During fast point-to-point motion, the inherent joint flexibility could be detrimental in terms of residual vibration. Aiming to minimize the vibration, the command shaping method has been developed so as to remove critical energy from the input profile at resonant frequencies. Since this method requires information of a physical model in order to find the target frequencies, the quality of the shaped command profile relies on the accuracy of the model parameter estimation. Therefore, in this work, a system identification method using Instrumental Variables is applied from the literature. Compared with the classic Ordinary Least Square method, the IV approach has successfully improved the estimation of parameters, based on simulation results. The accuracy of parameter estimation influences the command profile, as does the feedback controller. In this work, starting from a mathematical derivation with a mismatch model due to a feedback controller called Computed Torque Control, insight for the closed-loop system is given with regard to the interaction between control gains and the actual resonant frequencies. It is found that the control gain is able to modify the actual resonant frequency curve, and push it into or out of the shaping bounds which are generated from the command shaping method. Further analysis based on the simulation results shows that the overlap area between the shaping bounds and the actual frequencies affects the level of residual vibration. In light of this fact, an optimal control gain exists and is found when the estimation error is in a certain range. At the end, recommendations for choosing the control gains are provided.

Traction efficiency of tractors barely reaches 50 % in field operations. On the other hand, modern trends in agriculture show growth of the global tractor markets and at the same time increased demands for greenhouse gas emission reduction as well as energy efficiency due to increasing fuel costs. Engine power of farm tractors is growing at 1.8 kW per year reaching today about 500 kW for the highest traction class machines. The problem of effective use of energy has become crucial. Existing slip control approaches for tractors do not fulfil this requirement due to fixed reference set-point. The present work suggests an optimal control scheme based on set-point optimization and on assessment of soil conditions, namely, wheel-ground parameter identification using fuzzy-logic-assisted adaptive unscented Kalman filter.:List of figures VIII List of tables IX Keywords XI List of abbreviations XII List of mathematical symbols XIII Indices XV 1 Introduction 1 1.1 Problem description and challenges 1 1.1.1 Development of agricultural industry 1 1.1.2 Power flows and energy efficiency of a farm tractor 2 1.2 Motivation 9 1.3 Purpose and approach 12 1.3.1 Purpose and goals 12 1.3.2 Brief description of methodology 14 1.3.2.1 Drive torque feedback 14 1.3.2.2 Measurement signals 15 1.3.2.3 Identification of traction parameters 15 1.3.2.4 Definition of optimal slip 15 1.4 Outline 16 2 State of the art in traction management and parameter estimation 17 2.1 Slip control for farm tractors 17 2.2 Acquisition of drive torque feedback 23 2.3 Tire-ground parameter estimation 25 2.3.1 Kalman filter 25 2.3.2 Extended Kalman filter 27 2.3.3 Unscented Kalman filter 27 2.3.4 Adaptation algorithms for Kalman filter 29 3 Modelling vehicle dynamics for traction control 31 3.1 Tire-soil interaction 31 3.1.1 Forces in wheel-ground contact 32 3.1.1.1 Vertical force 32 3.1.1.2 Tire-ground surface geometry 34 3.1.2 Longitudinal force 36 3.1.3 Zero-slip condition 37 3.1.3.1 Soil shear stress 38 3.1.3.2 Rolling resistance 39 3.2 Vehicle body and wheels 40 3.2.1 Short description of Multi-Body-Simulation 40 3.2.2 Vehicle body and wheel models 42 3.2.3 Wheel structure 43 3.3 Stochastic input signals 45 3.3.1 Influence of trend and low-frequency components 47 3.3.2 Modelling stochastic signals 49 3.4 Further components and general view of tractor model 53 3.4.1 Generator, intermediate circuit, electrical motors and braking resistor 53 3.4.2 Diesel engine 55 4 Identification of traction parameters 56 4.1 Description of identification approaches 56 4.2 Vehicle model 58 4.2.1 Vehicle longitudinal dynamics 58 4.2.2 Wheel rotational dynamics 59 4.2.3 Tire dynamic rolling radius and inner rolling resistance coefficient 60 4.2.4 Whole model 61 4.3 Static methods of parameter identification 63 4.4 Adaptation mechanism of the unscented Kalman filter 63 4.5 Fuzzy supervisor for the adaptive unscented Kalman filter 66 4.5.1 Structure of the fuzzy supervisor 67 4.5.2 Stability analysis of the adaptive unscented Kalman filter with the fuzzy supervisor 69 5 Optimal slip control 73 5.1 Approaches for slip control by means of traction control system 73 5.1.1 Feedback compensation law 73 5.1.2 Sliding mode control 74 5.1.3 Funnel control 77 5.1.4 Lyapunov-Candidate-Function-based control, other approaches and choice of algorithm 78 5.2 General description of optimal slip control algorithm 79 5.3 Estimation of traction force characteristic curves 82 5.4 Optimal slip set-point computation 85 6 Verification of identification and optimal slip control systems 91 6.1 Simulation results 91 6.1.1 Identification of traction parameters 91 6.1.1.1 Comparison of extended Kalman filter and unscented Kalman filter 92 6.1.1.2 Comparison of ordinary and adaptive unscented Kalman filters 96 6.1.1.3 Comparison of the adaptive unscented Kalman filter with the fuzzy supervisor and static methods 99 6.1.1.4 Description of soil conditions 100 6.1.1.5 Identification of traction parameters under changing soil conditions 101 6.1.2 Approximation of characteristic curves 102 6.1.3 Slip control with reference of 10% 103 6.1.4 Comparison of operating with fixed and optimal slip reference 104 6.2 Experimental verification 108 6.2.1 Setup and description of the experiments 108 6.2.2 Virtual slip control without load machine 109 6.2.3 Virtual slip control with load machine 113 7 Summary, conclusions and future challenges 122 7.1 Summary of results and discussion 122 7.2 Contributions of the dissertation 123 7.3 Future challenges 123 Bibliography 125 A Measurement systems 137 A.1 Measurement of vehicle velocity 137 A.2 Measurement of wheel speed 138 A.3 Measurement or estimation of wheel vertical load 139 A.4 Measurement of draft force 140 A.5 Further possible measurement systems 141 B Basic probability theoretical notions 142 B.1 Brief description of the theory of stochastic processes 142 B.2 Properties of stochastic signals 144 B.3 Bayesian filtering 145 C Modelling stochastic draft force and field microprofile 147 D Approximation of kappa-curves 152 E Simulation parameters 156

