Heating, ventilation, and air conditioning (HVAC) is an indoor environmental technology that is extensively instrumented for large-scale buildings. Among all subsystems of buildings, the HVAC system dominates the energy consumption and accounts for 57% of the energy used in U.S. commercial and residential buildings. Unfortunately, the HVAC system may fail to meet the performance expectations due to various faults, including not only complete hardware failures, but also non-optimal operations. These faults waste more than 20% of the energy HVAC consumes. Therefore, it is of great potential to develop automatic, quick-responding, intelligent, and reliable monitoring and diagnosis tools to ensure the normal operations of HVAC and increase the energy efficiency of buildings.

To achieve these goals, increasing attentions have been attracted to two research areas, i.e., models that monitor the indoor thermal environment, and fault detection and diagnosis (FDD) tools that capture abnormal HVAC performance. Despite contributions of the existing works, there are still many challenges in these two areas. For the thermal models, the major concerns lie in 1) most of the models are determined empirically, 2) optimal structures and orders of the models are often determined through simulations, 3) the predictions of the models degrade quickly over longer time intervals, and 4) a lack of studies to incorporate architectural parameters and control variables into the models. For the FDD, we face the challenges of 1) the inherent complexity, coupled hardware and software, and increasing scale of HVAC significantly complicate the nature of faults, 2) faults occur at different levels with various degrees of impacts on upper-level HVAC units, 3) practical FDD tools at the system-level are scarce, and 4) the computational efficiency and calibration onerousness of the simulation-based FDD is a concern.

In this thesis, we address these challenges by innovating a system-level monitoring and diagnosis tool for HVAC. For the monitoring, we study and establish a parametric modeling approach to present indoor air temperature and thermal comfort. The resulting models take advantages of both analytical and numerical modeling techniques. These models have a two-stage regression structure, and explicitly include both architectural parameters and control variables as its predictors. As a result, they allow parametric studies of influence of the building envelope on indoor thermal behavior, serve as an efficient foundation for intelligent HVAC control design, and help optimize the design of and the material selection for office buildings. For the diagnosis, we innovate and develop a system-level FDD architecture for detecting faults across different levels of the HVAC system. Specifically, this architecture monitors and detects faulty HVAC units in a top-down manner. By monitoring HVAC units at higher level, instead of lower level components, the proposed FDD strategy reduces the computational effort in real-time monitoring of the HVAC system, obtains a system-level view of the HVAC operation, and provides a way to integrate the existing methods for component fault detection when needed. Based on extensive data collected from an office building on the campus of the University of California at Merced, numerical validations of the models, and examples of detected faults demonstrate the effectiveness of the proposed monitoring and diagnosis tool.

Heating, ventilation and air conditioning (HVAC) is a mechanical system that provides thermal comfort and accepted indoor air quality often instrumented for large-scale buildings. The HVAC system takes a dominant portion of overall building energy consumption and accounts for 50% of the energy used in the U.S. commercial and residential buildings in 2012. The performance and energy saving of building HVAC systems can be significantly improved by the implementation of better and smarter control strategies. Therefore, it is of great benefits to develop automatic, intelligent, optimal and consistent model and control tools to ensure the normal operations of HVAC systems and increase the building energy efficiency.

Motivated by these goals, this thesis presents a parametric modeling approach and a system-level control design for HVAC systems. For the modeling, we establish dynamical models for return air for the air-handiling-unit (AHU) of HVAC systems. The models include temperature and air flow rate models. These models follow the structure of the finite impulse response (FIR) model, and explicitly include the control variables from the AHU, the dynamical states, and the disturbances. Therefore, it is easy to apply this model to control design. Also, the model is flexible with prediction horizon and control horizon with a stability of the accuracy. In data processing, a convolution based low-pass digital filter, Savitzky-Golay filter is used for data smoothing. As a result, the return air flow rate model becomes more feasible with the smoothed data.

Secondly, this thesis study develops a model preidctive control (MPC) algorithm with the application of the dynamical models for AHU optimization problems. This control strategy optimizes the energy consumption of the AHU, and tracks the set points the room temperatures, supply air flow and return air flow rate of the building. The strategy provides physical-based inherent connection between components in AHU by applying damper positions, supply air temperature and outside air flow rate as manipulated variables. The control inputs are explicitly implemented into both the models and objective functions and the optimization structure is computationally efficient. The optimal results show an energy saving average percentage over and track the supply air flow rate and set point of room temperatures in the building effectively. The thermal load, supply air flow rate set points are calculated from thirty-two VAVs, that ensures the internal cooling demand, the static pressure and the ventilation level of the building.

