Systémy včasné výstrahy průmyslových systémů

Abstract

As manufacturing companies expand or transform production, they use ever more complex production equipment and technology to be able to promptly adapt to immediate customer demands and needs. However, as the complexity of individual machines, components and devices around us grows, the approach to their care (maintenance) changes accordingly, and the complexity of the relationships (models) that describe the reliability of the devices also increases. The main goal of the dissertation is to create a methodology for the creation of a technical-economic model for optimizing the maintenance of industrial equipment to support maintenance management, the basis of which will be the neurogenetic system. This model is created using the Monte Carlo method, which allows the creation of a model using a minimal amount of data. The proposed model is subsequently adapted during further use based on current technical and economic data. With this created model, it will be possible to optimize the term of preventive maintenance. The methodology is therefore based on three basic elements: creation of a training set using the Monte Carlo method, creation of a neuro-genetic system algorithm, parameterization and testing of the model. The main output of the dissertation is the creation of a methodology for creating a technical-economic model to support maintenance management. The input data are both technical and economic parameters. For this methodology of creating a technical-economic model to support maintenance management, a so-called combined approach is proposed, which means that not only artificial intelligence methods are, used to solve a certain problem, but a connection with some other method, for this model the Monte Carlo method. The output of the entire methodology is an updated optimal interval for any technical object. The novelty of the solution lies in the fact that the result is a methodology for creating an optimization model, and not a general optimization function, and also maintenance optimization is performed on the entire system, and not on individual components. For optimization, a technical-economic objective function is used, which is implemented using a neural network, so the proposed objective function does not have a general character, but is always implemented for a specific device and takes into account all its specifics.