Mathematical modeling and optimization techniques are key elements in modern industrial production and the maintenance of services vital to modern society. System models and optimization tools are used in controlling the massive flows of goods, energy, and information necessary for the everyday functioning of society, managing the resources for competitive production, the design of effective services, etc. Most conventional optimization techniques start from the assumption that all the relevant requirements and resources are known and available at the outset, and the optimal solution or process is determined under the given constraints. However, the continuing digitalization of manufacturing and services is making it increasingly feasible to continuously monitor the requirement and resource status of the systems at hand, and adapt their behavior to the changing circumstances. This kind of online optimization has great potential in increasing the flexibility, efficiency and service level provided by the systems of interest. Research into the methodology and applications of online optimization techniques is, however, still at its early stages, particularly in the case of the difficult combinatorial optimization tasks that often arise in practical applications.

In this PhD project we develop the methodology of the emerging area of online optimization, especially in the context of industrial production planning. The results are validated both analytically and by simulations reflecting real-world conditions.