⚡ electrical engineering

Hybrid ANFIS-Based Fuzzy-PID Controller Design for Multi-Area Load Frequency Control in the Northwest Ethiopian Power Grid

This paper proposes a hybrid ANFIS-based fuzzy-PID controller for multi-area load frequency control in the Northwest Ethiopian power grid, demonstrating through MATLAB/Simulink simulations that it significantly outperforms conventional PID, fuzzy-PID, and standalone ANFIS controllers in terms of settling time, error metrics, and robustness against parameter variations.

Original authors: Fekadu Gebey Moges, Girmaw Teshager Bitew, Yechale Amogne Alemu, Birhane Wondmaneh Getahun

Published 2026-06-24

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Fekadu Gebey Moges, Girmaw Teshager Bitew, Yechale Amogne Alemu, Birhane Wondmaneh Getahun

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the Ethiopian power grid as a massive, interconnected orchestra. Every power plant (the musicians) must play at the exact same speed (frequency) to keep the music harmonious. If one section plays too fast or too slow, the whole song gets out of tune, which can damage instruments or even stop the concert entirely.

This paper is about a new, smarter "conductor" designed to keep this orchestra in perfect time, specifically for the hydropower plants in Northwest Ethiopia.

The Problem: The Old Conductor is Too Slow

Currently, the grid uses a "conductor" called a PID controller. Think of this like a strict, old-school teacher who follows a fixed rulebook.

How it works: If the music gets slightly off-key, the teacher adjusts the volume based on a pre-set formula.
The flaw: The teacher's rulebook is static. It can't change its mind quickly if the situation gets chaotic. If a sudden demand for electricity spikes (like a sudden crowd cheering), the old teacher takes a long time to react, causing the music to wobble (frequency deviation) for nearly 50 seconds before settling down.

The Solution: A Hybrid "Super-Conductor"

The researchers designed a new, hybrid conductor called ANFIS-based Fuzzy-PID. To understand this, imagine combining two types of experts:

The Fuzzy Logic Expert: This is like a seasoned jazz musician who can "feel" the music. Instead of following rigid rules, they use common sense (e.g., "If the pitch is a little flat, nudge it up a little").
The ANFIS (Neuro-Fuzzy) Expert: This is like a student who learns by watching the jazz musician. It uses artificial intelligence to study how the system reacts to mistakes and then updates the jazz musician's rulebook in real-time.

The Hybrid Approach:
The new system works like a team where the AI (ANFIS) constantly watches the grid, learns from its mistakes, and instantly tweaks the settings of the Fuzzy Logic expert. This allows the system to adapt to changes instantly, rather than waiting for a slow, pre-set reaction.

The Experiment: A Three-Area Test

The researchers tested this new conductor on a model of the Northwest Ethiopian grid, which they divided into three "areas" (like three sections of the orchestra: Tana Beles, Fincha, and Tis Abay-2). They simulated three different scenarios:

Scenario 1 (One Section Stumbles): They simulated a small problem in just one area.
- Result: The old PID teacher took 48.8 seconds to fix the rhythm. The new hybrid conductor fixed it in just 12.9 seconds. That's a massive improvement, like going from a slow jog to a sprint.
Scenario 2 (Everyone Stumbles at Once): They simulated a huge, simultaneous problem in all three areas.
- Result: The new conductor kept the music stable with far less "wobble" (overshoot) and settled down much faster than the old system or even the other smart controllers tested.
Scenario 3 (The Weather Changes): They tested if the new conductor could handle "bad weather" by changing the system's physical parameters by 25% (simulating aging equipment or unexpected changes).
- Result: The new conductor remained stable and effective, proving it is robust and reliable even when things aren't perfect.

The Verdict

The paper concludes that this new ANFIS-based Fuzzy-PID controller is significantly better than the current system. It fixes frequency errors faster, reduces the "wobble" in power flow, and handles unexpected changes with ease.

However, the paper also notes the hurdles:
While the "music" sounds perfect in the computer simulation, actually hiring this new conductor for the real orchestra is hard.

Computational Cost: The AI brain requires a lot of computing power to think in real-time.
Human Resources: There is a shortage of engineers in Ethiopia who know how to maintain these advanced AI systems.
Infrastructure: The current communication systems might need upgrading to talk to this new smart conductor fast enough.

In short, the paper proves that a smarter, AI-driven conductor can keep the Ethiopian power grid in perfect tune, but getting it to work in the real world will require new technology and training.

