Guy A Dumont

Particle accelerators are mainly used as research tools for various scientific topics such as subatomic particles, astrophysical nuclear reactions, molecular and materials science. Apart from being research tools, particle accelerators can also be used for various medical purposes. The work in this thesis motivated by the control requirement for the electron linear accelerator (E-LINAC) of Advanced Rare IsotopE Laboratory (ARIEL), currently under construction in TRIUMF, Canada’s particle accelerator center. In the e-linac, electrons are accelerated up to 50 MeV along a linear beamline. Radiofrequency (RF) cavity resonators are metallic structures that accelerate chargedparticles along a linear axis, by providing an oscillating electric field known as the accelerating field. The particles are accelerated in bunches arriving at the cavity when the accelerating field is close to its maximum. In order to deliver a high quality beam, it is desired that all the particles receive the same amount of energy from the RF cavities. Therefore, the amplitude and phase of the accelerating field should be precisely controlled. When a bunch of particles pass through the cavity, energy is transferred from the cavity to the beam. As a result of the interaction between the cavity field andthe beam, the accelerating field decreases. This phenomenon is called the beam loading effect. The control goal is to cancel the beam loading effect so that it does not affect the acceleration of the following particles. The feedback controller in the loop is not fast enough to meet the control requirements. Therefore, a feedforward controller is needed for fast response. In this thesis, an Iterative learning control (ILC) system is developed to cancel the beam loading effect. Two implementations of ILC are presented, filter-based ILC and norm-optimal ILC. The transient response of filter-based ILC is analyzed using pseudospectra, the eigenvalues of the perturbed transfer matrix. The norm-optimal ILC is designed as a finite-state machine (FSM) to be implemented on field-programmable gate array (FPGA) hardware. The filter-based ILC is tested on the cavities in TRIUMF e-linac, and the test results show successful cancellation of the beam loading effect, meeting the control requirements.

View record

To convince the public to employ new technologies in their daily life, the reliability and safety of such technologies must be demonstrated. Verification of safety is more crucial in applications which may directly jeopardize human safety. A self- driving car is an example of such a technology in which the safety of passengers and pedestrians depends on the car’s actions made based on its perception of the environment. A closed-loop drug delivery system is another example of a safety-critical technology for which there are risks of drug over/under-dosing and adverse side effects. Uncertainty is the key challenge in safety verification of these technologies. In the self-driving car example, the uncertainty is in the car’s knowledge of the environment. In the context of closed-loop drug delivery systems, uncertainty is in the patient response to drugs due to inter-patient variability (model uncertainty). Unmeasurablity of variables indicating the system’s safety (e.g. drug concentrations in the plasma) is another source of problems in safety verification of technologies such as closed-loop anesthesia. Motivated by closed-loop anesthesia, this thesis aims to develop a mathematical framework for formal safety verification and safety-preserving control of safety-critical systems with model uncertainty. This extends formal methods and existing safety-preserving control techniques to systems with a certain class of multiplicative model uncertainty. Moreover, this work proposes one of the first safety- preserving control schemes for output-feedback control systems in which safety- critical variables are not measurable. We also extend this technique to uncertain output-feedback systems. In this work, we employ the developed techniques to design and formalize a safety system for closed-loop anesthesia. Finally, we propose a novel approach to reduce conservatism of the formalized safety system by reducing model uncertainty using model falsification. The results discussed in this thesis may facilitate the process of obtaining regulatory approvals for closed-loop anesthesia, which helps the emergence of this technology (and other closed-loop drug delivery systems) in clinical environments.

