The development of fast and efficient computer hardware technology has resulted in the rapid development of numerous computational software tools for making statistical inferences. The computational algorithms, which are the backbone of these tools, originate from distinct areas in science, mathematics and engineering. The main focus of this thesis is on computational tools which can be employed for estimating unmeasured variables in a process using all the available prior information. Specifically, this thesis demonstrates the application of a variety of tools for soft sensing of process variables and uncertain parameters of physiochemical process models, using routine data available from the process. The application examples presented in this thesis come from broad areas where process uncertainty is inherent and includes petrochemical processes, mechanical valve actuators, and upstream production processes in petroleum reservoirs. The mathematical models that are employed in different domains vary significantly in their structure and their level of complexity. In the petrochemical domain, the focus was on developing empirical soft sensors which are essentially nonparametric mathematical models identified using routine data from the process. The Support Vector Regression technique was applied for identifying such nonparametric empirical models. On the other hand, in all the other application examples in this thesis the physical parametric models of the process were utilized. The latter application examples, which cover a major portion of this thesis, demonstrate the application of modern state and parameter estimation algorithms which are firmly grounded on Bayesian theory and Monte Carlo techniques. Prior to the chapters on the application of state and parameter estimation techniques, a tutorial overview of the Monte Carlo simulation based state estimation algorithms is provided with an attempt to throw new light on these techniques. The tutorial is aimed at making these techniques simple to visualize and understand. The application case studies serve to illustrate the performance of the different algorithms. All case studies presented in this thesis are performed on processes that exhibit significant nonlinearity in terms of the relationship between the process input variables and output variables. / Process Control

This work describes a procedure for a soft sensor design to predict the top composition of a methanol-water distillation column. Soft sensor is a mathematical model that is used to estimate variables of interest from secondary variables easy to measure. This technique comes from an operational difficulty or high cost obtaining the desired variable. The approach to build a soft sensor was an artificial intelligence modeling, a black-box type, using a hybrid neuro-fuzzy technique. The data acquisition to train and validate the soft sensor comes from a mathematical model validated from pilot plat data. One of the limitations of neuro-fuzzy system is that it works with a limited number of inputs, depending on the combinatorial explosion of fuzzy rules. To minimize these effects and to reduce the number of rules in the training data sets of virtual analyzer, a data clustering technique called substractive clustering was used. To obtain a better performance of soft sensor for the dynamic process, distillation column, a regression of lone sampling time in selected variables was used, changing the number of entries from 9 to 18 variables, nine variables at actual sampling time and nine variables at previous sampling time. The distillation column is a good process for the present study because composition measurements are the main objective of this process and are difficult to obtain. The computational strategy for a soft sensor design produced good results in estimating the top composition of the methanol-water distillation column. / Este trabalho descreve um procedimento para o desenvolvimento de um analisador virtual, para predição da composição de topo de uma coluna destilação metanol-água em uma planta piloto. Analisador virtual é um modelo matemático que é usado para estimar variáveis de interesse a partir de variáveis secundárias de fácil medição. Esta tecnologia surge de uma real dificuldade operacional ou do alto custo de obtenção da variável desejada. O modelo utilizado nesta abordagem de construção do analisador virtual utiliza técnicas de sistemas inteligentes, tipo caixa preta, através da técnica híbrida neuro-fuzzy. A aquisição dos dados para treinar e validar o analisador virtual foi feita através de um modelo matemático validado a partir de dados experimentais da planta piloto. Uma das limitações do sistema neuro-fuzzy é que ele trabalha com um número limitado de entradas, dependendo da explosão combinatória das regras fuzzy. Para minimizar estes efeitos e conseguir reduzir o número de regras nos conjuntos de treinamento da rede neuro-fuzzy, foi utilizada a técnica de agrupamento de dados, denominada agrupamento substrativo. Com a intenção de se obter um melhor desempenho do analisador virtual no processo dinâmico, que é a coluna de destilação, foi empregada uma regressão de um tempo de amostragem nas variáveis de entrada selecionadas, alterando o número de entradas de 9 para 18, sendo 9 variáveis no tempo de amostragem atual e 9 variáveis em um tempo de amostragem anterior. O processo de destilação mostrou-se adequado para o presente estudo, pois as medições de composições são de difíceis obtenções. A estratégia computacional para um projeto de analisador virtual produziu bons resultados, de forma a estimar a composição do topo da coluna de destilação binária metanol-água.

