New AI model to forecast load, reduce power wastage
Source: Chronicle News Service / Premchand Thongam

Imphal, April 04 2025: Amid the persistent power shortage and frequent load shedding in the state, a research scholar from Manipur University's Computer Science Department has developed an Artificial Intelligence (AI) model that can forecast power consumption and help optimise electricity distribution.

Kshetrimayum Nilakanta, a resident of Kakching Moirangthem Leikai, has designed the model as part of his research project titled Development of AI Model for Load Forecasting.

Nilakanta, who completed his MSc from Delhi University, is currently conducting research at Manipur University.

His AI model, designed using deep learning techniques, can predict the potential load for the upcoming week with significantly higher accuracy and efficiency than existing models.

By forecasting power consumption trends, his model could assist power companies in acquiring the required amount of electricity, preventing both shortages and over-procurement.

This would enable a more balanced and efficient power distribution system, reducing wastage and ensuring uninterrupted supply to consumers.

To train and test his model, Nilakanta relied on open-source data, primarily from the United States and Morocco.

He utilised datasets that included hourly power consumption patterns to refine the model's accuracy.

However, when he approached Manipur State Power Distribution Company Limited (MSPDCL) for relevant data, he was informed that the company does not maintain detailed hourly power consumption records.

The only available data pertained to prepaid electricity recharges, which were unsuitable for training the model.

He emphasised that if the state had maintained hourly consumption records, his model could deliver highly precise load forecasts tailored to the state's specific needs.

The AI model offers two key forecasting approaches: single-step prediction and multi-step prediction.

The single-step prediction model can estimate immediate potential load or forecast demand for the following day, while the multi-step prediction model can project power consumption trends for the next seven days.

Given that Manipur relies primarily on hydroelectric power, which often becomes insufficient during the dry season, his model could help power companies efficiently manage demand and supply, mitigating the impact of seasonal shortages.

Nilakanta's research, which concluded in December, highlights the importance of advanced technology in addressing the state's power crisis.

If adopted by the state's power authorities, his model could significantly improve electricity distribution, ensuring a more stable .power supply for both urban and rural consumers.

(This article is published under the 5th Scientific Journalism Program on 'Artificial Intelligence and Its Usage' of the Public Relations and Media Management (PRMM) Cell, Central Agricultural University, Imphal) .

Comment