View record

We developed novel algorithms for monitoring sleep, sleep breathing disorder (SBD)and instantaneous respiratory rate (IRR) in children using the characterization ofpulse oximetry photoplethysmogram (PPG). To evaluate the algorithms, we recordedthe oxygen saturation (SpO₂) and PPG signals from 160 children using a phone-basedoximeter consisting of a microcontroller-based pulse oximeter module interfacinga smartphone. This mobile oximeter was further developed to perform allprocessing on the smartphone through the audio interface.We evaluated the relative impact of SBD on sympathetic and parasympatheticactivity in children through the characterization of PPG and concluded that sympatheticactivity was higher in 30-second epochs with apnea/hypopnea event(s). Welater characterized the SpO₂ pattern in SDB and then combined SpO₂ pattern characterizationand PPG analysis to design a model with two binary logistic classifiersto identify the epochs with apnea/hypopnea events.We developed a novel model for identifying the cycles of random eye movement(REM) and non-REM of the overnight sleep based on the activity of cardiorespiratorysystem using the overnight PPG. We extracted the features associated withpulse rate variability (PRV), respiratory rate (RR), vascular tone and movementfrom PPG to build a model with two binary classifiers to identify wakefulness fromsleep (wake/sleep classifier) and REM from non-REM sleep (non-REM/REM classifier).We also developed a novel algorithm for extracting the instantaneous respiratoryrate (IRR) from PPG. The algorithm was performed in three steps: extractionof three respiratory-induced variation signals from PPG, estimation of IRR fromeach extracted respiratory-induced variation signal and fusion of IRR estimates. A time-frequency transform called synchrosqueezing transform (SST) was usedto extract the respiratory-induced variation signals from PPG. Later, a second SSTwas applied to estimate IRR from respiratory-induced variation signals. To fuseIRR estimates, a novel algorithm was proposed.This study would expand the functionality of conventional pulse oximetry beyondthe measurement of heart rate and SpO₂ to monitor sleep, to screen SBDs andmeasure the respiratory rate continuously and instantly.

View record

This thesis investigates the design and performance of a model predictive controller (MPC) for the automatic control of hypnosis. It constitutes the first step towards automatic control of anesthesia with constraints on important parameters such as drug concentrations in the body and hemodynamic variables such as blood pressure. The literature suggests, that closed-loop control of anesthesia can significantly reduce drug consumption and lessen recovery times, thus improving the safety and quality of anesthesia care while reducing costs. However, automation of anesthesia is challenging because of shortcomings associated with drug-response modeling, in particular limited data for children and disagreement between published models, inadequate predictive capacity of models owing to inclusion of monitor dynamics in the models, and significant inter/intra patient variability and uncertainty in models. The first part of this thesis introduces a new approach to dose-response modeling and presents models with different clinical end-points for propofol in children and adults. This thesis also presents a new monitor-decoupledmodel of propofol pharmacodynamics (PD) where the monitor model is clearly excluded from the identified PD. The second part of the thesis concentrates on design of a constrained MPC for hypnosis. While the anesthesia closed-loop concept has already been investigated in the past, there is still a need for a closed-loop control system that explicitly includes robustness in the design step, allows constraints on drug concentrations and physiological parameters, and can incorporate multivariable control of multi drug and multi sensor systems. In this thesis, robust MPC controllers are presented for closed-loop control of depth of hypnosis in adults and children. Robustness in the presence of inter-patient variability is taken into account in the controller design. A novel idea is introduced on how to define and implement physiological constraints in closed-loop control of hypnosis using MPC with a parallel PKPD model. Evaluation of the proposed MPC meets the design specifications and shows that the required robustness against patient uncertainty is achieved and the proposed safety constrained control strategy can potentially reduce the risk of under/over-dosing for most patients by providing controller enforced safety bounds without sacrificing the performance of the closed-loop control system.

