This work presents an identification method for neuro-fuzzy models that not only allows models grow in function of the existent complexity but that beforehand they try to self-adapt to avoid the inclusion of new rules.
In general, that growing occurs in function of redundant rules or rules that cover a very little domain of the problem. Unfortunately, neuro-fuzzy models can grow up to become of hard interpretation because of the complexity of the systems under modelling. This is a very intuitive human form of interpretation and neuro-fuzzy model are gathering increasing acceptance. Neuro-fuzzy models have been used for the last years in a attempt of suppress these difficulties because of their ability of modelling in linguist form even a system which behavior is extremely complex. This difficulty does not happens only in nuclear power plants but in almost every a little more complex control system. Nevertheless their operation be automatic most of the time, the integral understanding of their internal- logic can be away of the comprehension of even experienced operators because of the poor interpretability those controls offer. Neuro-fuzzy models for systems identification applied to the operation of nuclear power plants Sistemas neuro-fuzzy para identificacao de sistemas aplicados a operacao de centrais nuclearesĮnergy Technology Data Exchange (ETDEWEB)Ī nuclear power plant has a myriad of complex system and sub-systems that, working cooperatively, make the control of the whole plant.
0 kommentar(er)
