Search

This paper presents a study on applying machine learning algorithms for the classification of a two-phase flow regime and its internal structures. This research results may be used in adjusting optimal control of air pressure and liquid flow rate to pipeline and process vessels. To achieve this goal the model of an artificial neural network was built and trained using measurement data acquired from a 3D electrical capacitance tomography (ECT) measurement system. Because the set of measurement data collected to build the AI model was insufficient, a novel approach dedicated to data augmentation had to be developed. The main goal of the research was to examine the high adaptability of the artificial neural network (ANN) model in the case of emergency state and measurement system errors. Another goal was to test if it could resist unforeseen problems and correctly predict the flow type or detect these failures. It may help to avoid any pernicious damage and finally to compare its accuracy to the fuzzy classifier based on reconstructed tomography images -- authors' previous work. [ABSTRACT FROM AUTHOR]

The paper presents an intelligent module to control dynamic two-phase gas–liquid mixtures pipelines flow processes. The module is intelligent because it uses the algorithm based on AI methods, namely, fuzzy logic inference, to build the fuzzy regulator concept. The developed modification has allowed to design and implement the black-box type regulator. Therefore, it is not required to determine any of the complicated computer models of the flow rig, which is unfortunately necessary when using the classic regulators. The inputs of the regulator are four linguistic variables that are decomposed into two classes and two methods of fuzzification. The first input class describes the current values of gas and liquid pipe flows, which at the same time are the controlled values manipulated to generate desired flow type. The second class of the input signals contains a current flow state, namely, its name and the name preferred by the operator flow type. This approach improves the control accuracy since the given flow type can be generated with different gas and liquid volume fractions. Those values can be optimized by knowing the current flow type. Moreover, the fuzzification algorithm used for the input signals included in the first-class covers the current crisp signal value and its trend making the inference more accurate and resistant to slight measurement system inaccuracy. This approach of defined input signals in such environments is used for the first time. Considering all mentioned methods, it is possible to generate the desired flow type by manipulating the system input signals by minimum required values. Furthermore, a flow type can be changed by adjusting only one of the input signals. As an output of the inference process, two linguistic values are received, which are fuzzified adjustment values of the liquid pump and gas flow meter. The regulator looks to be universal, and it can be adopted by multiple test and production rigs. Moreover, once configured with a dedicated rig, it can be easily operated by the non (domain) technical staff. The usage of fuzzy terms makes understanding both the control strategy working principles and the obtained results easy. [ABSTRACT FROM AUTHOR]

In the present research work, an electrical resistance tomography (ERT) system is utilized as a means for real-time fault detection and diagnosis (FDD) during a reactive crystallization process. The calcium carbonate crystallization is part of the carbon capture and utilization scheme where process monitoring and malfunction diagnostics strategies are presented. The graphical logic representation of the fault tree analysis methodology is used to develop the system failure states. The measurement consistency due to the use of a single electrode from a set of ERT electrodes for malfunction identification is experimentally and quantitatively investigated based on the sensor sensitivity and standard deviation criteria. Electrical current measurements are employed to develop a LabVIEW-based process automation program by using the process-specific knowledge and historical process data. Averaged electrical current is correlated to the mechanical failure of the stirrer through standard deviation evaluation, and slopes of the measured data are used to monitor the pump and concentrations status. The performance of the implemented methodology for detecting the induced faults and abnormalities is tested at different operating conditions, and a basic signal-based alarming technique is developed. [ABSTRACT FROM AUTHOR]