View record

For continuous-time LTI systems under human control and under shared control, this thesis studies techniques to determine whether or not a given user-interface provides the information required to accomplish a certain task. It is well known that attaining Situation Awareness (SA) is essential to the safe operation of the systems involving human-automation interaction. Hence, through two different approaches, the work in this thesis evaluates and designs user-interfaces based on the satisfaction of SA requirements by the user. In the first approach, observability-based conditions under which a user-interface provides the user with adequate information to accomplish a given task are identified. The user is considered to be a special type of observer, with capabilities corresponding to different levels of knowledge regarding the input and output derivatives. Through this approach, the ``user-observable/user-predictable'' subspaces for systems under shared control are defined and formulated. In addition, state estimation is considered to incorporate a processing delay. Hence, the ``delay-incorporating user-observable/user-predictable'' subspaces are formulated and are compared with the space spanned by the combination of the states which create the task. If the task subspace does not lie in the relevant space, then the user-interface is incorrect, meaning that the user cannot accomplish the desired task with the given user-interface. In the second approach, to determine the required information to be displayed, a model of attaining SA for the users is proposed. In this model, the user is modeled as an extended delayed functional estimator. Then, the information needed for such an estimator to make correct estimations as well as the desired expansion of the functional of the states to let the user precisely reconstruct and accurately predict the desired task is determined. Additionally, it is considered that in practice, to attain the situation awareness, the estimation of the task states does not necessarily need to be precise but can be bounded within certain margins. Hence, the model of the user attaining SA is also modified as a ``bounded-error delayed functional observation/prediction''. Such an observer/predictor has to exist for a system with a given user-interface, otherwise, the safety of the operation may be compromised.

View record

We have developed a novel real-time cardiorespiratory coherence (CRC) algorithm to monitor nociception during general anesthesia. We have made two novel and significant contributions to the field.We have developed a novel filter for measuring the respiratory sinus arrhythmia (RSA) in the heart rate variability (HRV) in real-time. The filter uses information from a secondary signal source (a respiration rate derived from a respiration signal) to track the RSA as it moves in time and frequency. It then uses this information to dynamically tune the centre frequency and bandwidth of a band-pass filter to isolate and measure the RSA. This filter is very effective at tracking the RSA in time and frequency, and it may provide the most robust measure of RSA yet devised.We have integrated multiple signals and algorithms together into an end-to-end software system for robust continuous real-time nociception monitoring during general anesthesia. The software system incorporates not only our novel RSA filter to measure nociception, but also many peripheral algorithms for detecting and rejecting artifacts in the input signals. The input signals required for real-time nociception monitoring can be extremely noisy, and artifacts are a very significant challenge. To our knowledge, no other nociception monitor includes such robust artifact handling using redundant signal channels.We estimated the sensitivity of our real-time CRC algorithm to nociception and antinociception, and compared it to traditional univariate HRV measures and standard clinical vital signs. Following ethics approval and informed parental consent, data were collected from 48 children receiving general anesthesia during dental surgery. A total of 42 dental dam insertion (nociception) and 57 anesthetic bolus (antinociception) events were noted. A nociception index was created for each HRV algorithm, ranging from 0 (no nociception) to 100 (strong nociception). Dental dam insertion changed the CRC nociception index by an average of 27 [95% CI from 21 to 33] (P
View record

