Skip to main content

Advanced control advice for power systems with large scale integration of renewable energy source

Exploitable results

The objective of this project is the development of CARE, an adaptable advanced control software that achieves optimal utilization of renewable energy sources, in medium and large size isolated systems with diverse structures and operating conditions by advising operators of possible actions. The insurance of increased security and reliability of the system will allow maximization of renewables penetration. The software includes various modules for: Forecasting Load and Renewables (mainly wind) Forecasts are required for both short (several minutes up to several hours ahead) and longer time scales (of the order of several hours up to few days ahead), in order to provide input to the economic dispatch and unit commitment algorithms. A number of modules based on auto-regression, fuzzy neural networks and simpler methods have been developed. Operational Planning Unit Commitment functions to advice operators about possible switching on or off of one or more thermal units in the next time step based on load and renewables forecasts and operating status of units and taking into account stochastic nature of wind power, uncertainties in thermal generation, etc. Genetic Algorithms and combinatorial techniques have been applied. In addition, on-line Generation Dispatch functions determine the power output of each generating unit (both thermal and renewables) aiming to minimize operation cost, while satisfying operational constraints. Evolutionary Programming, Genetic Algorithms and Optimization Techniques, like Constrained Linear programming, have been applied to Generation Dispatch. Security Assessment Security rules and security functions are required for the detection of insecure dispatching recommendations to the operator (preventive mode) and for on-line security monitoring of pre-specified, probable disturbances. Advanced artificial intelligence techniques, like Decision Trees, Kernel Regression Trees and Artificial Neural Networks have been used to achieve the essential on-line performance. The above modules are integrated in the CARE package providing optimal advice to isolated power system operators via a user-friendly Man-Machine interface. A prototype of CARE has been interfaced to the on-line SCADA Data Base of the Control Center of Crete. Initial evaluation of this pilot installation has shown satisfactory forecasting results, clear economic gains provided by the economic dispatch advice, timely and accurate assessment of dynamic security. The modular configuration of the CARE software and the standard hardware used enable further installations in other isolated networks in Europe and in developing countries.