Display omitted
•Component contents continuous estimation in the isomerization process products.•Support vector machine and dynamic polynomial model structures are employed.•Soft sensors for a ...physical analyser replacement.
A novel data-driven soft sensor models for application in the refinery isomerization process are presented. Soft sensor models based on support vector machine regression (SVM) and dynamic polynomial linear Finite Impulse Response (FIR), Autoregressive with Exogenous Inputs (ARX), Output Error (OE), and Nonlinear Dynamic Autoregressive with Exogenous Inputs (NARX) and Hammerstein–Wiener (HW) models are developed. They are intended for continuous estimation of key component contents in the products of a low-temperature isomerization process equipped with a deisohexanizer distillation column. Experimental data from the refinery distributed control system are employed. A significant attention is paid on collection, analysis and pre-processing of the data as well as selection of influential input variables. Developed models were evaluated on an independent data set, and the results show that selected models can reliably estimate the component contents. SVM regression model has better generalization ability in comparison with standard dynamic models on the data set with a low diversity. Developed soft sensors are suitable as analyser replacement and application in the deisohexanizer column advanced process control strategies.
This paper describes the design, construction and experimental testing of a single-joint manipulator arm actuated by pneumatic artificial muscles (PAMs) for the tasks of transporting and sorting work ...pieces. An antagonistic muscle pair is used in a rotational sense to produce a required torque on a pulley. The concept, operating principle and elementary properties of pneumatic muscle actuators are explained. Different conceptions of the system realizations are analyzed using the morphological-matrix conceptual design framework and top-rated solution was practically realized. A simplified, control-oriented mathematical model of the manipulator arm driven by PAMs and controlled with a proportional control valve is derived. The model is then used for a controller design process. Fluidic muscles have great potential for industrial applications and assembly automation to actuate new types of robots and manipulators. Their characteristics, such as compactness, high strength, high power-to-weight ratio, inherent safety and simplicity, are worthy features for advanced manipulation systems. The experiments were carried out on a practically realized manipulator actuated by a pair of muscle actuators set into an antagonistic configuration. The setup also includes an original solution for the subsystem to add work pieces in the working space of the manipulator.
Dynamic neural networks (DNNs) are a type of artificial neural network (ANN) designed to work with sequential data where context in time is important. Unlike traditional static neural networks that ...process data in a fixed order, dynamic neural networks use information about past inputs, which is important if the dynamic of a certain process is emphasized. They are commonly used in natural language processing, speech recognition, and time series prediction. In industrial processes, their use is interesting for the prediction of difficult-to-measure process variables. In an industrial isomerization process, it is crucial to measure the quality attributes that affect the octane number of gasoline. Process analyzers commonly used for this purpose are expensive and subject to failure. Therefore, to achieve continuous production in the event of a malfunction, mathematical models for estimating product quality attributes are imposed as a solution. In this paper, mathematical models were developed using dynamic recurrent neural networks (RNNs), i.e., their subtype of a long short-term memory (LSTM) architecture. The results of the developed models were compared with the results of several types of other data-driven models developed for an isomerization process, such as multilayer perceptron (MLP) artificial neural networks, support vector machines (SVM), and dynamic polynomial models. The obtained results are satisfactory, suggesting a good possibility of application.
This paper presents the development of soft sensor empirical models using support
vector machine (SVM) for the continual assessment of 2,3-dimethylbutane and 2-methylpentane mole percentage as ...important product quality indicators in the refinery isomerisation process. During the model development, critical steps were taken, including selection and pre-processing of the industrial process data, which are broadly discussed in this paper. The SVM model results were compared with dynamic linear output error model and nonlinear Hammerstein-Wiener model. Evaluation of the developed models on independent data sets showed their reliability in the assessment of the component contents. The soft sensors are to be embedded into the process control system, and serve primarily as a replacement during the process analysersb failure and service periods.
Raman spectroscopy is a useful tool for polymorphic form-monitoring during the crystallization process. However, its application to solute concentration estimation in two-phase systems like ...crystallization is rare, as the Raman signal is influenced by various changing factors in the crystallization process. The development of a robust calibration model that covers all variations is complex and represents a major challenge for the implementation of Raman spectroscopy for in-line monitoring and control of the solution crystallization process. This paper describes the development of a Raman-based calibration model for estimating the solute concentration of the active pharmaceutical ingredient ceritinib. Several different calibration approaches were tested, which included both temperature and spectra of clear solutions and slurries/suspensions. It was found that the concentration of the ceritinib solution could not be accurately predicted when suspended crystals were present. To overcome this challenge, the approach was enhanced by including additional variables related to crystal size and solid concentration obtained via in-line process microscopy (chord-length distribution percentiles D10, D50 and D90) and turbidity. Partial least squares regression (PLSR) and artificial neural network (ANN) models were developed and compared based on root mean square error (RMSE). ANN models estimated the solute concentration with high accuracy, with the prediction error not exceeding 1% of the nominal solute concentration.