No description available.

Industrial automation systems normally consist of four different hierarchy levels: planning, scheduling, real-time optimization, and control. At the planning level, the goal is to compute an optimal production plan that minimizes the production cost while meeting process constraints. The planning model is typically formulated as a mixed integer nonlinear programming (MINLP), which is hard to solve to global optimality due to nonconvexity and large dimensionality attributes. Uncertainty in component qualities in gasoline blending due to measurement errors and variation in upstream processes may lead to off-specification products which require re-blending. Uncertainty in product demands may lead to a suboptimal solution and fail in capturing some potential profit due to shortage in products supply. While incorporating process uncertainties is essential to reducing the production cost and increasing profitability, it comes with the disadvantage of increasing the complexity of the MINLP planning model. The key contribution in the planning level is to employ the inventory pinch decomposition method to consider uncertainty in components qualities and products demands to reduce the production cost and increase profitability of the gasoline blend application. At the control level, the goal is to ensure desired operation conditions by meeting process setpoints, ensure process safety, and avoid process failures. Model predictive control (MPC) is an advanced control strategy that utilizes a dynamic model of the process to predict future process dynamic behavior over a time horizon. The effectiveness of the MPC relies heavily on the availability of a reasonably accurate process model. The key contributions in the control level are: (1) investigate the use of different system identification methods for the purpose of developing a dynamic model for high-purity distillation column, which is a highly nonlinear process. (2) Develop a novel hybrid based MPC to improve the control of the column and achieve flooding-free control. / Dissertation / Doctor of Philosophy (PhD) / The operation of a chemical process involves many decisions which are normally distributed into levels referred to as process automation hierarchy. The process automation hierarchy levels are planning, scheduling, real-time optimization, and control. This thesis addresses two of the levels in the process automation hierarchy, which are planning and control. At the planning level, the objective is to ensure optimal utilization of raw materials and equipment to reduce production cost. At the control level, the objective is to meet and follow process setpoints determined by the real-time optimization level. The main goals of the thesis are: (1) develop an efficient algorithm to solve a large-scale planning problem that incorporates uncertainties in components qualities and products demands to reduce the production cost and maximize profit for gasoline blending application. (2) Develop a novel hybrid-based model predictive control to improve the control strategy of an industrial distillation column that faces flooding issues.