Autonomic-cardiac regulation operates through interactions between the autonomic nervous system (ANS) and the cardiovascular system (CVS). In order to maintain homeostasis in the CVS, the ANS adjusts it effectors, such as the stiffness of blood vessels and the pace of heartbeats, against physical and psychological stressors, so that it can maintain adequate blood flow. This allows oxygen and nutrients to be delivered to organs and enables the performance of other essential functions.Autonomic-cardiac regulation can be described by a mathematical model and it can be analyzed under different scenarios such as a stressful condition or an increased arterial stiffness. This may help researchers to obtain new understandings of the autonomic-cardiac regulation. This thesis is built upon a physiology-based mathematical model of autonomic-cardiac regulation describing the regulation of heart rate (HR) and blood pressure (BP), using a set of nonlinear, coupled differential equations with delay.Non-invasive and subject-specific monitoring of autonomic-cardiac regulation has the potential to improve current treatments of autonomic-cardiac disorders. A parameter estimation method has been used to specify time-varying subject-specific model parameters associated with autonomic-cardiac regulation. The proposed method will help to improve monitoring of autonomic-cardiac variables, such as sympathetic and parasympathetic nerve activities affecting the heart and sympathetic nerve activity affecting the arterial tree.The complex dynamic interactions between nonlinearities and delays in the autonomic-cardiac regulation may result in the onset of instabilities in BP and HR regulation. In this thesis, we propose a model-based approach to stability analysis and introduce a quantitative stability indicator of the autonomic-cardiac regulation. We can prevent irregularities in cardiovascular rhythms (e.g., HR and BP) by knowing their causes and developing an intelligent method to control them.An artificial bionic baroreflex can be an effective treatment for baroreflex failure in, for example, individuals with severe orthostatic hypotension. We propose a method to design an artificial bionic baroreflex by mimicking the baroreflex mechanism in the body. This could then be potentially used to adjust existing neurostimulator devices that regulate BP.

View record

The focus of this thesis is to develop an advanced control system for existing and to-be developed Thermo-Mechanical Pulping (TMP) refining processes. Therefore, the thesis has two parts. The first part developed two different but yet complementary closed-loop identification methods to update the models used in the low-level control layer of the current advanced control systems for TMP refiners. These newly developed identification methods are specific to closed-loop systems controlled by Model Predictive Control (MPC) techniques, and successfully tailored to the presence of the MPC. By updating the existing process models, the current advanced control systems can be operated to give better performance. However, these control systems may not be able to provide optimal performance due to process disturbances. Hence, a novel economically oriented advanced control system is developed for the existing two-stage TMP refiner processes for their optimal operation when disturbances are present. This novel technique dynamically optimizes the TMP processes as opposed to conventional two-layer methods which perform process optimization at steady-state, and has shown a potential economical benefit through reduction of total specific energy of a two-stage TMP process in a simulation study.In the second part, a novel TMP process with multi-stages of Low Consistency (LC) refining is studied for its optimal operation. The proposed Nonlinear Model Predictive Control (NMPC) technique dynamically optimizes the process and provides better performance when disturbances are present. This economically oriented NMPC (econNMPC) minimizes the total specific energy consumptionof the process while respecting all the process constraints and achieving final desired pulp quality. In simulation studies, a TMP process with multiple stages of LC refining was able to save significant specific energy consumption with setpoint tracking control when disturbances are present. Moreover, further reduction of specific energy has been achieved with the developed econNMPC technique.

View record

Oxygen is a critical component in living organisms and its concentration in tissue is an important parameter indicative of tissue metabolism, level of activity and health condition. As a result, measuring oxygen concentration in the tissue is essential in many clinical and research applications. Near Infrared Spectroscopy (NIRS)is a non invasive method of measuring tissue oxygenation using diffusion of light in the tissue. NIRS as a safe, non invasive and low cost monitoring technology has been used in a wide range of applications including monitoring muscle andbrain oxygenation, brain computer interface and rehabilitation. The motivation for this thesis has been to develop new signal processing methods and to investigate potential new applications for NIRS.One major characteristic of NIRS is its sensitivity to movement of the target tissue during the measurement. The effects of movements, known as motion artifacts, have limited clinical applications of NIRS in ambulant patients as well as experimental applications of NIRS monitoring in areas such as exercise science and sports medicine. In this thesis, we present a new method of reducing the effect of motion artifacts on NIRS signal using Discrete Wavelet Transform (DWT).One of the areas of application which can significantly benefit from reduction of motion artifacts is NIRS-based wearable sensors. In particular, a potential and unexplored application of NIRS is providing a monitoring method for people with bladder control problems, which occurs in a variety of conditions including spinal cord injury and stroke. We investigate the application of NIRS for detection of bladder filling to capacity using a wearable wireless monitoring sensor which canbe used to warn the subject once the bladder content reaches a predefined percentageof the full capacity. NIRS can be used as a functional neuroimaging method to identify brain activations during practice of a motor/cognitive task. One important question in this field is how the activated brain areas are interconnected. We thus investigate the use of phase information in NIRS channels to identify cortical connections and in particular, show the applicability of this approach in identifying language networkin human infants.

