Abstract : One of the most popular models in scheduling theory is that the job shop, as it is strongly motivated by practical requirements and has earned a reputation for being difficult to solve. It is probably the most studied and well developed model in deterministic scheduling theory. We study in this thesis the deterministic job shop scheduling problem. In first time we consider the problem referred to J/ / Cmax under the machine availability constraints. We propose a novel use of Knowledge Discovery from Data (KDD) process to solve the job shop. We adapt the KDD process step by step to explore the patterns in good solutions of job shop instances generated by a genetic algorithm and to develop a scheduling rule set which approximates the genetic algorithm's scheduler. Genetic algorithms often provide fast solutions to traditional numeric problems. However, they do not demonstrate repeatability or provide an explanation of how a solution is developed. Using KDD, the thesis presents a method for inducing dispatching rules from the genetic algorithm solutions. These rules have been applied differently to solve the job shop. Firstly, the rules were transformed into a heuristic to solve different-size problems. Secondly, the rules have been applied with success to similar job shop cases with the same size of the learning example. In the second part of the thesis, we study the combined maintenance and production scheduling. Such problem is traditionally treated independently specially for the multi product environment. We optimize jointly two criteria, one for production: the makespan and one for maintenance: the total cost. These two criteria are antagonist, that is why we develop a multiobjective and genetic algorithm based method to schedule simultaneously production and maintenance. The genetic algorithm uses a Pareto optimal selection keeping only the more adapted solutions, which are validated using some lower bounds. For the purpose of reducing the frequency of breakdowns, we apply the preventive maintenance and we compare two types of maintenance periods.