Optimization of Power System Reliability and Energy Management via an Enhanced Firefly Algorithm

Optimization of Power System Reliability and Energy Management via an Enhanced Firefly Algorithm

Osadebamwen Kenneth Ojo (1), Onyegbadue Ikenna (2), Ehimare Osadebamen Airiohuodion (3)

(1) Department of Electrical and Electronic Engineering, Edo State Polytechnic Usen, Edo State, Nigeria

(2) Department of Electrical and Electronic Engineering, Igbinedion University Okada, Edo State, Nigeria

(3) Department of Computer Engineering, Edo State Polytechnic Usen, Edo State, Nigeria

Abstract – Modern electrical power grids face escalating complexities due to fluctuating demand profiles, high penetration of volatile renewable energy resources, and stringent operational constraints. To address these challenges, this paper presents an integrated framework for power-system optimization, reliability enhancement, and energy management using a nature-inspired meta-heuristic approach. We leverage the Firefly Algorithm (FA) to simultaneously minimize operational costs, mitigate transmission power losses, and maximize the Loss of Load Probability (LOLP) reliability indices. The proposed optimization engine is mathematically formulated as a non-linear constrained problem and evaluated on a simulated IEEE 30-bus test system configured with hybrid generation assets (conventional thermal, wind, and solar photovoltaic systems). Available results demonstrate that the FA optimization model consistently locates global minima without premature convergence. Comparatively, the FA achieves an 8.4% reduction in overall fuel costs and a 12.1% reduction in active power losses relative to conventional Economic Load Dispatch (ELD) models. Crucially, the system reliability profile was successfully enhanced, keeping the system risk metrics safely within utility standards while maintaining an optimal battery energy storage state-of-charge. The findings confirm that FA is a highly robust, computationally efficient, and scalable solution for modern energy management architectures.

1. INTRODUCTION

   Modern electrical utility networks are transitioning from predictable centralized architectures to highly dynamic decentralized grids. This evolution is driven by the global imperative to integrate clean energy technologies, reduce greenhouse gas emissions, and optimize fuel expenses. However, integrating intermittent renewable energy sources like wind and solar PV introduces severe technical challenges regarding power system stability, reliability, and economic scheduling.

   Optimal Power Flow (OPF) and Energy Management Systems (EMS) represent the foundational computational layers required to handle these grid complexities. The fundamental objective of an EMS is to schedule generating units economically while operating within security constraints, keeping transmission network losses low, and guaranteeing uninterrupted service to end- users. Classical mathematical optimization techniques, such as Interior Point Methods and Successive Linear Programming, scale poorly when dealing with non-convex, non-linear, multi-modal search spaces populated with complex discrete constraints (like transformer tap positions and battery degradation bounds).

   To overcome these deterministic limits, meta-heuristic algorithms have emerged as powerful alternatives. Algorithms like Particle Swarm Optimization (PSO) and Genetic Algorithms (GA) have been widely deployed, yet they frequently suffer from premature local convergence when handling highly correlated variables. Introduced by Dr. Xin-She Yang, the Firefly Algorithm (FA) offers a compelling solution because it excels at multi-modal optimization by utilizing an attraction mechanism based on luminescence. This journal article provides a holistic optimization paradigm for a hybrid power network utilizing the Firefly Algorithm. The target is an integrated model optimizing three core pillars simultaneously: economic dispatch (energy management), transmission loss reduction, and system risk mitigation (reliability).

2. MATHEMATICAL PROBLEM FORMULATION

   The optimization problem is formulated as a mixed-integer non-linear programming (MINLP) problem containing multiple objective functions subjected to diverse equality and inequality constraints.

   1. Objective Functions

      1. Economic Dispatch Cost Minimization (F1):

         The total generation cost includes the quadratic fuel costs of conventional thermal units coupled with the operational expenses of renewable generation and storage systems:

         1 = (2 + + ) + ,() + ()

         = =1

         Where ai, bi, ci are the cost coefficients of the i-th conventional generator; PGi is the active power output of the thermal generator; CR,j and PR,j represent the cost models and power outputs for wind/solar assets; and CB(PB) tracks the operational battery degradation or storage usage cost.

      2. Minimization of Real Power Losses (F2):

         Transmission system active losses are minimized to maximize network efficiency:

         2 = {} = [2 + 2 2( )]

         Where is the conductance of line k connecting bus m and bus n, V represents bus voltage magnitude, and represents bus voltage angle.

      3. Reliability Index Enhancement (F3):

      To improve grid structural reliability, the objective function minimizes the Loss of Load Probability (LOLP) across the optimized schedule:

      F3 = LOLP = Probability(PGeneration,t + PStorage,t < PLoad,t)

   2. System Constraints

      The objective landscape is tightly governed by the following constraints:

      • Active and Reactive Power Balance (Equality): PGi + PRenewable + PBattery – PLoss – PLoad = 0

      • Generation Unit Operating Bounds (Inequality): PGimin PGi PGimax

      • Voltage Profile Security Limits: Vmmin Vm Vmmax

      • Battery State of Charge (SoC) Boundaries: SoCmin SoC(t) SoCmax

3. THE FIREFLY ALGORITHM ARCHITECTURE

   The Firefly Algorithm acts as our global optimizer. It assumes that all fireflies are unisexual, meaning they attract each other regardless of sex. The core mechanics rely on two foundational criteria: Light Intensity and Attractiveness.

   1. Light Intensity and Attractiveness

      The light intensity (brightness I) of a firefly is linked directly to its objective function value: I(x) 1 / Objective Function(x). Because light dampens over distance due to air absorption, the operational attractiveness varies dynamically with the relative

      ij 0 0

      cartesian distance r between two fireflies i and j: (r) = 2, where is the base attractiveness at distance r = 0, is the fixed light absorption coefficient, and rij is the Euclidean distance between vectors xi and xj.

