An artificial immune system with continuous-learning for voltage disturbance diagnosis in electrical distribution systems

• We propose a new artificial immune algorithm with continuous-learning.
• This means to learn without reinitialization when occur new disturbances.
• Application to voltage disturbance diagnosis in electrical distribution systems.
• The results show robustness, precision and efficiency.

This paper presents a new artificial immune algorithm with continuous-learning, which is inspired by the biological immune system, to realize the voltage diagnosis in electrical distribution systems. This conception allows one to compose a diagnosis system that can continuously learn without reinitialization when new disturbances occur due to the evolution of the electrical system. Two artificial immune algorithms, which are the negative selection algorithm and the clonal selection algorithm, are used for the pattern recognition process and the learning process, respectively. The principal application of this new method aids the operation during failures, supervises the protection system, and can evolve with the power systems to continuously acquire new knowledge. This new methodology has a direct impact in the area of diagnosis in electrical systems, as well as, in the pattern recognition problem, because the main contribution and novelty of this method is the continuous learning capability, which enables the system to learn unknown patterns without having to restart the knowledge. This is the major advantage of this methodology. To evaluate the efficiency and performance of this new method, failure simulations were performed in a real distribution system with 134 buses using the EMTP software. The results show robustness and efficiency.

Figure optionsDownload as PowerPoint slide