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.
Jedan od čestih problema koji se javlja u postrojenjima je nemogućnost kontinuiranog mjerenja i analize ključnih procesnih veličina, posebice kad se radi o sastavima procesnih struja i svojstvima ...proizvoda. Razvoj naprednih senzora, koji se temelje na novim tehnologijama analitičke kemije i suvremenim elektroničkim napravama, važno je područje znanstvenog istraživanja, ali je cijena njihova razvoja vrlo visoka.
Softverski senzori (virtual soft sensors, software sensors, soft analyzers) postaju važna alternativa skupim mjerenjima on-line u primjerima gdje se na temelju fundamentalnih i empirijskih modela može zaključivati o teško mjerljivim veličinama. Razvoj softverskih senzora danas postaje područje velikog interesa, pri čemu se na osnovi analitičkih i empirijskih modela mogu zaključivati i predviđati vrijednosti teško mjerljivih veličina stanja procesa, posebice kad se radi o složenim i nelinearnim procesima. Pri tome se znanja o procesu povezuju sa statističkim metodama za identificiranje i primjenjuju u svrhu optimiranja procesa.
Mogućnosti primjene softverskih senzora posebno su široke u procesnoj industriji, ali i u drugim područjima, primjerice u biokemijskim istraživanjima. Budući da se softverski senzori realiziraju računalno, moguće je procjenjivati veličine unaprijed i na taj način optimirati djelovanje i vođenje
procesa.
Control systems and optimization procedures require regular and reliable measurements at the appropriate frequency. At the same time, legal regulations dictate strict product quality specifications ...and refinery emissions. As a result, a greater number of process variables need to be measured and new expensive process analyzers need to be installed to achieve efficient process control. This involves synergy between plant experts, system analysts and process operators. One of the common problems in industrial plants is the inability of the real time and continuous measurement of key process variables.Absence of key value measurement in a timely manner aggravates control, but it does not mean that it is always an impossible step. As an alternative, the use of soft sensors as a substitute for process analyzers and laboratory testing is suggested. With the soft sensors, the objective is to develop an inferential model to estimate infrequently measured variables and laboratory assays using the frequently measured variables. By development of soft sensors based on measurement of continuous variables (such as flow, temperature, pressure) it is possible to estimate the difficult- -to-measure variables as well as product quality and emissions usually carried by laboratory assays.Software sensors, as part of virtual instrumentation, are focused on assessing the system state variables and quality products by applying the model, thus replacing the physical measurement and laboratory analysis. Multiple linear/nonlinear regression methods and artificial intelligence methods (such as neural network, fuzzy logic and genetic algorithms) are usually applied in the design of soft sensor models for identification of nonlinear processes.Review of published research and industrial application in the field of soft sensors is given with the methods of soft sensor development and nonlinear dynamic model identification. Based on soft sensors, it is possible to estimate product properties in a continuous manner as well as apply the methods of inferential control. By real plant application of the soft sensors, considerable savings could be expected, as well as compliance with strict legal regulations for product quality specifications and emissions.
▶ Linear and nonlinear soft sensor models for light naphtha quality estimation in refinery plant are developed. ▶ The best results were achieved by nonlinear soft sensor models using artificial ...neural networks. ▶ Developed nonlinear models can be used in light naphtha quality monitoring and estimation as an alternative to laboratory assays.
Due to the strict norm requirements of keeping products in crude refining units within specifications, laboratory testing and quality control of the products are necessary. Given this reason, virtual soft sensor for continuous quality estimation of light naphtha as the crude distillation unit (CDU) product was developed. Experimental data included available continuous measurements of CDU process streams (temperatures, pressures and flowrate) and laboratory analyses undertaken twice a day. The results are soft sensor models for light naphtha vapor pressure (RVP) estimation.
Soft sensor models have been developed conducting multiple linear regression analysis and using neural network-based models such as LNN, MLP and RBF. Considering statistical and sensitivity analysis, the best results for both oils were obtained with MLP and RBF neural networks. The results show possible application of the soft sensor models for estimating light naphtha RVP as an alternative for laboratory testing.
