A Gaussian-process (GP) model for the control of sequencing batch-reactor (SBR) for wastewater treatment has been developed. The GP model is a probabilistic, nonparametric model with uncertainty predictions. In the case of SBR control, it is used for the on-line optimisation of the batch-phases duration. The control algorithm follows the course of the indirect process variables (pH, redox potential and dissolved oxygen concentration) and recognises the characteristic patterns in their time profile. The control algorithm uses GP-based regression to smooth the signals and GP-based classification for the pattern recognition. When tested on the signals from an SBR laboratory pilot plant, the control algorithm provided a satisfactory agreement between the proposed completion times and the actual termination times of the biodegradation processes.
COBISS.SI-ID: 26698535
This paper presents a temperature model of an industrial, semi-batch, emulsion-polymerisation reactor, which together with the already designed chemical reactions model is able to predict the temperature in the reactor as a result of varying operating conditions. The model was used to analyse the influence of reactants dosing during the batch on the reactor temperature. The analysis shows that during the batch dosing of the two reactants, initiator and monomer, needs to be mutually balanced and adjusted to the current process situation, otherwise, the temperature in the reactor may become oscillatory and unstable towards the end of the batch because of the limited heat removal capacity of the condenser. To keep the reactor temperature in a narrow region also an improved reactants dosing control strategy was proposed. The designed control algorithms and the corresponding control parameters were adjusted so that limited and stable temperature oscillations are achieved with small deviations from the desired temperature set-point, while at the same time the batch processing time was not prolonged because of the shortage of the reacting chemicals.
COBISS.SI-ID: 26605607
An experimental industrial prototype of a development environment for modelling and automatic generation of the procedural part of process control software was developed. The development environment is based on a previously developed methodology for development and automatic generation of process control software, called MAGICS (which has been adapted for industrial use) and on a previously developed laboratory prototype development environment for the mentioned methodology. In the frame of the new industrial prototype a functionality was developed, which allows creating and modifying software models in the modelling language ProcGraph, validating the conformance of models with the rules and syntax of the language, saving models in the project files and exporting models in the XML format, which is a basis for the transformation of models into the code of industrial controllers.
COBISS.SI-ID: 26723623
It is crucial to properly identify the most influential manipulative variables, when applying model-based production control and optimisation. On one hand, the complexity of the models and optimisation problem has to be reduced, and on the other hand enough manipulative influence on the controlled variables has to be provided. Input selection approach is presented where two decisive steps are considered. First, an input variable selection methodology, typically applied for selecting model regressors, is applied. To determine the most appropriate input variable selection approach, different selection methodologies are compared with synthetic data sets. Secondly, the appropriateness of the selected inputs and their manipulative strength is validated by observing achievable operating-space of controlled variables. A complete input variable selection procedure for model-based production control and optimisation was demonstrated on complex simulation production process (Tennessee Eastman process).
COBISS.SI-ID: 26628135
A vast amount of production is being collected in modern production processes. Although the historians offer representative production data, there is still little or no idea how to efficiently exploit the data. Article discusses the problem of finding an appropriate empirical model of production performance, which could be employed to adjust manipulative variables to enhance production performance. The main steps of production performance modelling are described. Special attention is given to neural network identification and representation of the developed assisting tool, which ease the identification of a production performance model. In the article, a simulation of a complex production process (Tennessee Eastman process) is used as a case study. Identification of the production performance models and their practical application are illustrated on a case study, where models for three main production performance indicators (i.e. costs, quality and production rate) are identified.
COBISS.SI-ID: 26263079