Le travail de recherche présenté dans ce mémoire concerne le développement de capteurs logiciels de position pour la commande de la machine synchrone à aimants permanents. La commande vectorielle de la MSAP nécessite une connaissance précise de la position rotorique. Traditionnellement, cette position est obtenue à partir de l’utilisation d’un capteur mécanique.Depuis des années, l’attention de la communauté scientifique s’est portée sur la limitation du nombre de capteurs vu que leur présence, non seulement augmente le coût et la complexité matérielle totale, mais aussi réduit sa fiabilité avec une sensibilité additionnelle aux perturbations extérieures. Dans une première partie, nous présentons plusieurs types de capteurs logiciels deposition pour la MSAP. En fonction du régime de fonctionnement de la machine, nous proposons le capteur, selon nous, le mieux adapté pour une application automobile. Ce capteur est basé sur le couplage intelligent entre un observateur et un capteur logiciel basé sur la technique d’injection de signaux. Dans une deuxième partie, nous proposons une méthode simple et rapidepermettant l’estimation de la résistance et des inductances statoriques à l’arrêt. La méthode proposée, basée sur la technique d’injection de signaux de haute fréquence, exploite la mise en oeuvre des filtres à variable d’état afin d’obtenir un modèle linéaire par rapport aux paramètres. La combinaison de l’identification à l’arrêt et du capteur logiciel permet une bonne estimationde la position de la MSAP sur une large plage de vitesse y compris les basses vitesses et à l’arrêt. Nous abordons également certains aspects de commande de robustesse vis-à-vis de l’ensemble des paramètres incertains de la machine, mais ce de manière plus prospective. / This thesis focuses on the development of soft sensors for position control of the permanent magnet synchronous machine. Vector control of PMSM requires accurate knowledge of the rotor position. Traditionally, this position can be obtained from a mechanical sensor. Many years ago, the attention of the scientific community has focused on reducing the number of sensorsbecause their presence not only increases the cost and the total hardware complexity, but also reduces its reliability with an additional sensitivity to external disturbances. As a first step, we present several known types of soft sensors for PMSM. We present the complete design of a soft sensor for speed measurement of permanent magnet synchronous motor. The rotor speedand position can be estimated in a wide speed range even at low speed and standstill. We introduce two soft sensors operating in two different ranges of speed. Secondly, a simple method based on high frequency signal injection and exploiting the implementation of state variable filters to obtain a linear model with respect to the parameters is presented. Thus, a simplifiedprocedure of identification based on a least squares algorithm can be used. In an automotive application, the PMSM parameters can change due to temperature variation and aging of the material. Therefore, the coupling of the hybrid soft sensor and the simplified pocedure of identification provides a good estimate of the PMSM position over a wide speed range including standstill. We also consider a new approach to the robust control of the PMSM, but just as a newtrack for further investigations.

With increasing digitization, models are more important than ever. Especially their use as soft sensors during operation offers opportunities in cost saving, easy data acquisition and therefore additional functionality of systems. In soft sensor networks there is redundant data acquisition and consequently the occurrence of inconsistent values from different soft sensors is encouraged. The resolution of these data-induced conflicts allows for the detection of changing components characteristics. Hence soft sensor networks can be used to detect wear in system components. In this paper this approach is validated on a test rig. It is found, that the soft sensor network is capable to determine wear and its extent in eccentric screw pumps and valves via data induced conflicts with relatively simple models.

The aim of this project was to examine if a machine learning model could be used to predict snow density from six different weather parameters. These were artificially generated snow density, air temperature, ground temperature, relative humidity, windspeed and the snow depth change. The questions asked were what parameters correlates to the snow density, what model will perform best and could this approach be a better alternative to measure snow density manually. The research was performed in the application Regression Learner in MATLAB by testing five different premade machine learning models on a dataset. The premade models were, Linear Regression, GPR Matern 5/2, SVM Medium Gaussian, Wide Neural Network and Trilayered Neural Network. Also, the project includes data collection, data cleaning, data modification, data generation, training, testing, and evaluating the models. The results show that air temperature and windspeed overall are the most important parameters and the GPR Matern 5/2 and the Wide Neural Network had the highest performance. Lastly, it was concluded that the machine learning model could be a better alternative to measuring snow density with a real sensor. / Målet med detta arbete var att undersöka om en maskininlärningsmodell kunde användas för att förutse snödensitet utifrån sex olika väderparametrar. Dessa var artificiell genererad snödensitet, lufttemperatur, marktemperatur, relativ luftfuktighet, vindhastighet och snödjupsförändring. Frågeställningarna som skulle besvaras var vilka väderparametrar som korrelerar med snödensiteten, vilken eller vilka modeller som presterade bäst samt om maskininlärningsmodellen skulle kunna vara att bättre alternativ till att mäta snödensitet manuellt. Undersökningen utfördes i applikationen Regression Learner i MATLAB genom att testa fem olika förhandsgjorda modeller vilka var Linear Regression, GPR Matern 5/2, SVM Medium Gaussian, Wide neural network och Trilayered neural network. Projektet inkluderar även datainsamling, städning av data, datamodifiering, datagenerering, träning, testning och evaluering av modellerna. Resultaten visar att lufttemperaturen och vindhastigheten över lag är viktigast för modellerna och att GPR Matern 5/2 samt Wide neural network presterade bäst. Slutligen kunde man argumentera för att maskininlärningsmodellen är ett bättre alternativ till att mäta snödensitet manuellt.