In this thesis, all the data processing and modeling, model validation and implementation of the control algorithm are based on extensive data measurements collected from an office building on the campus of the University of California, at Merced. The control strategy is implemented into the online building automation system (BAS) of the building and can be easily incorporated with other BAS as well because of the explicit formulation.

Massive dissipated kinetic energy of roadway traffic is a sustainable energy source that can be harvested via piezoelectric effect. This approach provides a robust and reliable solution of roadway power generation. However, to achieve high energy density remains challenging. This thesis focuses on the development of a high density piezoelectric energy harvesting system from roadway traffic, which reaches an energy density as high as 15.37 J/(m.pass.lane) in drive tests. The high energy density is achieved by incorporating a compression-to-compression force amplification mechanism which amplifies the harvested energy by a factor of 100, as well as an optimized system configuration to take full advantage of dynamic load from vehicles. To the best of our knowledge, the scalable prototype system has achieved the highest energy density, which holds the great potential to enable self- powered smart road on a large scale. Based on the drive test results, implementation of this technology in one lane of 1609.34 m can generate at least electric energy of 7.22 × 10^4 kWh in one year.In this dissertation, the details of proposed piezoelectric energy harvesting technology are presented, including mathematical modeling and analysis, optimization of design parameters, mechanical design at all levels, lab test method and results, road implementation and drive test results. The chapters are organized as follows. Chapter 1 reviews recent developments of piezoelectric energy harvesting, including materials, configurations, and performances, followed by the discussion on the challenges and trends. Chapter 2 introduces the mathematical modeling of proposed design idea, derives the equations of motion for theoretical exploration. Chapter 3 demonstrates the procedure of dynamic analysis for this modeling, and presents simulation results of a selected design. Chapter 4 explains how we con- ducted parameters optimization for high energy density based on the mathematical modeling, afterwards, illustrates the mechanical design for the realization of an aggressive selected design and its laboratory validation. Chapter 5 introduces the vertical packaging of PEHUs, piezoelectric energy harvesting tower, including mechanical design, quasi-static compression test results, battery charging test results. Chapter 6 shows the system design for the implementation on road from both the electrical and mechanical perspectives. The drive test results based on a prototype system on the campus of UC Merced are presented. Chapter 7 discusses the optimal designs for the traffic of heavy truck for future exploration. Chapter 8 summarizes the thesis and discusses future directions.

Multi-Objective Optimization Problems (MOPs) deal with optimizing several objectives simultaneously and have diverse applications in engineering, economics, logistics, etc. The methods for solving MOPs can generally be classified into stochastic and deterministic approaches. Deterministic approaches are capable of finding the global solution even though they are computationally burdensome. Stochastic methods, on the other hand, can save on computations significantly, although they do not guarantee to find the global solution.

In engineering applications, MOPs can become nonlinear, multi-modal, high dimensional, and have complex structured solutions that makes them more challenging.

This theses follows two major goals. Firstly, it presents new methods and algorithms for solving engineering MOPs by hybridizing the existing methods and comparing their effectiveness by using benchmark problems. The hybrid method combines an evolutionary algorithm with a cell mapping method in order to reduce the computational time while maintaining the quality of the solution. Implementation details on parallel CPU/GPU programming of such methods are discussed as well. The second goal of this thesis is to introduce new applications of MOPs in different areas of engineering such as control design, path planning, fractional systems and airfoil design.

This dissertation presents a series of studies on the analysis and control of time-delayed dynamical systems. We first review the discrete methods of semidiscretization and digital control approach. Then we introduce a new method of continuous time approximation (CTA) for the response of linear and nonlinear dynamical systems with time delay. We investigate the accuracy of temporal responses as well as the spectral properties of the CTA method, and present several variants of the CTA method with improved accuracy. Finally, we apply the CTA method to control problems of time-delayed systems including optimal control, supervisory control and sliding mode control of uncertain systems. The CTA method creates a set of state equations for the delayed portion of the response leading to a high and finite dimensional state space formulation of the time-delayed system. The resulting state equations are in the standard state space form, making all the existing solution methods and control design tools for linear and nonlinear dynamical systems amenable to time-delayed systems. The CTA method can handle multiple independent time delays in a natural way. The work in this dissertation indicates that the CTA method is a promising method for analysis and control design of complex nonlinear time-delayed dynamical systems.

