Artificial intelligence (AI) is widely regarded as the key to unlocking the door to the fourth industrial revolution. Every industrial revolution implies a dramatic change in the productivity of society, and the fourth industrial revolution is bound to have another widespread and far-reaching impact on the way humans produce and live, one that is even disruptive.
Food, energy and water are the foundations of human survival and development, and under huge population pressure, the food, energy and water crises are constantly impacting the beautiful home we all share.
David Wallerstein, Tencent’s Chief Exploration Officer, has confronted the global crisis with a new concept and explored in depth the prospects and pathways of AI technology in solving the three major problems of food, energy and water, writing the book “Reframing the Earth: AIFORFEW”.
The application of artificial intelligence technology in energy is just beginning at the moment, but it has unlimited potential and broad prospects for development. The issue of energy sustainability is the ultimate challenge shared by humanity, and AI technology becomes a powerful weapon in our fight against these challenges.
AI and big data combine to “catch the wind” of the grid
According to the book “Reconstructing the Earth: AIFORFEW”, AI technology can be used to better guarantee the consumption of renewable energy and to tap into the potential of new energy development. Along with the integration of technology, AI technology and big data technology have been used in the field of renewable energy consumption, and have achieved significant results.
In recent years, with the digitalisation and information technology of power grids becoming more and more mature, data from all aspects of generation, transmission, distribution and use have been effectively collected, and the amount of data is increasing day by day. However, in the face of massive amounts of power data, traditional technology has been unable to meet the data processing needs, not to mention the value of data mining, so AI technology and big data technology came into being.
The combination of AI and big data technologies can integrate advanced sensing and measurement technologies, information and communication technologies, analysis and decision-making technologies, automatic control technologies and grid infrastructure to drive changes in the grid system.
In addition to real-time monitoring and testing of the grid and ensuring the safe operation of the system, AI can further tap into historical and real-time data, facilitating grid diagnosis, optimisation and forecasting, improving the control and resource optimisation of the grid, and uncovering the laws of grid operation, thus ensuring the safety, reliability and economy of grid operation.
The combination of AI and big data can also facilitate renewable energy access and prepare for clean energy development. Due to the lack of system-level consideration, the forecast error of the current traditional renewable energy generation forecasting method exceeds 50% in some periods; at the same time, the large forecast error often makes it difficult for the system to fully track the renewable energy generation output, thus making it difficult to guarantee the effective consumption of renewable energy.
The AI technology and big data technology are based on four categories of dynamic and static wind power data resources: regional observation data, unit operation data, meteorological data and geographic information data. Using mature technologies such as Hadoop, data mining algorithms and behavioural analysis frameworks are built for a massive data environment to achieve terabytes and petabytes of big data processing capacity, with prediction accuracy reaching over 90% in coastal areas and The prediction accuracy is over 90% in coastal areas and around 80% in inland provinces. In terms of photovoltaic power output prediction, based on big data to establish a power prediction model for photovoltaic power plants, linked with numerical weather forecasting, the impact of sand, dust and haze on solar radiation can be finely considered, with ultra-short-term prediction accuracy reaching 95%, short-term prediction accuracy reaching more than 90%, and medium- and long-term prediction accuracy above 80%.
AI technology helps intelligent operation and maintenance based on knowledge mapping of grid faults
Reconstructing the Earth: AIFORFEW points out that AI technology can contribute to intelligent operation and maintenance based on knowledge mapping of grid faults. In fact, in response to the increasingly complex grid operation situation, dispatch operation control has been transformed by using intelligent technology to support the whole process before, during and after the event.
By distilling experience into knowledge and forming a fault knowledge map and combining it with AI technology, control staff can actively, quickly and comprehensively grasp key information on fault handling and provide corresponding auxiliary decisions for fault handling, thus effectively controlling the occurrence and development of grid accidents.
With the acceleration of technology and power reform process, energy-using enterprises have become more aware of their own energy management. At the same time, the rise of new energy sources and energy storage on the user side has made many energy users no longer just consumers of energy, but also suppliers of energy.
Intelligent electricity distribution monitoring systems and energy management systems enable monitoring, maintenance and optimisation of customer-side energy systems and reduce energy costs, while demand-side response and further reduction of customer energy costs can be achieved in response to grid demand or tariff signals. Some studies have pointed out that simply by using energy users to visualise and understand the energy use of an enterprise, energy users can make 10% of energy saving optimisation decisions.
On the demand side, demand-side management, one of the key functions in a smart grid, allows customers to reduce peak load demand on their energy suppliers and reshape the load curve. This enhances the sustainability of the smart grid and reduces overall operating costs and carbon emission levels. Most of the existing demand-side management strategies in traditional energy management systems use system-specific technologies and algorithms that can only handle a limited number and a limited type of controllable loads.
Smart devices now provide the basis for an accurate grasp of customer-level loads. By combining customer load sensing to tap into the flexibility of loads under the electricity market, the scope for flexible regulation can be increased.
AI technologies such as Hidden Markov Models, clustering algorithms, genetic algorithms and machine learning have excellent applications in load identification, multi-user coordinated control and peak staggering control. In addition, in the context of the evolving electricity market, it is also possible to use market instruments to make some users actively cut or increase their loads instead of regulating the output of conventional power sources, thus smoothing out changes in generation-side output and enabling the optimisation of system scheduling operations through demand-side management.
At the level of predictive analytics, energy suppliers need to forecast demand changes, system overloads and possible failures as accurately as possible, as the cost of error in the energy sector is very high and there is an urgent need for energy suppliers to improve their predictive analytics in order to reduce costs, save electricity, improve the use of renewable energy and prepare for the changing environment in order to provide better services to customers.