Cilj rada bio je izraditi računalni program koji služi za proračun optimalnog temperaturnog profila hlađenja šaržnog kristalizatora. Kao modelni sustav za istraživanje procesa uzet je kalijev nitrat ...otopljen u vodi. Za izradu matematičkog modela procesa primijenjene su populacijske bilance, odnosno njihova momentna transformacija. Za dobivanje optimalnog temperaturnog profila primijenjena je diskretizacija temperaturnog profila uz globalni algoritam optimizacije. Za provođenje optimizacije primijenjen je genetički algoritam, dok je sustav običnih diferencijalnih jednadžbi rješavan metodom Runge-Kutta 4,5. Funkcija cilja bila je minimiziranje omjera trećeg momenta sekundarnom nukleacijom nastalih kristala i trećeg momenta kristala cjepiva na kraju procesa. U radu je najprije ispitan utjecaj uvjeta zaustavljanja genetičkog algoritma na vrijeme proračuna i vrijednost funkcije. Nakon što je određen optimalni uvjet zaustavljanja, ispitan je utjecaj broja točaka diskretizacije temperaturnog profila na iznos funkcije cilja i potrebno vrijeme proračuna. Ustanovljeno je da je optimalni uvjet za zaustavljanje proračuna kad petnaest članova generacije imaju funkcije cilja koje se ne razlikuju više od tolerancije. Ustanovljeno je da se optimalno rješenje dobiva podjelom temperaturnog profila na osam dijelova. Da bi se ispitala ponovljivost proračuna za optimalne uvjete, proračun je ponavljan devet puta. Optimalni temperaturni profil uspoređen je s linearnim hlađenjem istog trajanja. Rezultati simulacijskih eksperimenata ukazuju na znatno poboljšanje procesa primjenom optimalnog temperaturnog profila naspram linearnog.
As vehicle emission standards become more stringent, there is an increasing need for continual monitoring of benzene content in gasoline. Since on-line analyzers are often unavailable, and laboratory ...analyses are infrequently obtained, soft sensors for the estimation of benzene content of light reformate are developed. Soft sensors are developed using linear and nonlinear identification methods. Experimental data are acquired from the refinery distributed control system (DCS) and include continuously measured variables and analyzer assays available on-line. In the present work, the development of a finite impulse response (FIR) model, an output error (OE) model, and a Hammerstein–Wiener (HW) model is presented. To overcome the problem of selecting the best model parameters by trial and error, genetic algorithms and pattern (direct) search were used. On the basis of developed soft sensors, it is possible to entirely replace on-line analyzers with soft sensors by embedding the model in a DCS on-site.
Industrial facilities nowadays show an increasing need for continuous measurements, monitoring and controlling many process variables. The on-line process analyzers, being the key indicators of ...process and product quality, are often unavailable or malfunction. This paper describes development of soft sensor models based on the real plant data that could replace an on-line analyzer when it is unavailable, or to monitor and diagnose an analyzer's performance. Soft sensors for continuous toluene content estimation based on the real aromatic plant data are developed. The autoregressive model with exogenous inputs, output error, the nonlinear autoregressive model consisted of exogenous inputs and Hammerstein-Wiener models were developed. In case of complex real-plant processes a large number of model regressors and coefficients need to be optimized. To overcome an exhaustive trial-and-error procedure of optimal model regressor order determination, differential evolution optimization method is applied. In general, the proposed approach could be, of interest for the development of dynamic polynomial identification models. The performance of the models are validated on the real-plant data.
Due to growing fuel quality demands, continuous measurements of process variables and product quality properties in the crude distillation unit (CDU) are necessary. One of the key diesel fuel ...properties is kerosene cold filter plugging point (CFPP). CFPP is usually determined only by laboratory assays. On the basis of available continuous measurements of temperatures and flows of appropriate process streams, soft sensor models for the estimation of kerosene CFPP have been developed. Data preprocessing includes: detection and outlier removal, generating additional output data by Multivariate Adaptive Regression Splines (MARSplines) algorithm, detrending data and filtering data. Soft sensors are developed using linear and nonlinear identification methods. Model structures are optimized by Genetic Algorithm (GA) and ANFIS (Adaptive Neuro-Fuzzy Inference System) algorithm. Results of the Output Error (OE) model, Hammerstein–Wiener (HW) model and neuro-fuzzy model are shown. Developed models were evaluated based on the final prediction error (FPE), root mean square error (RMSE), mean absolute error (AE) and FIT values. The best results are achieved with neuro-fuzzy model.
•Dynamic black box models for estimation of cold filter plugging point are developed•Model structures are optimized by Genetic Algorithm and ANFIS algorithm•Results obtained by the OE and HW models show good agreement•The best results are achieved with neural fuzzy model
Due to reduced energy efficiency and productivity loss caused by fouled heat exchangers in industrial plants there is an obvious need for detection of fouling formation. Based on continuously ...measured temperatures and flow rates collected from the refinery plant history database, a semi-empirical number of transfer units (NTU) model and neural network-based models are developed. In order to confirm the reliability of proposed fouling factor calculation, the entire procedure was performed by developing a dynamic nonlinear finite impulse response (NFIR) model. Developed models are intended for fouling detection for industrial shell and tube heat exchangers. The performance criteria of developed models together with residual monitoring indicate that not only neural networks but the NTU and NFIR models effectively detect fouling formation.