This article describes the process of multicriterial optimization using the Pareto efficiency method. A large-scale industrial plant was taken as a controllable object. The object was decomposed and represented as a hierarchy of embedded orgraphs. The orgraph’s vertices mark the current state of the product while the edges stand for technological operations. Based on the object’s technical documentation, a list of influencing factors is created. The list contains every technological parameter affecting the quality of the final product. A neural network trained on a set of statistical data is utilized to identify dependencies between discrete influencing factors and the product quality. These dependencies are then processed with the SPEA2 algorithm, outputting a set of combinations of optimized parameters values known as the Pareto front.