View record

Controlling and eliminating defects, such as macro-porosity, in die casting processes is an on-going challenge for manufacturers. Current strategies for eliminating macro-porosity focus on the execution of pre-set casting cycles, die structure design or the combination of both. To respond to process variability and mitigate its negative effects, advanced process control methodology has been developed to dynamically drive the process towards optimal dynamic or static operational conditions, hence minimizing macro-porosity in the casting.In this thesis, a Finite Element heat transfer model has been developed to predict the evolution of temperatures and the volume of encapsulated liquid in a casting with a high propensity to form macro-porosity. The model was validated by comparison to plant trial data. A virtual process has then been developed based on the model to simulate the continuous operation of a real process, for use as a platform to evaluate a controller’s performance.Since macro-porosity cannot be measured during casting, die temperature has been used as an indirect indicator of this defect. A model-based methodology has been developed to analyze the correlation between die temperature and encapsulated liquid volume, a precursor to the formation of macro-porosity. This methodology is employed to assess the suitability of different in-cycle die temperatures for use as indicators of macro-porosity formation. The optimal locations have then been determined to monitor die temperatures for the purpose of minimizing macro-porosity.A nonlinear state-space model, based on data from the virtual process, has been developed to provide a reliable representation of this virtual process. The control variable-driven portion exhibits linear dynamic behavior with nonlinear static gain. The resulting MIMO state-space model facilitated the design of a controller for this process.Finally, the performance of the nonlinear model-based predictive controller was evaluated using the virtual process. Independent of the initial state of the process - i.e. steady state or startup, the controller exhibited the capability to automatically adjust the process toward the dynamic or static optimal operational condition during disturbances examined. The advanced control methodology developed for LPDC provides a novel solution to improve the operational conditions in die casting process.

View record

The thesis focuses on mixing fluids, and it has three main parts. The first part investigates mixing in agitated pulp chests. A new model for identifying the performance of mixing in the pulp chest is developed. This newly developed model simplifies the previous pulp chest model and successfully describes non-ideal flow behavior, including channeling and dead-zones which occurs in the agitated pulp chests. The model is verified through experimental data obtained in a laboratory-scale agitated pulp chest.In the second part of the thesis, a novel method for identifying residence time (the average time that it takes for material to exit the mixer) from input and output data is developed. The main benefit of this new method is that it does not require that an individual have prior knowledge of the process to evaluate the mixing inside the flow system. The same idea is used to estimate the higher moments of linear time invariant transfer functions from a batch of input and output data. The ability to estimate higher moments of a transfer function enables one to reconstruct the original transfer function without having knowledge of its structure. This opens new possibilities in non-parametric system identification and residence-time theory. It is also shown how a bound on the delay of all pole transfer functions can be found using the first and second moments.In the third part of the thesis, a new class of mixers is designed. These mixers do not have agitators, yet they are as effective as stirred tanks in reducing fluctuations in their inlet stream. However, they are only effective for reducing the concentration fluctuations in the flow direction (axial mixing). This design splits the fluid into different channels and recombines these streams to achieve mixing. These mixers are especially useful in microfluidic applications where there is a physical limitation in using agitators for mixing; therefore, in the final section of the thesis, a mixer is designed and its geometry is calculated for microfluidic purposes. Moreover, the effectiveness of the mixer in reducing inlet concentration fluctuations is shown through computational fluid dynamic simulations (CFD).

