frequent outages are frequently caused by an imbalance between the supply and demand for electricity. The

majority of traditional load shedding techniques are reactive, manual, and unable to adjust to changing

operating conditions. In order to optimize load shedding decisions, this paper suggests an Artificial

Intelligence (AI) method based on fuzzy logic. The approach incorporates a number of variables into a fuzzy

inference engine, such as the supply-demand ratio, system frequency, and meteorological conditions. A

MATLAB/Simulink model was created and evaluated in a variety of real-world situations, including weather

disruptions, supply shortages, and generator failure. According to the results, the AI-controlled method

outperforms classical methods in terms of frequency and voltage stability, outage duration, and response time

to disturbances. The suggested plan has a great deal of potential to improve Barishal's power system

dependability and can be expanded to other parts of Bangladesh.

equilibrium between power supply and demand. The power grid is complex and interconnected, and even minor

disruptions can trigger a cascade of events, leading to system instability[1][2]. In such critical situations, power

system operators must resort to emergency operation control strategies, including load shedding, to regain

stability and prevent system-wide disasters[3].

In Bangladesh, the lack of generation capacity has led to a shortage of electricity, affecting industrial and

agricultural growth and the country's economy. Load shedding is done to balance power demand and supply,

forcing industries and businesses to close or relocate. To address this shortage, several options are under

consideration, such as increasing generation capacity, developing renewable energy technologies, and power

system optimization[4].

Fig. 1 illustrates the map of Barishal Upazila, the main administrative and commercial sub-district of Barishal,

production of electricity using biomass. Barishal Upazila is a good candidate for the implementation of

localized renewable energy solutions and intelligent load management strategies due to its unique geographic

and resource characteristics.

Fig. 1. The map of Barishal Upazila[5]

The Barishal upazila of Bangladesh faces a significant challenge in maintaining an uninterrupted power

distribution network. Load shedding, a practice where power is intentionally cut off due to fluctuations in

demand and supply, continues to be a persistent issue, causing disruptions and inconveniences for residents and

adversely impacting local businesses. Traditional methods for load shedding management often fail to adapt

swiftly to the dynamically changing energy landscape, resulting in suboptimal outcomes[6].

Conventional load shedding relies on predetermined thresholds or operator intervention[9]. While effective in

preventing total blackouts, these methods are reactive, rigid, and often lead to unnecessary outages.

Studies in Pakistan and Libya have explored AI-based control systems, employing neural networks, particle

swarm optimization, and fuzzy logic for load management. For example, Alarbi (2019) used AI to reduce

customer inconvenience during shedding, while Alamri (2020) focused on ANN-based load optimization in

Pakistan’s grid[1][2]. In microgrids, fuzzy logic has been successfully applied to integrate renewables and

balance fluctuating demand[8][10][11].

Despite these advances, minimal research has focused on regional applications within Bangladesh. Barishal,

with its chronic electricity shortfall, presents a critical case where intelligent load shedding can enhance system

reliability and minimize consumer disruption[12][13]. This study addresses that gap.

The proposed AI-based load shedding system consists of three major modules. Here, Fig. 2 illustrates the

flowchart of the proposed system architecture. System block diagram of the AI-based load-shedding controller

showing measurement inputs, fuzzy inference block, and feeder CB outputs

Fig. 2. The flowchart of the system architecture

Fig. 3. Flowchart of fuzzy logic controller

imbalance, and Weather accounts for demand surges and renewable generation variability. Together, these

variables enable the controller to make context-aware, priority-based load-shedding decisions.

1) Input variables:

 Supply-load ratio (0.5 — 2.0). Fig. 4 is a graph showing the fuzzy membership functions of the supply-load

 Frequency (49–51 Hz). Fig. 5 shows the fuzzy membership functions of frequency for three categories: poor,

Fig. 6. The graph of weather vs membership

2) Output Variables:

 Load shedding level (None, Low, Medium, High). Fig. 7 illustrates the fuzzy membership functions of load

Fig. 7. The graph of load shedding vs membership

Membership functions were defined for each variable, and a rule base was constructed (e.g., if frequency is low

and demand is high, Then apply high load shedding)[15].

levels.

Simulation Setup

The fuzzy logic controller was implemented in MATLAB/Simulink. To facilitate the integration of our Fuzzy

Logic-based load shedding system with the power grid infrastructure, we create interfaces and connectors that

enable seamless data exchange and communication between the AI system and grid components. This includes:

 Establishing protocols for bidirectional communication, allowing the system to receive real-time grid status

Sensor data and IoT devices play a pivotal role in real-time grid monitoring and load shedding control. Our

methodology involves:

diagram. The diagrams illustrate different levels of complexity in control systems, from a basic fuzzy to an

advanced AI-driven solution.

Fig. 8. Controller for the system

Fig. 9. AI load shead controller system and model

#!/usr/bin/env python

# coding: utf-8

# # Import Library

time = str(current_time_seconds)

freqs.automf(3)

weathers.automf(3)

sheds.automf(3)

# # View Membership Function

sl_ratios.view()

rule3 = ctrl.Rule(sl_ratios['good'] | freqs['average'] | weathers['good'],sheds['poor'])

sim.compute()

load_shed = sim.output['Load Shedding']

sheds.view(sim=sim)

load_shed = round(load_shed, 1)-0.2

S_load = load_shed*load

load shedding results was investigated by evaluating the suggested fuzzy logic-based controller under various

operating conditions.

Table 1 Comparison of AI-Fuzzy vs. Classical Load Shedding[18]