A small team of information technology engineers of Tangedco is engaged in the production of daily reports of energy demand prediction through artificial intelligence (AI) in a bid to automise the process.
The AI team competing with the those engaged in manual prediction is attempting to have a better reading of the electricity demand pattern in maintaining the grid management. The energy demand prediction report through AI has been started by the IT officials from November 1 and is being perfected to have minimum of errors in the data crunching. The daily energy data helps in the procuring of power for maintaining uninterrupted supply in the city during evening peak hours and also to meet the agricultural demand throughout the State.
A pet project of Chairman and Managing Director Rajesh Lakhoni, it follows the path of the demand data already in operation in the electricity department of Andhra Pradesh (AP).
A senior official of the State Load Despatch Centre (SLDC), which is the department engaged in round-the-clock monitoring of the grid management throughout the State, said daily prediction of energy demand was done manually by compiling the weather and consumption patterns based on the energy data available for several years. However, a small team was formed based on the direction of Mr. Lakhoni to automise the daily prediction report for which Tangedco entered into a knowledge sharing agreement with the electricity department of Andhra Pradesh.
As part of this venture, the electricity department has tied up with the Tamil Nadu Agricultural University for three years for procuring daily and weekly weather data through the network of automatic weather stations maintained by them throughout the State. The IT officials, using the daily weather data and the daily record of consumption pattern available with the SLDC for several years, have created an AI software for data crunching to produce the daily demand report to be available in excel sheets.
The senior electricity official said the correlation between the manual data, automated data and actual consumption, was also being studied wherein slight variations had been found. The margin of error in manual data against actual demand remains at 3-4% whereas the automated data remains at 5-7% during a particular time with actual demand. Once the margin of error comes down, the automated data could play major roles in not only daily demand prediction but also in making power purchase agreements of short- and long-term every year, the electricity official added.
The AI team which is at present using an open source software plans to migrate to fool proof software to protect the reports.
