In industrial systems, the K-out-of-N: G system is a prominent type of redundancy. The load sharing protects such system from malfunctioning/destroying and avoids overload problem that affects the system reliability in a significant manner. In this paper we develop a Markovian model of load-sharing K-out-of-N: G system having non-identical repairable components wherein the server may on working vacation. During his vacation period, the server repairs the failed components with different service rates rather than completely terminating service rate. The failed component gets immediately repaired by the server if not on vacation, and unequal load is distributed among remaining surviving components. The lifetime of each component is load dependent followed by non-identical exponential distribution with different failure rates. The system is failed down due to common cause with failure density which is also exponentially distributed. We suggest closed structure analytic expressions for reliability, cost estimation and other performance measures of the load-sharing K-out-of-N: G repairable system by incorporating the concept of working vacation. For the solution aspiration, Runge-Kutta method is utilized to solve the system of differential equations. Furthermore, we perform the numerical analysis for two illustrations 1-out-of-3: G system and 3-out-of-4: G system. The numerical simulation is carried out for the validation of analytical results which are exhibited and compared by giving numerical outcomes and neuro-fuzzy outcomes based on fuzzy interference system with the help of MATLAB.

No abstract

To assure the reliability and quality of the final product, testing is an essential and crucial part in the software development cycle. During this process, fault correction/detection activities are carried out to increase the reliability of the software. The non-homogeneous Poisson Process (NHPP) is the basis of the investigated software reliability growth models (SRGMs), which are based on the supposition that the number of faults found is affected by the amount of code covered during testing and that the amount of code covered during testing depends on the testing effort expended. This research takes into consideration several testing coverage functions: exponential, delayed S-shaped and logistic distributions, to propose three SRGMs that are based on testing efforts. For testing effort expenditure Weibull distribution has been employed. Two real failure datasets have been utilised to validate the proposed models, and their performance is evaluated using four goodness-of-fit metrics, including predictive ratio risk (PRR), coefficient of determination (R^2 ), predictive power (PP) and mean square error (MSE). Sensitivity analysis of cost requirement-based release time of software for exponential function has been done by using a genetic algorithm, which minimized the overall cost of the software subject to the requirement for reliability.

No abstract

No abstract