A novel approach to the gain estimation problem,using a multi-armed bandit formulation, is studied. The gain estimation problem deals with the problem of estimating the largest L2-gain that signal of bounded norm experiences when passing through a linear and time-invariant system. Under certain conditions, this new approach is guaranteed to surpass traditional System Identification methods in terms of accuracy.The bandit algorithms Upper Confidence Bound, Thompson Sampling and Weighted Thompson Sampling are implemented with the aim of designing the optimal input for maximizing the gain of an unknown system. The regret performance of each algorithm is studied using simulations on a test system. Upper Confidence Bound, with exploration parameter set to zero, performed the best among all tested values for this parameter. Weighted Thompson Sampling performed better than Thompson Sampling.

In this work, an improved Learning Model Predictive Control (LMPC)architecture for autonomous racing is presented. The controller is referencefree and is able to improve lap time by learning from history data of previouslaps. A terminal cost and a sampled safe set are learned from history data toguarantee recursive feasibility and non-decreasing performance at each lap.Improvements have been proposed to implement LMPC on autonomousracing in a more efficient and reliable way. Improvements have been doneon three aspects. Firstly, system identification has been improved to be runin a more efficient way by collecting feature data in subspace, so that thesize of feature data set is reduced and time needed to run sorting algorithmcan be reduced. Secondly, different strategies have been proposed toimprove model accuracy, such as least mean square with/without lifting andGaussian process regression. Thirdly, for reducing algorithm complexity,methods combining different model construction strategies were proposed.Also, running controller in a multi-rate way has also been proposed toreduced algorithm complexity when increment of controller frequency isnecessary. Besides, the performance of different system identificationstrategies have been compared, which include strategy from newton’s law,strategy from classical system identification and strategy from machinelearning. Factors that can possibly influence converged result of LMPCwere also investigated, such as prediction horizon, controller frequency.Experiment results on a 1:10 scaled RC car illustrates the effectiveness ofproposed improvements and the difference of different system identificationstrategies. / I detta arbete, presenteras en förbättrad inlärning baserad modell prediktivkontroll (LMPC) för autonom racing, styralgoritm är referens fritt och har visatsig att kunna förbättra varvtid genom att lära sig ifrån historiska data från tidigarevarv. En terminal kostnad och en samplad säker mängd är lärde ifrån historiskdata för att garantera rekursiv genomförbarhet och icke-avtagande prestanda vidvarje varv.förbättringar har presenterats för implementering av LMPC på autonom racingpå ett mer effektivt och pålitligt sätt. Förbättringar har gjorts på tre aspekter.Först, för system identifiering, föreslår vi att samlar feature data i delrummet,så att storlek på samlade datamängd reduceras och tiden som krävs för attköra sorteringsalgoritm minskas. För det andra, föreslår vi olika strategierför förbättrade modellnoggrannheten, såsom LMS med/utan lyft och Gaussianprocess regression. För det tredje, För att reducerar komplexitet för algoritm,metoder som kombinerar olika modellbygg strategier föreslogs. Att körastyrenhet på ett multi-rate sätt har också föreslagits till för att reduceraalgoritmkomplexitet då inkrementet av styrfrekvensen är nödvändigt.Prestanda av olika systemidentifiering har jämförts, bland annat, Newtonslag, klassisk systemidentifierings metoder och strategier från maskininlärning.Faktorer som eventuellt kan påverka konvergens av LMPC resultat har ocksåundersökts. Såsom, prediktions horisont, styrfrekvensen.Experimentresultat på en 1:10 skalad RC-bilen visar effektiviteten hos föreslagnaförbättringarna och skillnaderna i olika systemidentifierings strategier.