View record

As a chronic neurological disorder, epilepsy is associated with recurrent, unprovoked epileptic seizures resulting from a sudden disturbance of brain function. Long-term monitoring of epileptic patients' Electroencephalogram (EEG) is often needed for diagnosis of seizures, which is tedious, expensive, and time-consuming. Also, clinical staff may not identify the seizure early enough to determine the semiology at the onset. This motivates EEG-based automated real-time detection of seizures. Apart from their possible severe side effects, common treatments for epilepsy (medication and surgery) fail to satisfactorily control seizures in ~25% of patients. EEG-based seizure prediction systems would significantly enhance the chance of controlling/aborting seizures and improve safety and quality of life for patients. This thesis proposes novel EEG-based patient-specific techniques for real-time detection and prediction of epileptic seizures and also presents a pilot study of scalp EEGs acquired in a unique low-noise underground environment.The proposed detection method is based on the wavelet packet analysis of EEG. A novel index, termed the combined seizure index, is introduced which is sensitive to both the rhythmicity and relative energy of the EEG in a given channel and considers the consistency among different channels at the same time. This index is monitored by a cumulative sum procedure in each channel. This channel-based information is then used to generate the final seizure alarm.In this thesis, a prediction method based on a variational Bayesian Gaussian mixture model of the EEG positive zero-crossing intervals is proposed. Novel indices of similarity and dissimilarity are introduced to compare current observations with the preictal and interictal references and monitor the changes for each channel. Information from individual channels is finally combined to trigger an alarm for upcoming seizures.These methods are evaluated using scalp EEG data. The prediction method is also tested against a random predictor. Finally, this thesis investigates the capability of an ultra-shielded underground capsule for acquiring clean EEG. Results demonstrate the potential of the capsule for novel EEG studies, including establishing novel low-noise EEG benchmarks which could be helpful in better understanding of the brain functions and mechanisms deriving various brain disorders, such as epilepsy.

View record

Cylindrical agitated chests are frequently used to facilitate manufacturing processes in pulp and paper industry and one of their main functions is to attenuate any process disturbances. However, owing to the inherited non-Newtonian nature of pulp suspensions, it is not easy to achieve complete mixing and with the improper chest design, these agitated chests do not always perform ideally or satisfactorily. The cavern formation in incomplete mixing may induce bypassing and dead zones, which significantly affect the chest performance. A study of cavern formation in a cylindrical agitated chest was thus carried out. Also, a dynamic model developed by Soltanzadeh et al. (2009) was used to quantify the mixing dynamics of the cylindrical chest. In addition, using computational fluid dynamics (CFD), the simulated results of the flow in the chest were compared with the experimental results to verify the applicability of the CFD model on the study of pulp suspension agitation.To investigate the cavern formation in a lab-scale cylindrical chest, electrical resistance tomography (ERT) and ultrasonic Doppler velocimetry (UDV) were applied to estimate the cavern shape and size. Both methods gave satisfactory results and as expected, the cavern size was found to increase with impeller speeds. The cavern shape was best described as a truncated right-circular cylinder. Based on this observation, a model considering the interaction between the cavern and chest walls was developed to calculate the cavern volume.With the dynamic model, a series of dynamic tests were carried out to characterize the mixing behavior of the lab-scale cylindrical chest. It was found that the proposed flow configuration with the outlet close to the cavern could minimize the bypassing which affects mixing quality. Also, ERT verified the presence of cavern and dead zone when the chest was not completely agitated in continuous-flow operation.Numerical simulations using CFD were compared with the experimental results under different operating conditions. Pulp suspensions are a mixture of water and wood fibres that can entangle each other to form flocs affecting the mixing flow. Owing to this complex rheology, it is not easy to model the agitation precisely in CFD using a homogeneous fluid model. The floc formation and air entrapment observed in experiments were difficult to be numerically taken into account in the simulations. Although the CFD model could not exactly predict the mixing situation of pulp suspensions, it still can be used to estimate the mixing flow patterns, e.g., flow directions, in the proposed chest designs.

