<p>In industrial control systems PID-control remains the prevalent strategy, also for processes that would benefit from model based control. The purpose of this thesis is to evaluate whether model based control can be readily implemented in an industrial control system. To this end a simulated surge tank with a simulated industrial control system is studied. For evaluation two scenarios with specified objectives are selected.</p><p>Following a review of LQR and versions of MPC, Predictive Functional Control (PFC) is considered the most suitable for implementation. PFC is a form of MPC developed with industrial applications in mind and therefore has several advantages for implementation in an industrial control system. Controllers for the surge tank-system are developed and implemented in the control system.</p><p>Basic analysis of stability, sensitivity and robustness suggests that PFC has some advantages that might be important in a non-simulated implementation. Compared with PID-controllers adjusted for control performance, PFC does not show any notable improvements in performance.</p><p>In conclusion, it is possible to implement model based control in an industrial control system and with PFC the implementation is considered easy.</p>

Recent technical progress has made new forms of controller implementations on computers possible. Especially the technique of Fuzzy Logic Control has found a growing number of applications. Also the development of fast A/D converters has made the acquisation of data with vision based systems possible. In this project we combine the two techniques of Fuzzy Logic Control and Vision Feedback to control an inverted pendulum and to determine their usefulness and limitations. The experiment was conducted and provided us with the data necessary to judge the performance of the new control strategy. The gathered data support the hypothesis that it is possible to control the inverted pendulum with Fuzzy Logic Control using Vision Feedback, though not without limitations. / Graduation date: 1995

Most manufacturing processes produce parts that can only be correctly measured after the process cycle has been completed. Even if in-process measurement and control is possible, it is often too expensive or complex to practically implement. In this paper, a simple control scheme based on output measurement and input change after each processing cycle is proposed. It is shown to reduce the process dynamics to a simple gain with a delay, and reduce the control problem to a SISO discrete time problem. The goal of the controller is to both reduce mean output errors and reduce their variance. In so doing the process capability (e.g. Cpk) can be increased without additional investment in control hardware or in-process sensors. This control system is analyzed for two types of disturbance processes: independent (uncorrelated) and dependent (correlated). For the former the closed-loop control increased the output variance, whereas for the latter it can decrease it significantly. In both cases, proper controller design can reduce the mean error to zero without introducing poor transient performance. These finding were demonstrated by implementing Cycle to Cycle (CtC) control on a simple bending process (uncorrelated disturbance) and on an injection molding process (correlated disturbance). The results followed closely those predicted by the analysis. / Singapore-MIT Alliance (SMA)

The Chemical Mechanical polishing (CMP) process is now widely employed in the Integrated Circuit Fabrication. However, due to the complexity of process parameters on the material removal rate (MRR), mechanism of material removal and pattern effect are not well understood. In this paper, three contact regimes between the wafer surface and the polishing pad were proposed: direct contact, mixed or partial contact, and hydroplaning. The interfacial friction force has been employed to characterize these contact conditions. Several polishing models are reviewed with emphasis on the mechanical aspects of CMP. Experiments have been conducted to verify the mechanical polishing models and to identify the dominant mechanism of material removal under typical CMP conditions. / Singapore-MIT Alliance (SMA)

In this thesis, a Reinforcement Learning (RL) method called Sarsa is used to dynamically tune a PI-controller for a Continuous Stirred Tank Heater (CSTH) experimental setup. The proposed approach uses an approximate model to train the RL agent in the simulation environment before implementation on the real plant. This is done in order to help the RL agent initially start from a reasonably stable policy. Learning without any information about the dynamics of the process is not practically feasible due to the great amount of data (time) that the RL algorithm requires and safety issues. The process in this thesis is modeled with a First Order Plus Time Delay (FOPTD) transfer function, because almost all of the chemical processes can be sufficiently represented by this class of transfer functions. The presence of a delay term in this type of transfer functions makes them inherently more complicated models for RL methods. RL methods should be combined with generalization techniques to handle the continuous state space. Here, parameterized quadratic function approximation compounded with k-nearest neighborhood function approximation is used for the regions close and far from the origin, respectively. Applying each of these generalization methods separately has some disadvantages, hence their combination is used to overcome these flaws. The proposed RL-based PI-controller is initially trained in the simulation environment. Thereafter, the policy of the simulation-based RL agent is used as the starting policy of the RL agent during implementation on the experimental setup. As a result of the existing plant-model mismatch, the performance of the RL-based PI-controller using this primary policy is not as good as the simulationresults; however, training on the real plant results in a significant improvement in this performance. On the other hand, the IMC-tuned PI-controllers, which are the most commonly used feedback controllers are also compared and they also degrade because of the inevitable plant-model mismatch. To improve the performance of these IMC-tuned PI-controllers, re-tuning of these controllers based on a more precise model of the process is necessary. The experimental tests are carried out for the cases of set-point tracking and disturbance rejection. In both cases, the successful adaptability of the RL-based PI-controller is clearly evident. Finally, in the case of a disturbance entering the process, the performance of the proposed model-free self-tuning PI-controller degrades more, when compared to the existing IMC controllers. However, the adaptability of the RL-based PI- controller provides a good solution to this problem. After being trained to handle disturbances in the process, an improved control policy is obtained, which is able to successfully return the output to the set-point. / Process Control