京都大学 / 新制・課程博士 / 博士(情報学) / 甲第24041号 / 情博第797号 / 新制||情||135(附属図書館) / 京都大学大学院情報学研究科システム科学専攻 / (主査)教授 加納 学, 教授 下平 英寿, 教授 石井 信 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

Varje studie av en verklig process eller ett verkligt system är baserat på mätdata. Förr var den tillgängliga datamängden vid undersökningar ytterst begränsad, men med dagens teknik är mätdata betydligt mer lättillgängligt. Från att tidigare enbart haft få och ofta osammanhängande mätningar för någon enstaka variabel, till att ha många och så gott som kontinuerliga mätningar på ett större antal variabler. Detta förändrar möjligheterna att förstå och beskriva processer avsevärt. Multivariat analys används ofta när stora datamängder med många variabler utvärderas. I det här projektet har de multivariata analysmetoderna PCA (principalkomponentanalys) och PLS (partial least squares projection to latent structures) använts på data över avloppsvatten insamlat på Hammarby Sjöstadsverk. På reningsverken ställs idag allt hårdare krav från samhället för att de ska minska sin miljöpåverkan. Med bland annat bättre processkunskaper kan systemen övervakas och styras så att resursförbrukningen minskas utan att försämra reningsgraden. Vissa variabler är lätta att mäta direkt i vattnet medan andra kräver mer omfattande laboratorieanalyser. Några parametrar i den senare kategorin som är viktiga för reningsgraden är avloppsvattnets innehåll av fosfor och kväve, vilka bland annat kräver resurser i form av kemikalier till fosforfällning och energi till luftning av det biologiska reningssteget. Halterna av dessa ämnen i inkommande vatten varierar under dygnet och är svåra att övervaka. Syftet med den här studien var att undersöka om det är möjligt att utifrån lättmätbara variabler erhålla information om de mer svårmätbara variablerna i avloppsvattnet genom att utnyttja multivariata analysmetoder för att skapa modeller över variablerna. Modellerna kallas ofta för mjukvarusensorer (soft sensors) eftersom de inte utgörs av fysiska sensorer. Mätningar på avloppsvattnet i Linje 1 gjordes under tidsperioden 11 – 15 mars 2013 på flera ställen i processen. Därefter skapades flera multivariata modeller för att försöka förklara de svårmätbara variablerna. Resultatet visar att det går att erhålla information om variablerna med PLS-modeller som bygger på mer lättillgänglig data. De framtagna modellerna fungerade bäst för att förklara inkommande kväve, men för att verkligen säkerställa modellernas riktighet bör ytterligare validering ske. / Studies of real processes are based on measured data. In the past, the amount of available data was very limited. However, with modern technology, the information which is possible to obtain from measurements is more available, which considerably alters the possibility to understand and describe processes. Multivariate analysis is often used when large datasets which contains many variables are evaluated. In this thesis, the multivariate analysis methods PCA (principal component analysis) and PLS (partial least squares projection to latent structures) has been applied to wastewater data collected at Hammarby Sjöstadsverk WWTP (wastewater treatment plant). Wastewater treatment plants are required to monitor and control their systems in order to reduce their environmental impact. With improved knowledge of the processes involved, the impact can be significantly decreased without affecting the plant efficiency. Several variables are easy to measure directly in the water, while other require extensive laboratory analysis. Some of the parameters from the latter category are the contents of phosphorus and nitrogen in the water, both of which are important for the wastewater treatment results. The concentrations of these substances in the inlet water vary during the day and are difficult to monitor properly. The purpose of this study was to investigate whether it is possible, from the more easily measured variables, to obtain information on those which require more extensive analysis. This was done by using multivariate analysis to create models attempting to explain the variation in these variables. The models are commonly referred to as soft sensors, since they don’t actually make use of any physical sensors to measure the relevant variable. Data were collected during the period of March 11 to March 15, 2013 in the wastewater at different stages of the treatment process and a number of multivariate models were created. The result shows that it is possible to obtain information about the variables with PLS models based on easy-to-measure variables. The best created model was the one explaining the concentration of nitrogen in the inlet water.