No abstract

No abstract

No abstract

This paper deals with reliability analysis of transformer using stochastic or random process through supplementary variable technique (SVT) and neural network approach. The failure of the considered complex system can occur due to the failure of breakdown voltage (BDV) and moisture content (MC) of insulating paper and oil. The repair rates for both insulation oil and insulation paper are presumed to be general, whereas for insulation, oil repair rates are exponentially distributed. Further, whenever there are two possibilities of repairs, it has been coupled using Gumbel–Hougaard copula. All types of failures and repair rates are treated as weights while analyzing with the help of neural network approach which are governed by exponential distribution. System state probabilities, up and down states probabilities and reliability are evaluated for the proposed model by using Markovian process with the help of Laplace transforms. Numerical examples are included to illustrate the results.

The world’s ever increasing demand for energy and abating global warming, suitable renewable sources of energy are highly in demand. The wastes from industries such as plant’s biomass could meet the energy requirements. In this paper authors analyze bio fuel plant system which produces ethanol fuel. This system is divided into various subsystems considering multiple phases in the production of ethanol. The structure of this system consists of interconnected networks of components on very large dimensional scales escalates the complexity of systems that can increase the degradation of system's functioning. In view of this, one of the computational intelligence approach, neural network (NN), is useful in predicting various reliability parameters. To improve the accuracy and consistency of parameters, Feed Forward Back Propagation Neural Network (FFBPNN) is used. All types of failures and repairs follow exponential distributions. System state probabilities and other parameters are developed for the proposed model using neural network approach. Failures and repairs are treated as neural weights. Neural network's learning mechanism can modify the weights due to which these parameters yield optimal values. Numerical examples are included to demonstrate the results. The iterations are repeated till the convergence in the error tends up to 0.0001 precision using MATLAB code. The reliability and cost analysis of the system can help operational managers in taking the decision to implement it in the real time systems