Thesis (M.Eng.)--Memorial University of Newfoundland, 2005. / Bibliography: leaves 124-128.

The main objective of this thesis is to use available knowledge about a process and combine this with measurement data from the same process to extract more information about the process. The combination of knowledge and measurement data is referred to as Multi Sensor Data Fusion, MSDF. This added information is then used to control the process towards a specified goal. The process studied in this thesis is the process of drilling wells in a petroleum reservoir, while the oil is flowing from the reservoir. In the petroleum industry, this is defined as underbalanced drilling (UBD), where the bottom hole pressure (BHP) in the well is below the pore pressure in the reservoir. Detailed knowledge of the process is of paramount importance when using multi sensor data fusion. Due to this, various process modelling efforts are examined and evaluated, from simple relations between parameters to a finite-element approach of modelling the fluid flow in the well during drilling. Several sensors are used in the various cases, and existing sensors such as pressure sensors and flow sensors are the main data source in the analysis. Future scenario with sensors such as pressure arrays and non-intrusive multiphase flow meters are evaluated. In addition, new positions of existing sensor systems are discussed. The methods available for fusing the knowledge of the process represented as models together with the available data is ranging from artificial intelligent methods such as neural networks, to methods incorporating statistical analysis such as various Kalman filters. History matching techniques using gradient techniques are also examined. The migration of reservoir fluids into the well during UBD influences the BHP of the well. The results in the thesis show that this reservoir influx can be calculated by estimating some of the important reservoir parameters such as reservoir pore pressure or reservoir permeability. These reservoir parameters can be estimated most efficiently by performing an MSDF using a detailed nonlinear model of the well and reservoir dynamic behaviour together with real-time measurements of the fluid flow parameters such as fluid temperature, fluid pressure and fluid flow rates. The unscented Kalman filter shows the best performance when evaluating both estimation accuracy and computational requirements. Regarding available instrumentation for use during UBD, the analysis shows that there is a major potential in introducing new sensors. As new data transmission methods are emerging and making data from sensors distributed along the drillstring available, this can generate a shift in paradigm regarding real-time analysis of reservoir properties during drilling. Controlling the process is an important usage of the information gained from the MSDF analysis. Various control methods for controlling the most important process variables are examined and evaluated. The results show that acceptable pressure control can be obtained when using the choke valve opening as the primary control parameter. However, the choke valve operation has to be closely coordinated with drilling fluid flow rate adjustments. The choke valve opening control is able to compensate for pressure variations during the whole drilling operation. A suggested nonlinear model predictive control algorithm gives best results when looking at the control accuracy, and can easily be expanded to handle multiple control inputs and system constraints. This control algorithm uses a detailed model of the well and reservoir dynamics. The Levenberg-Marquardt algorithm is used to calculate the optimal future control variables. The main drawback of the control algorithm is computational burden. A linear control algorithm, which also is evaluated, uses less computational resources, but has less control accuracy and is more difficult to expand into a multivariable control system. Recommendations for further work are to expand the suggested model predictive control algorithm to handle more control inputs, while reducing the computational burden by incorporating low-order models for describing the future behaviour of the well.