View record

According to the World Health Organization (WHO), cardiovascular (CV) diseases are the number one cause of death globally, and are projected to remain the single leading cause of death by 2030. Central pulse pressure and several arterial stiffness indices, especially aortic pulse wave velocity (PWV), are among the strongest predictors of CV events, including stroke, myocardial infarction (heart attack) and angina (chest pain). Therefore, non-invasive methods for the assessment of the central pressure and central arterial stiffness are very important for the diagnosis of CV diseases at their early stage of development.In this thesis, non-invasive methods are developed to (i) measure the aortic PWV, and (ii) measure the aortic pressure waveform (APW).In the intensive care unit, cardiac output (CO) is an important measure of the adequacy of circulation among post-operative patients of cardiovascular diseases. Current methods of monitoring use the heart rate, peripheral blood pressure, and urine output as surrogates of CO, but these indices are inadequate. CO can be monitored directly using thermodilution but the procedure is highly invasive.In this thesis, a minimally-invasive method is developed to monitor the CO using the radial artery pressure waveform.The underlying algorithms of the methods developed in this thesis are interrelated. In this thesis, the arterial system is modeled as a single uniform lossless transmission line terminated by a complex load.Each measuring and monitoring task is a problem of identification, simulation or data acquisition.The measurement of the aortic PWV is formulated as an identification problem. The method developed in this thesis was shown to be able to differentiate a group of children with Marfan syndrome from the healthy children.The monitoring of CO is formulated as a problem of identification and simulation. The method developed in this thesis was applied to five post-surgical infants. It showed clinically acceptable agreement with the more established echocardiographic technique.The measurement of APW is a problem of data acquisition. A method was developed to estimate APW from the aortic distension waveform obtained using B-mode ultrasound. This method showed good agreement with carotid artery applanation tonometry when applied to nine healthy children.

View record

Advances in monitoring technology have resulted in the collection of a vast amount of data that exceeds the simultaneous surveillance capabilities of expert clinicians in the clinical environment. To facilitate the clinical decision-making process, this thesis solves two fundamental problems in physiological monitoring: signal estimation and trend-pattern recognition. The general approach is to transform changes in different trend features to nonzero level-shifts by calculating the model-based forecast residuals and then to apply a statistical test or Bayesian approach on the residuals to detect changes. The EWMA-Cusum method describes a signal as the exponentially moving weighted average (EWMA) of historical data. This method is simple, robust, and applicable to most variables. The method based on the Dynamic Linear Model (refereed to as Adaptive-DLM method) describes a signal using the linear growth model combined with an EWMA model. An adaptive Kalman filter is used to estimate the second-order characteristics and adjust the change-detection process online. The Adaptive-DLM method is designed for monitoring variables measured at a high sampling rate. To address the intraoperative variability in variables measured at a low sampling rate, a generalized hidden Markov model is used to classify trend changes into different patterns and to describe the transition between these patterns as a first-order Markov-chain process. Trend patterns are recognized online with a quantitative evaluation of the occurrence probability. In addition to the univariate methods, a test statistic based on Factor Analysis is also proposed to investigate the inver-variable relationship and to reveal subtle clinical events. A novel hybrid median filter is also proposed to fuse heart-rate measurements from the ECG monitor, pulse oximeter, and arterial BP monitor to obtain accurate estimates of HR in the presence of artifacts. These methods have been tested using simulated and clinical data. The EWMA-Cusum and Adaptive-DLM methods have been implemented in a software system iAssist and evaluated by clinicians in the operating room. The results demonstrate that the proposed methods can effectively detect trend changes and assist clinicians in tracking the physiological state of a patient during surgery.

