engineers provides complementary depth.
The gap between conventional power systems curricula and the competencies that current practice now demands is structural rather than incremental, and closing it requires deliberate curriculum redesign rather than the addition of isolated elective courses. The framework developed in this paper identifies three competency areas — computational and data science foundations, data-driven analysis methods, and the professional judgment and liability reasoning that distinguishes engineering from calculation — that must be woven through the core sequence rather than appended to it. For educators and program directors, the operative conclusion is that the engineer's evolving role from calculator to integrator changes what the degree must certify: graduates must be able to direct, validate, and assume professional responsibility for AI-assisted analysis, which requires deeper grounding in the physical principles that let an engineer recognize when a tool's output is wrong. The recommended curriculum framework preserves the irreplaceable core of circuit theory, machine theory, protection, and power systems analysis while building the computational and judgment competencies on top of that foundation, producing graduates equipped for a profession in which the tools change continuously but the responsibility for correct engineering does not.
[1] ABET, Criteria for Accrediting Engineering Programs, 2024–2025, ABET Engineering Accreditation Commission, 2024.
[2] IEEE Power and Energy Society Education Committee, PES Education Report: Power Engineering Curriculum for the Energy Transition, IEEE PES, 2023.
[3] A. G. Phadke and J. S. Thorp, Computer Relaying for Power Systems, 2nd ed., Wiley-IEEE Press, 2009.
[4] P. Kundur, Power System Stability and Control, McGraw-Hill, 1994.
[5] U.S. Department of Energy, Workforce Development for the Energy Transition: Skills and Training, DOE, 2023.
[6] National Academy of Engineering, Educating the Engineer of 2020: Adapting Engineering Education to the New Century, NAE, 2005.
[7] EPRI, Artificial Intelligence Applications for Power Systems, EPRI Technical Report 3002018003, 2021.
[8] J. Rawlings, D. Mayne, and M. Diehl, Model Predictive Control: Theory, Computation, and Design, 2nd ed., Nob Hill Publishing, 2017.
[9] IEEE Standard 1547-2018, Standard for Interconnection and Interoperability of Distributed Energy Resources, IEEE, 2018.