The power transmission companies are adopting digitally empowered technologies such as artificial intelligence (AI) and machine learning (ML) to unlock efficiency gains and economic benefits. The technologies will increase grid reliability by augmenting the grid stabilisation competencies of the transmission companies; provide real-time diagnosis and analysis of demand-supply dynamics, giving greater insight into the future energy flows; and improve grid security. AI and ML technologies improve the reliability and predictability of grid operations and maintenance.
Reliability centred management
Reliability centred management (RCM) is part of the digital strategy of transmission companies. It involves AI and ML technologies to enhance the reliability of the grid system. The strategy focuses on the prevention of cumbersome repair and maintenance activities with the help of real-time diagnosis, and advanced repair of faults before they evolve into something dangerous, hampering the integrity of the grid. The AI and ML technologies can be integrated with data, which will allow them to model, estimate and predict the eventuality of such incidents over the long term.
The other school of reliability management called condition-based management focuses on regularly checking on vital and vulnerable components such as towers, foundation revetments, and current and potential transformers. AI and ML incorporated in SCADA systems can be employed to derive analytical insights. This will help the transmission companies maintain a defect-free infrastructure and prioritise repairs for defective and critical components.
Emergence of renewable technologies
The rising share of renewable technologies is making it necessary for transmission companies to go digital and integrate AI and ML in their systems to manage the yield unpredictability, intermittency issues and ramping up issues associated with wind and solar power. As per several estimates, renewable technologies will contribute to 30-40 per cent of electricity requirements by 2030.
The existing grid assets and transmission systems will not be able to manage and facilitate the growth of renewable technologies. The incorporation of MI and AL into the technological systems will help in improving the precision of demand-supply predictions and addressing issues that may arise from such volatility.
The inclusion of AI and ML will also enable the transmission companies to transform their traditional operations and become lean enterprises. It has been observed that planning and modelling by transmission companies is off the mark. They schedule surplus electricity to meet their requirements in order to avoid blackouts and grid instability, which are likely to be prompted by sudden and unanticipated demand for electricity. AI and ML will improve the precision of transmission companies and therefore reduce excessive electricity acquisition by them.
The application of long short-term neural networks and ML can be useful in determining the potential renewable electricity generation by taking into account wind, temperature, sunlight and humidity forecasts. Similarly, it is possible to feed historical data into machine learning algorithms like support vector machines (SVMs) to accurately forecast energy usage and ensure continuous supply.
The electricity system in India suffers from huge aggregate technical and commercial (AT&C) losses of 15-20 per cent. A substantial portion of these AT&C losses are attributable to theft or fraud. Using machine learning, the transmission companies can identify the bad actors that are defrauding and tampering with the grid. ML helps in detecting such anomalies and unusual spikes, and distinguishing it from noise.
Application of AI and ML technologies
There is immense opportunity to enhance efficiencies and reliability by integrating AI and ML capacities into digital technologies such as geographic information systems (GIS), which combines hardware, software and data to furnish an exhaustive and in-depth picture of the utility status.
Other applications of AI and ML include trip management systems connected to SCADA and asset supervision of components and their conditions by using models integrated with AI and ML to understand them better. The competencies of these systems are amplified and thus transmission companies are able to predict the stresses that will arise in the system by modelling and extrapolating the available data.
Watson, an AI-system pioneered by IBM Corporation, is installed by default with information on engineering and equipment standards, as well as industry regulations. It imports the data from EAM (enterprise asset management) systems of transmission assets to obtain the point-of-impact insights, understanding of technical issues and aggregated data views. These analyses permit the plant management to identify the right course of action to preclude any incidents.
Furthermore, combining SVM with discrete wavelet transformation to observe transient voltage can help determine the fault location by calculating the eaerial mode voltage wavelets (for above ground transmission wire) and ground mode voltage wavelets (for in-ground transmission wires). So far, this method has successfully detected fault inception angles, fault locations, loading levels, and non-linear high-impedance faults for both aerial and underground transmission lines.
Some of the micro drones deployed by transcos and discoms weigh less than 2 kg, and have a flying capacity of below 200 feet. They are equipped with an integrated thermal vision camera to render infrared radiations, LiDAR (light detection and ranging) technology to measure distances with the use of laser lights, a high resolution camera for electrical asset inspection, monitoring and mapping, and a GPS-enabled autopilot system guided by a ground control station.
These visualisation technologies integrated with AI help in vegetation management and minimise right-of-way issues for the commissioning of a new project or expansion of an existing one. Furthermore, AI and ML technologies can be employed to predict the future developments of vegetation around the transmission infrastructure and effectively preclude any trespassing and land acquisition issues.
The way forward
AI and ML have the potential to significantly transform the way power is generated and distributed. These can be used across the value chain, based on the needs of the utilities and the desired outcomes. AI and ML modules could be developed based on analytics of the data collected by sensors installed in the system. AI and ML can help turn information into actionable insights that can help predict network failures, plan timely interventions and avoid customer interruptions.
With inputs from a presentation by Satish Talmale, COO, IndiGrid, and Deepak Desai, Partner and Cognitive Business Decision Services Leader, IBM India/South Asia, at a recent Power Line conference