View record

A key issue in paper-machine Cross-Directional (CD) Control is alignment. Typically, this mapping problem is a non-linear and slowly time-varying phenomenon for individual machines. The first part of this thesis specifically focuses on different causes of the misalignment and reviews recent developments in CD mapping. It summarizes the results of a comprehensive survey of different alignment schemes that have been used in CD control processes. This dissertation presents a novel, deterministic, tensor-based modeling of the closed loop controlled CD process and an alignment method. First, we link the CD data to the tensor model. Exploiting this link, we derive a deterministic blind PARAFAC decomposition as an alignment method. The proposed PARAFAC capitalizes on the physical location of the actuators, scanning databoxes and their temporal diversities. Its performance is verified in several simulations then tested, evaluated and implemented as an alignment tool on real industrial paper machine.In the second section we present a new novel Adaptive Robust Control approach to the multivariable CD process of continuous web manufacturing. We have applied discretized Internal Model Control(IMC)-based classical Windsurfing to each individual separated spatial frequency. This approach allows the dynamical bandwidth of the closed-loop system to be increased progressively at each spatial frequency through an iterative control relevant system identification and control design procedure. The method deals with both model uncertainty and measurement noise issues. We also have applied Discrete-time H-infinity Windsurfing to each individual separated spatial frequency, starting with an initial model and a robust stabilizing controller at each spatial frequency. This modified approach provides robust stability through iterations. The approach is evaluated through a number of simulation experiments. Finally, in a closed-loop approach to modifying the existing industrial CD controller, the objective is turned to the modification of the Windsurfing method. An input signal is designed for the system identification and control design. The nu-gap approach to robust control is fed into the Windsurfing model. Since the nu-gap metric establishes the upper and lower bounds, this approach reduces the number of iterations before the design is completed.

View record

The aim of this work is to improve the identification of cross-directional (CD)models as well as to simplify the robust control design problem. Providingcomputationally efficient techniques for identification and robust control willfacilitate the implementation of an autonomous CD control system. These are challenging problems as CD processes are large scale distributed parameter systems. The conventional model for a CD process is a spatial interaction matrix cascaded by a low order transfer function in the temporal domain. This representation results in a large dimensional multi-variable model.Model uncertainties in the process are inevitable and rise from several sources. As the CD process models are usually identified by input-output data from bump tests, there is a demand for better identification techniques that minimizes the uncertainties in CD mapping and response shape models.Mapping and misalignment detection are problems that are specific to CD processes due to the configuration of the paper machine and its unconventional scanning method. These problems are of great practical importance for industrial implementations. This work proposes modeling the CD process as a two-dimensional (2D)system that is spatially noncausal. The spatial noncausal transfer functions facilitate input design and identification in the spatial domain. Both noncausal finite impulse response (FIR) models and rational transfer functions are used to model the CD response. The FIR model representation isconvenient for input design and identification in the CD while the spatialnoncausal rational transfer function is more suitable for robust control design. The 2D representation for the plant and the controller is convenient for implementation in an autonomous control scheme with iterative feedback tuning or adaptive control. Robust stability criteria are developed to investigate the stability of theprocess under feedback. Employing the 2D representation results in criteriathat are computationally efficient as the 2D stability conditions are replacedby a set of simple 1D problems. A robust stability criterion that is based on the v-gap stability criterion provides bounds for robust stability against perturbations in the plant or the controller. This feature permits designing a two-dimensional controller in an adaptive control scheme or a simple re-tuning of an existing controller through iterative feedback tuning. Thisproperty is convenient when switching between different grades of paper.Another stability criterion based on the concept of phase margins is developed to determine the closed-loop’s tolerance to misalignment uncertainties. Finally, a simple loop-shaping technique using spatial transfer functions is proposed to shape the closed-loop’s two-dimensional frequency response. The noncausal spatial transfer function provides a convenient tool for improving the closed-loop’s performance at low temporal frequencies. An adaptive control scheme is implemented to retune the system after a grade change. The developed techniques for identification and robust control can be used as a part of a more sophisticated autonomous control system.

View record