Soft sensors for the on-line estimation of kerosene 95 % distillation end point (D95) in crude distillation unit (CDU) are developed. Experimental data are acquired from the refinery distributed ...control system (DCS) and include on-line available continuously measured variables and laboratory data which are consistently sampled four times a day. Additional laboratory data of kerosene D95 for the model identification are generated by Multivariate Adaptive Regression Splines (MARSplines). Soft sensors are developed using different linear and nonlinear identification methods. Among the variety of dynamic models, the best results are achieved with Box Jenkins (BJ), Output Error (OE) and Hammerstein-Wiener (HW) model. Developed models were evaluated based on the Final Prediction Error (FPE), Root Mean Square Error (RMSE), mean Absolute Error (AE) and FIT coefficients. The best results for diagnostic purposes show BJ model. For continuous estimation of D95, OE and HW models can be used. Key words: Crude distillation unit, distillation end point, soft sensor, identification
A soft sensor was developed for quality estimation of diesel fuel as the crude distillation unit product. Due to the stringent fuel quality standards and the growing need to produce various ...gradations of quality with different feeds, laboratory testing and quality control of the products have become a necessity. On the basis of available continuous temperature measurements and flows of appropriate process streams, software sensors for estimating the cold filter plugging point of diesel fuel were developed. Nonlinear software sensor models were built with the best results achieved by using multilayer neural networks. Statistical data analysis was carried out and the data were critically evaluated. The results showed that soft sensors can be applied for refinery product quality estimation of diesel fuel as an alternative to laboratory testing. So, it becomes possible to continuously estimate fuel quality and to apply methods of inferential control for plant operation optimization.
Soft sensors for diesel fuel quality estimation in refinery production are studied. Nonlinear soft sensor models have been developed with the best results achieved by using neural networks. The developed model estimates the cold filter plugging point as an alternative to laboratory assays. The results are validated at an industrial facility.
U ovome radu opisana je nova praktična metoda i softver za identificiranje sustava, ugađanje parametara regulatora i optimiranje sustava za vođenje procesa. Prikazani pristup i alati omogućuju ...inženjerima projektiranje i primjenu različitih metoda vođenja procesa unutar DCS-a i PLC-a. Opisanom metodom identificiraju se empirijski modeli regulacijskih krugova na temelju podataka iz postojećih regulacijskih krugova i optimiraju se parametri regulatora. Na temelju određenih dinamičkih i statičkih karakteristika procesa moguće je razviti prilagodljive i napredne metode vođenja. Softverski alati služe za obuku inženjera i operatora, ali i za praktičnu primjenu. Optimiranjem standardne i napredne regulacije stabilizirat će se proces, a postrojenje će raditi bliže procesnim, sigurnosnim i ekonomskim granicama. Na taj način povećava se kapacitet proizvodnje, smanjuju se troškovi energenata i troškovi održavanja, a raste kvaliteta proizvoda.
This paper describes application of the new method and tool for system identification and PID tuning/advanced process control (APC) optimization using the new 3G (geometric, gradient, gravity) ...optimization method. It helps to design and implement control schemes directly inside the distributed control system (DCS) or programmable logic controller (PLC). Also, the algorithm helps to identify process dynamics in closed-loop mode, optimizes controller parameters, and helps to develop adaptive control and model-based control (MBC). Application of the new 3G algorithm for designing and implementing APC schemes is presented. Optimization of primary and advanced control schemes stabilizes the process and allows the plant to run closer to process, equipment and economic constraints. This increases production rates, minimizes operating costs and improves product quality.
Pri razvoju inteligentnih sustava u posljednjih dvadesetak godina ostvarena su brojna unapređenja inspirirana biološkim neuronskim sustavom. Istraživači s različitih znanstvenih područja kreirali su ...i primijenili umjetne neuronske mreže za rješavanje niza zadataka - od prepoznavanja uzoraka, predviđanja, dijagnosticiranja stanja, softverskih senzora, modeliranja i identificiranja, vođenja i optimiranja procesa itd.
Umjetne neuronske mreže pokazale su se korisnim u primjeni kod složenih kemijskih i biokemijskih procesa gdje standardnim metodama nije moguće uspješno modelirati procese i dobivene modele primijeniti za vođenje procesa. Danas, zahvaljujući intenzivnom razvoju teorije i praktične primjene neuronskih mreža, stoje na raspolaganju brojne strukture i algoritmi.
U radu je dan pregled primjene neuronskih mreža s težištem na identificiranju i vođenju procesa na polju kemijskog inženjerstva. Istaknuti su primjeri primjene kod prediktivnog, inverznog i prilagodljivog vođenja procesa.