Applying artificial intelligence in energy systems
AI for power system operation and planning
The amount of distributed and intermittent energy production is increasing which changes the operation and planning of the energy system drastically. In the future flexible and consumer-centric energy system, also small-scale resources, need to be monitored and controlled in real-time. Also more accurate load and production forecasting, including control response forecasts, is a necessity. Overall the reliability requirements for electricity supply are increasing.
More flexibility and efficiency
Automated AI-assisted systems can be utilized for many purposes in different operational layers of the power system. For instance the following solutions have been developed at VTT:
- At building level, smart algorithms optimize the operation of the whole building taking into account different aspects such as maximizing self-consumption of produced energy, providing optimal comfort level to users and/or participation to whole system operation through different markets. Machine learning is also utilized for generating accurate forecasts on the behavior of the energy consumption profiles of the building and devices against environmental variables
- In power plants, machine learning-based fault detection and isolation algorithms can be used to quickly identify the fault location and to reduce the outage time
- At district level, AI algorithms are used for optimal smart city and district energy planning
- At system level, AI algorithms utilize data from a multitude of data sources and provide accurate short- and long-term load and production forecasts and also forecasts for their control responses
Following benefits can be achieved by the efficient usage of automated AI-assisted systems;
- Enhanced building automation systems enable utilization of new flexibility in power system operation without compromising user comfort level and technical constraints (e.g. in stores cold chain has to be maintained)
- Better operation of energy production units
- Cost savings and lower level emissions in district level
- More accurate load and production forecasts enable improved utilization rate of power system infrastructure and more efficient integration of renewables to the system
AI for better utilization of energy system data
The amount of data from the energy system is already now very large and increasing rapidly due to e.g. smart meters and low cost of IoT devices. This data is not utilized efficiently at the moment. There is a need to utilize the large amount of data that is already now available from different parts of the energy system more efficiently e.g. for preventive maintenance purposes or for development of smart control strategies.
AI technologies are used to find the essential information from large data sets. Smart meter data can be used to improve load profiles which are used for network planning purposes. Load measurements can be used for identification of different types of loads at individual buildings. On-line and off-line measurement data can be used for preventive maintenance purposes.
More accurate load profiles enable better network planning while improving the utilization rate of the network. Preventive maintenance enables better selection of components to be overhauled or replaced so that cost savings and better reliability is reached.