Use of a Fuzzy Neural Network to Evaluate the Cable Lines Insulation State

The article discusses the issues of monitoring and assessing the cable lines insulation state. Two approaches to the organization of cable line service are presented. The benefits of a risk-based approach are shown. The main factors affecting the state of cable insulation are indicated. Their division into three categories according to the mechanisms of influence on insulation is given. It is shown that numerous heterogeneous factors makes the task of assessing the cable lines insulation state difficult to formalize. The complex nature of the task of assessing the insulation status of cable lines also leads to the need to use the expert knowledge to increase the validity of the final result. It is shown that at present such problems are easily solved by using the mathematical apparatus of fuzzy systems or fuzzy logic. Systems based on fuzzy logic make it possible to adequately represent the experts knowledge of a subjective or incomplete nature. But these systems cannot automatically acquire knowledge for use in the mechanisms of conclusions, which does not allow to fully overcome the subjectivity of expert knowledge. It is shown that in order to overcome this problem it is necessary to switch to the fuzzy neural networks which allow to acquire new knowledge in the process of their work. The modernized structure of the hybrid network is presented, which allows to reduce the complexity of compiling the knowledge base rules for this network.