Inverse analysis of nonlinear dynamic systems is an important area of research in the eld of structural health monitoring for civil engineering structures. Structural damage usually involves localized nonlinear behaviors of dynamic systems that evolve into different classes of nonlinearity as well as change system parameter values. Numerous parametric modal analysis techniques (e.g., eigensystem realization algorithm and subspace identification method) have been developed for system identification of multi-degree-of-freedom dynamic systems. However, those methods are usually limited to linear systems and known for poor sensitivity to localized damage. On the other hand, non-parametric identification methods (e.g., artificial neural networks) are advantageous to identify time-varying nonlinear systems due to unpredictable damage. However, physical interpretation of non-parametric identification results is not as straightforward as those of the parametric methods. In this study, the Multidegree-ofFreedom Restoring Force Method (MRFM) is employed as a semi-parametric nonlinear identi- fication method to take the advantages of both the parametric and non-parametric identification methods. The MRFM is validated using two realistic experimental nonlinear dynamic tests: (i) largescale shake table tests using building models with different foundation types, and (ii) impact test using wind blades. The large-scale shake table test was conducted at Tongji University using 1:10 scale 12-story reinforced concrete building models tested on three different foundations, including pile, box and fixed foundation. The nonlinear dynamic signatures of the building models collected from the shake table tests were processed using MRFM (i) to investigate the effects of foundation types on nonlinear behavior of the superstructure and (ii) to detect localized damage during the shake table tests. Secondly, the MRFM was applied to investigate the applicability of this method to wind turbine blades. Results are promising, showing a high level of nonlinearity of the system and how the MRFM can be applied to wind-turbine blades. Fuiii ture studies were planned for the comparison of physical characteristic of this blade with blades created made of other material.

In this paper, a roll stability controller (RSC) is presented based on an eight degree of freedom dynamic vehicle model. The controller is designed for and tested on a scaled vehicle performing obstacle avoidance maneuvers on a populated test track. A rapidly-exploring random tree (RRT) algorithm is used for the vehicle to execute a trajectory around an obstacle, and examines the geographic, non-homonymic, and dynamic constraints to maneuver around the obstacle. A model predictive controller (MPC) uses information about the vehicle state and, based on a weighted performance measure, generates an optimal trajectory around the obstacle. The RSC uses the standard vehicle state sensors: four wheel mounted encoders, a steering angle sensor, and a six degree of freedom inertial measurement unit (IMU). An emphasis is placed on the mitigation of rollover and spin-out, however if a safe maneuver is not found and a collision is inevitable, the program will run a brake command to reduce the vehicle speed before impact. The trajectory is updated at a rate of 20 Hz, providing improved stability and maneuverability for speeds up to 10 ft/s and turn angles of up to 20°.

On-board gimbal systems for camera stabilization in helicopters are typically based on linear models. Such models, however, are inaccurate due to system nonlinearities and complexities. As an alternative approach, artificial neural networks can provide a more accurate model of the gimbal system based on their non-linear mapping and generalization capabilities. This thesis investigates the applications of artificial neural networks to model the inertial characteristics (on the azimuth axis) of the inner gimbal in a gyro-stabilized multi-gimbal system. The neural network is trained with time-domain data obtained from gyro rate sensors of an actual camera system. The network performance is evaluated and compared with measured data and a traditional linear model. Computer simulation results show the neural network model fits well with the measured data and significantly outperforms a traditional model.

In the field of structural engineering it is of vital importance that the condition of an ageing structure is monitored to detect damages that could possibly lead to failure of the structure. Over the past few years various methods for monitoring the condition of structures have been proposed. With respect to civil and mechanical structures several methods make use of modal parameters such as, natural frequency, damping ratio and mode shapes. In the present work four methods for modal parameter estimation and two methods for have been evaluated for their application to multi degree of freedom structures. The methods evaluated for modal parameter estimation are: Wavelet transform, Hilbert-Huang transform, parametric system identification and peak picking. Through various numerical simulations the effectiveness of these methods is studied. It is found that the simple peak-picking method performs the best and is able to identify modal parameters most accurately in all the simulation cases that were considered in this study. The identified modal parameters are then used for locating the damage. Herein the flexibility and the rotational flexibility approaches are evaluated for damage detection. The approach based on the rotational flexibility is found to be more effective. / Master of Science