When Artificial Intelligence & Machine Learning will enter into the distribution sector through the smart meter window

Power sector in India is going an unprecedented energy transition where the energy mix of the country/World is changing i.e.:-

Generation - from Synchronous to Non synchronous,

Power flow- from unidirectional to bidirectional,

Resiliency from- stable to highly intermittent,

Storage requirement-from zero storage requirements with conventional mode of generation to Energy storage requirements with Renewables Energy(RE).

As the Changes in the Grid are must required for lowering the emission of traces gases with high GWP(Global Warming Potential) such as CO2,CH4,N2O,CFC such that the global target for limiting the temperature growth can be restricted to 1.5 degree Celsius. Now with the increasing penetration of RE power in the grid that consist majorly of Solar & Wind where 500GW is the target for 2030 compared to 132GW today, so that Farsightedness of any grid turbulence will become a new challenge and if that can be catered with supply and load side data need to be studied and worked upon.

As the supply side data i.e. generation is available in the country at EHV to 11KV network but the load side data is still a challenge as only monthly billing data is available for most of the consumers. Now with implementation of smart meter the data will be available but the approach to handle the data will be a new challenge some aspects for the need of AI/ML for the meter data are as under :-

1- With the installation of 180 million smart meters expected in India in next 3 years the power sectors specially the Discom will be flooded with data. This data will be real time in 15 min time block and can be very helpful when scheduling load demand forecast, theft control, load analysis of particular segment of consumer be it on category wise, area wise, high loss pockets or high load pockets is concerned.

2- Only smart meter installation cannot solve the above problems it needs to be worked upon with AI or ML so that a trend is understood and actions are taken in form of Regulations/Standards to avoid any grid disturbance.

3- With AI/ML in smart meter data Discom can start promoting the RTPV /EV in target areas such as high load/high loss area with giving some additional incentives so that the losses in that particular division can be lowered.

4- Net metering and other incentive schemes of State Governments & Central Government, for Roof Top Photo Voltaic (RTPV) is growing at a much higher rate which results in bidirectional power flow into the grid that was initially built for unidirectional power flow, so system redevelopment with new bidirectional protection and other requirements can be checked on with the real time fault data of smart meters and with AI/ML the FAR (fault analysis Report) may result in system improvements.

5- The high penetration of solar/EV will result in high harmonics generation that will affect the transformer life as the transformer is designed against the rated frequency and slight change in frequency may abruptly increase the eddy and hysteresis loss creating other path to bypass the energy either it is humming sound and heating the coil and the heating that happened due to harmonics can reduce the transformer inter turn insulation and finally may result in failure of transformer in due course of time. The harmonics injection form the grid tied inverters can be continuously monitored and logic can be developed in the system with AI/ML in smart meter data so that harmonics are not injected into the grid.

6- P2P (peer to peer) trading that is released by U. P. Regulatory Commission is also going to be launched by Delhi Electricity Regulatory Commission (DERC) as the Draft of the regulation has already been released. These P2P logics needs to be developed in a blockchain that required mandatory smart meters and discom. AI/ML in smart meter data is required to avoid any system misuse by the consumers or the software provider.

7- AI/ML will be required for estimating the battery energy storage in a particular Discom, be it on LT side or HT side of the transformer.

8- Apart from quantifying the requirement of Battery Energy Storage System (BESS), AI/ML will also be informing the type of application required in BESS for a particular distribution transformer be it Energy shift mode, Valley filling mode, peak shaving mode so that Capex is deferred by making a flat load curve and enhancing the life of transformers and system.

9- With the help of AI/ML in smart meter data, Discom behavioral energy efficiencies scheme/programme can be developed and deployed where the consumers will be motivated for the use of energy efficient equipment's and reducing its energy bills hence promoting the Demand side management.

10-Behind the meter, battery is going to play a critical role in growth of RTPV as presently the prosumer are exporting/importing power as per there requirements and that may or may be as per Discom requirements in a particular time block. Hence, a control arrangements need to be developed with AI/ML so that Discom gets the support from the battery that can be in form of power at the time of peak price in the energy market or at the time of grid peak demand.

11- Reactive power support can be designed with the help of AL/ML with the smart meter data as the inverters are presently designed to export only active power and any reactive power need is taken care by the grid, hence this become a challenge for the Discom to serve reactive power even when the RTPV plant are in export mode.

12-In transition of Discom to DSO(Distribution system operators) the data logic need to be built that comes from smart meter so as to adhere the Virtual power plant, or participating the consumer in the discom peak where the consumer can get incentive apart from normal TOD (Time of the day benefits)