Markov machine repair model consisting of mixed spares under the supervision of two heterogeneous repairmen is investigated. Every time any component fails, it is quickly supplanted by a spare component if accessible. In the event when all spares are used up and the system operates with less than K functioning components then the failure of components happens in a degraded fashion. The distribution of failure time and repair time of the components are assumed to be exponential and FCFS discipline is used for repairing of failed components. The failed components may balk with constant probability as a result of impatience and may renege according to negative exponential distribution if the repairmen are busy. A reneged component can be retained in the queue by using some convincing ways to finish the repair process. After the repair, each component may rejoin the system as a feedback customer with some probability in case of imperfect repair. The repairmen switch to vacation state from busy state whenever there are no failed components in the system and repairmen switch back to the busy state as soon as any failed component arrives. The numerical technique named ‘Successive Over Relaxation (SOR) Technique’ is employed to acquire the system state probabilities at steady state which are then used to calculate the mean count of failed components in the system, throughput, carried load and other indices of the system. The numerical outcomes acquired are verified using the results generated by ANFIS.