Multi-State System Reliability Design via Particle Evolutionary Swarm Optimization

Many real-world systems are composed of multi-state items that have different performance levels and several failure modes. Items are characterized by their cost, performance and reliability. In a multi-state system, the system and/or its items can posses more than two performance levels. Thereby, the system can have several levels of performance from perfect operation to total failure. The system reliability is modeled to find the optimal items configuration that minimizes the cost of the system subject to minimum system performance levels. In many new designs, item reliabilities are often uncertain. Thus, the associated system reliability are uncertain as well. In this work, the reliability variance of multi-state systems is studied. Applying the Delta method for the propagation of item variances through system structure, a technique for the approximation of variance of the entire system is proposed. There are a lot of real-world systems that require a special analysis due to their complex structure and importance in everyday life. The vehicle routing problem has been recognized as one of the great success stories of the operations research area. It is faced everyday by thousands of companies and organizations engaged in the delivery and recollection of goods and people. This dissertation studies the vehicle routing problem in presence of unavailable paths, modeling it as a multi-state system. An evolutionary algorithm, a modified version of the particle swarm optimization algorithm with constraint handling rules, is introduced for solving system reliability optimization problems.