Technical Summary: Hybrid ANFIS-Based Fuzzy-PID Controller Design for Multi-Area Load Frequency Control in the Northwest Ethiopian Power Grid

Problem Statement
The Ethiopian power grid, particularly the Northwest Region, relies on a conventional Proportional-Integral-Derivative (PID) controller for primary Load Frequency Control (LFC). While PID is a standard mechanism, its static gain parameters are difficult to tune in real-time, limiting its effectiveness against dynamic system disturbances and nonlinearities. As electricity demand rises, mismatches between generation and load cause frequency deviations from the nominal 50 Hz, threatening system stability. The existing National Load Dispatch Center (NLDC) monitors system status but lacks a sophisticated supplementary mechanism to centrally regulate frequency and tie-line power flow during extreme faults or significant load variations. The study addresses the need for a robust, supplementary LFC strategy capable of real-time parameter tuning to replace or augment the existing primary PID mechanism.

Methodology
The research developed a mathematical model of a three-area interconnected hydropower system representing the Northwest Ethiopian Power Grid (Tana Beles, Fincha, and Tis Abay-2). The model incorporated distinct governor dynamics for each area: a mechanical-hydraulic governor (Area 1), an electro-hydraulic governor (Area 2), and a programmable PID-type electro-hydraulic governor (Area 3).

Four control strategies were designed and simulated in MATLAB/Simulink:

Conventional PID: Tuned using the Ziegler-Nichols sustained oscillation method.
Fuzzy-PID (FPID): A hybrid where a fuzzy logic system dynamically adjusts PID gains ( $K_p, K_i, K_d$ ) based on Area Control Error (ACE) and its rate of change ( $\Delta ACE$ ).
ANFIS Controller: An Adaptive Neuro-Fuzzy Inference System using a Sugeno-type structure to learn from data and adjust control signals.
Hybrid ANFIS-Based FPID: The proposed core contribution. This architecture integrates the learning capability of ANFIS with the rule-based adaptability of FPID. The trained ANFIS output serves as the ACE input for the FPID controller, which then generates the final governor reference signal.

The controllers were evaluated using the Integral Time Absolute Error (ITAE) as the primary objective function, alongside settling time, peak overshoot, and peak undershoot. Performance was tested under three scenarios:

Scenario 1: A 0.01 p.u. load disturbance in Area 1.
Scenario 2: Simultaneous multi-area disturbances (0.46 p.u. in Area 1, 0.0268 p.u. in Area 2, and 0.016 p.u. in Area 3).
Scenario 3: Robustness analysis with $\pm 25\%$ variations in system parameters (governor time constants, water time constants, inertia, and droop).

Key Results
The hybrid ANFIS-based FPID controller demonstrated superior performance across all scenarios compared to PID, FPID, and standalone ANFIS controllers:

Single-Area Disturbance (Scenario 1): The proposed controller reduced the frequency deviation settling time to 12.959 s (compared to 48.8824 s for PID) and achieved an ITAE of 0.00565 (compared to 0.0838 for PID). Tie-line power undershoot in Area 1 was reduced from -0.00805 p.u. (PID) to -0.00339 p.u. (Hybrid).
Multi-Area Disturbance (Scenario 2): Under simultaneous large load variations, the hybrid controller maintained the lowest ITAE of 0.02375, significantly outperforming the PID controller's ITAE of 0.97080. It consistently achieved the smallest peak overshoots and undershoots across all three areas.
Robustness (Scenario 3): The system remained stable under $\pm 25\%$ parameter variations, with the ITAE remaining within a narrow range of 0.00541 to 0.00580, confirming the controller's resilience to model uncertainties.

Significance and Claims
The paper claims that the proposed hybrid ANFIS-based FPID controller offers a technically viable pathway to enhance grid reliability in Ethiopia. By combining the adaptive learning of ANFIS with the nonlinear handling of FPID, the system effectively manages uncertainties and dynamic conditions that static PID controllers cannot. The study positions this controller not as a replacement for the primary LFC, but as a necessary supplementary, centralized mechanism to be deployed at the NLDC.

The authors emphasize that this approach enables real-time adaptation to the grid's fluctuating energy landscape without manual retuning, potentially preventing blackouts and protecting infrastructure. However, the paper modestly acknowledges significant implementation challenges, including the computational burden of ANFIS, the "curse of dimensionality," and the need for upgraded communication protocols and specialized human resources at the NLDC. The authors explicitly state that the current validation is limited to simulation (MATLAB/Simulink) and does not yet include non-linear constraints like governor dead bands (GDB) or generation rate constraints (GRC), nor does it incorporate thermal or wind plants.

Conclusion
The study concludes that the hybrid ANFIS-based FPID controller significantly outperforms existing mechanisms in terms of settling time, overshoot reduction, and robustness. It recommends that the Ethiopian Electric Power (EEP) consider piloting this controller to transition toward a more resilient, centralized LFC strategy, provided that strategic investments are made in modernization and engineer training. Future work is directed toward Hardware-in-the-Loop (HIL) testing, expanding the model to include more diverse generation sources, and integrating cybersecurity considerations.

The Problem: The Old Conductor is Too Slow

The Solution: A Hybrid "Super-Conductor"

The Experiment: A Three-Area Test

The Verdict

More like this