Recent developments in material joining, specifically arc-welding, have increased in scope and extended into the aerospace, nuclear, and underwater industries where complex geometry and hazardous environments necessitate fully automated systems. Even traditional applications of arc welding such as off-highway and automotive manufacturing have increased their demand in quality, accuracy, and volume to stay competitive. These requirements often exceed both skill and endurance capacities of human welders. As a result, improvements in process parameter feedback and sensing are necessary to successfully achieve a closed-loop control of such processes. <br ><br /> One such feedback parameter in gas-metal arc welding (GMAW) is acoustic emissions. Although there have been relatively few studies performed in this area, it is agreed amongst professional welders that the sound from an arc is critical to their ability to control the process. Investigations that have been performed however, have been met with mixed success due to extraneous background noises or inadequate evaluation of the signal spectral content. However, if it were possible to identify the salient or characterizing aspects of the signal, these drawbacks may be overcome. <br ><br /> The goal of this thesis is to develop methods which characterize the arc-acoustic signal such that a relationship can be drawn between welding parameters and acoustic spectral characteristics. Three methods were attempted including: Taguchi experiments to reveal trends between weld process parameters and the acoustic signal; psycho-acoustic experiments that investigate expert welder reliance on arc-sounds, and implementation of an artificial neural network (ANN) for mapping arc-acoustic spectral characteristics to process parameters. <br ><br /> Together, these investigations revealed strong correlation between welding voltage and arc-acoustics. The psycho-acoustic experiments confirm the suspicion of welder reliance on arc-acoustics as well as potential spectral candidates necessary to spray-transfer control during GMA welding. ANN performance shows promise in the approach and confirmation of the ANN?s ability to learn. Further experimentation and data gathering to enrich the learning data-base will be necessary to apply artificial intelligence such as artificial neural networks to such a stochastic and non-linear relationship between arc-sound and GMA parameters.

In industrial control systems PID-control remains the prevalent strategy, also for processes that would benefit from model based control. The purpose of this thesis is to evaluate whether model based control can be readily implemented in an industrial control system. To this end a simulated surge tank with a simulated industrial control system is studied. For evaluation two scenarios with specified objectives are selected. Following a review of LQR and versions of MPC, Predictive Functional Control (PFC) is considered the most suitable for implementation. PFC is a form of MPC developed with industrial applications in mind and therefore has several advantages for implementation in an industrial control system. Controllers for the surge tank-system are developed and implemented in the control system. Basic analysis of stability, sensitivity and robustness suggests that PFC has some advantages that might be important in a non-simulated implementation. Compared with PID-controllers adjusted for control performance, PFC does not show any notable improvements in performance. In conclusion, it is possible to implement model based control in an industrial control system and with PFC the implementation is considered easy.

Recent developments in material joining, specifically arc-welding, have increased in scope and extended into the aerospace, nuclear, and underwater industries where complex geometry and hazardous environments necessitate fully automated systems. Even traditional applications of arc welding such as off-highway and automotive manufacturing have increased their demand in quality, accuracy, and volume to stay competitive. These requirements often exceed both skill and endurance capacities of human welders. As a result, improvements in process parameter feedback and sensing are necessary to successfully achieve a closed-loop control of such processes. <br ><br /> One such feedback parameter in gas-metal arc welding (GMAW) is acoustic emissions. Although there have been relatively few studies performed in this area, it is agreed amongst professional welders that the sound from an arc is critical to their ability to control the process. Investigations that have been performed however, have been met with mixed success due to extraneous background noises or inadequate evaluation of the signal spectral content. However, if it were possible to identify the salient or characterizing aspects of the signal, these drawbacks may be overcome. <br ><br /> The goal of this thesis is to develop methods which characterize the arc-acoustic signal such that a relationship can be drawn between welding parameters and acoustic spectral characteristics. Three methods were attempted including: Taguchi experiments to reveal trends between weld process parameters and the acoustic signal; psycho-acoustic experiments that investigate expert welder reliance on arc-sounds, and implementation of an artificial neural network (ANN) for mapping arc-acoustic spectral characteristics to process parameters. <br ><br /> Together, these investigations revealed strong correlation between welding voltage and arc-acoustics. The psycho-acoustic experiments confirm the suspicion of welder reliance on arc-acoustics as well as potential spectral candidates necessary to spray-transfer control during GMA welding. ANN performance shows promise in the approach and confirmation of the ANN?s ability to learn. Further experimentation and data gathering to enrich the learning data-base will be necessary to apply artificial intelligence such as artificial neural networks to such a stochastic and non-linear relationship between arc-sound and GMA parameters.