   2. Movement Step

      When a firefly i is attracted to a brighter, more optimal firefly j, its change in position is computed as follows: -(+1) = () +

      2( ) + . . The second term captures the pull toward the brighter solution, while the third term adds

      0

      randomized exploration via a scaling factor and a random vector i drawn from a normal or uniform distribution.

      Algorithm 1: Firefly Engine for Power System Management

      Input: Objective Function (F1, F2, F3), System Bus Constraints Output: Optimal Dispatch Vector(PGi, P_Battery, Loss_Minimum)

      1: Initialize a population of fireflies x_i (i = 1, 2, …, n)

      2: Evaluate light intensities I_i at x_i via objective function mapping 3: Define light absorption coefficient () and random scaling factor () 4: while t < Maximum_Generations do

      5: for i = 1 to n do

      6: for j = 1 to n do

      7: if I_j > I_i then

      8: Calculate spatial Euclidean distance r_ij

      9: Update attractiveness using Equation 0*exp(-*r^2)

      10: Move firefly i toward j using the Movement Step formula

      11: Evaluate new position constraints & penalty functions

      12: Update light intensity I_i 13: end if

      14: end for

      15: end for

      16: Rank fireflies and identify the current global best option 17: t = t + 1

      18: end while

4. CONFIGURATION AND SETUP

   To evaluate the proposed algorithm, a standard IEEE 30-bus test system was simulated within a MATLAB/Python workflow. The basic architecture consists of 6 thermal generators, 24 load buses, and 41 transmissions interconnect.

   To simulate a modern hybrid grid environment, we introduced the following modifications to the test framework:

   1. Renewable Integration: Bus 2 and Bus 5 were integrated with a 15 MW Wind Farm and a 10 MW Solar PV Plant, respectively.

   2. Energy Storage System (ESS): A 5 MWh Lithium-ion Battery Bank was connected to Bus 6 to mitigate renewable volatility and smooth peak load demand.

   3. Algorithmic Hyperparameters: Population size n = 40; Maximum generations Tmax = 500; Light absorption coefficient = 1.0; Randomization parameter = 0.25; Base attractiveness 0 = 1.0.

   [Schematic Layout of the Modified IEEE 30-Bus Hybrid Test System Grid Infrastructure]

   Figure 1: Modified IEEE 30-bus network single-line topology showing integrated wind, solar PV, and ESS battery allocations.

5. RESULTS AND DISCUSSION

   The performance of the Firefly Algorithm was evaluated against an unoptimized base case, standard Particle Swarm Optimization (PSO), and a Real-Coded Genetic Algorithm (RCGA).

   1. Convergence Analysis

      The convergence profiles demonstrate that the Firefly Algorithm exhibits faster convergence rates and superior global searching metrics compared to PSO and RCGA. Thanks to its localized subgroup attraction mechanism, FA escapes local minima early in the optimization run. This avoids the premature flatline behaviors often seen in traditional genetic implementations.

   2. Parametric Performance Comparison

      The core performance metrics across 50 independent simulation runs are summarized in Table 1 below:

      Metric Evaluation Parameter	Base Case (No Opt.)	RCGA Framework	PSO Framework	Proposed FA
      Total Generation Cost ($/h)	912.45	864.20	848.15	835.80
      Active Power Loss (MW)	17.34	16.12	15.54	15.24
      Minimum Voltage Profile (p.u.)	0.921	0.954	0.962	0.981
      Loss of Load Probability (LOLP)	0.042	0.015	0.011	0.004
      Mean Computation		14.82	9.35	6.12

      Time (s)

      Table 1: Comparative operational simulation performance metrics matrix.

   3. Reliability and Energy Storage Behavior

      When managing high penetration of solar and wind generation, the FA engine utilizes the battery system to balance the grid. During periods of excess renewable production, the optimizer directs excess power into the battery storage system while satisfying voltage boundary limits. During peak demand intervals, the FA draws power from the battery storage bank rather than cycling expensive thermal units online. This energy management approach improves grid reliability, dropping the system’s LOLP value to a highly secure 0.004.

6. CONCLUSION

This study developed a complete optimization infrastructure for hybrid power grids using the meta-heuristic Firefly Algorithm (FA). The algorithm managed a complex search space to minimize production costs, reduce power transmission losses, and maximize system reliability under volatile renewable conditions.

Simulated results on the modified IEEE 30-bus test system confirm the proposed method’s engineering validity. The Firefly Algorithm achieved an 8.4% reduction in hourly fuel expenses and a 12.1% reduction in active power losses while ensuring a reliable, well-regulated voltage profile. Future research will focus on extending this FA framework to large-scale, multi-area smart grids and evaluating its performance under real-time hardware-in-the-loop (HIL) operational settings.

REFERENCES

1. Yang, X. S. (2008). Firefly Algorithms for Multimodal Optimization. Stochastic Algorithms: Foundations and Applications.

2. ResearchGate (2015). Firefly algorithm based solution to minimize real power loss in a power system transmission network.

3. National Center for Biotechnology Information (2024). Using firefly algorithm to optimally size a hybrid renewable power system with battery degradation tracking.

4. Egyptian Knowledge Bank (2021). Optimal Power Flow Using firefly-algorithm on standard IEEE configurations.

5. Journal of Operation and Automation in Power Engineering (2023). Hybrid Energy Systems Optimization Using Firefly Algorithm considering system uncertainty parameters.

6. Semantic Scholar (2018). Reactive Power Optimization and loss minimization models utilizing firefly heuristic parameters.