Underactuated mechanical system (UMS) is a special class of mechanical systems that play an important role in a wide variety of engineering applications. UMSs typically show a great difficulty in analysis and control design because of complex nonlinearity and loss of capability to configure arbitrary motions in some directions. Flatness-based control and active disturbance rejection control (ADRC) approach has been an active research topic about dealing with control problems of UMSs. Flatness, as a useful property of a class of dynamical systems, called flat systems, guarantees that all the states and inputs of them can be parameterized by a set of differentially independent coordinates, called flat output. The trajectory planning and tracking control of flat systems can be greatly simplified by prescribing the desired references in `flat space' without any other constraints except for necessary initial and final conditions. Despite its merits in control design, flatness is not a universal property of nonlinear dynamical systems and normally is not effortless to be characterized. Additionally, most UMSs are not inherently flat, i.e., not feedback linearizable, either statically or dynamically. The emerging framework that combines flatness of tangent linearization and ADRC has been experimentally proved feasible and robust for a large class of UMSs in recent years. While this approach continues to extend to more engineering applications, data-driven control design based on the flatness and ADRC has drawn attention of us, as fully detailed model information for flat output characterization is expensive and normally unavailable in control design. With few numbers of input-output measurements and little knowledge of the underlying UMSs, a data-driven approach for this framework enables an efficient, without modeling process, and systematic feedback control design for a class of UMSs with similar dynamical structures. Specifically, we first focus on the identification of flat output of nonlinear UMSs' tangent linearization using only input-output data collected. The identified flat outputs are naturally applied to the flatness-based control and ADRC framework, where several issues, such as trajectory planning and identification of state-flat output relations, are discussed and solved when extensive model information is no longer available. Frequency domain algorithms (FOID, MFOID) and time domain algorithm (FOID-Net) are proposed to solve flat output identification (FOID) problem of nonlinear UMSs. FOID and MFOID leverages the estimated transfer function of various flat output candidate in chosen frequency band to identify their relative degrees which is proved closely related to flatness of linearized UMSs. No detailed mathematical model of the system is needed. Flat output candidate is written as a linear combination of measured outputs. An optimal linear combination is identified so that the candidates achieve the highest relative degree. When the relative degree is equal to the order of the system, the output is flat. We have also developed data handling strategies to obtain the best estimate of the relative degree in the presence of measurement noise, high-frequency dynamics and Nyquist digitization effect. MFOID is an extension of FOID to deal with a special case of flat output of MIMO UMSs. The proposed algorithms are validated by numerical examples and an experimental study of a rotary crane system. The FOID-Net is a neural work framework designed to use time series data to estimate flat output out of linearized systems. We introduce the tracking differentiator into training of neural network which generates the time derivatives of flat output candidates and filter the noisy states. The idea of FOID-Net is based on the essential feature of flat systems that all states and inputs can be expressed by linear or nonlinear functions of flat coordinates. The training method, simulations of Furuta pendulum and a fourth-order nonlinear UMSs are discussed. A framework of designing data-driven robust tracking control based on identified flat output and sparse identification is proposed. Reduced linearized model are proposed to show that the number of components of linear flat outputs can be further reduced. Flat outputs can be identified by FOID algorithm or FOID-Net. Technique of sparse regression is applied to identify the relationships between flat output and system states, which reduces the order of the well-known extended state observer (ESO) and thereby make the ESO more effective for both trajectory planning and tracking in terms of the flat output. When FOID-Net is applied, the flat output-state relations can be directly found by weight matrix of well-trained layer. The proposed control scheme is validated by experimental studies of a rotary crane system in which a rest-to-rest tracking control is implemented.

Different challenges exist in the area of control, including but not limited to the high computational cost, model uncertainty, disturbances, nonlinearity, multi-objective optimization, constraints handling of partial differential equations, implementation, and real-time computation. In this dissertation, we focused on solving the issues of high computational cost, model uncertainty, nonlinearity, and disturbances in control design. The curse of dimensionality is a common issue for control design. Many mechanical systems with multi-degrees of freedom are underactuated by design. The dynamics of these systems live in a relatively high dimensional state space. However, many states are less controllable or observable for the high dimensional system.To keep the most controllable and observable states in the system, it is highly advantageous to develop a reduced order model. It turns out to be highly advantageous to develop a reduced order model to keep the most controllable and observable states in the system. Moreover, the curse of dimensionality also exists in the computation of neural networks solutions. The high computational cost is a severe issue in solving the neural network solutions in optimal control. In this dissertation, we adopted the balanced truncation and empirical balanced truncation to reduce the order of the dynamics systems. Moreover, the radial basis function neural networks (RBFNN) are adopted to solve the Hamilton-Jacobi-Bellman (HJB) equation in optimal control. The high-speed train bogie and the Quanser Qube2 system are adopted as practical applications to test the performance of the reduced order model-based control design. Especially for the Qube2 system, extensive numerical simulations and experimental validations show that the RBFNN with a reduced order model save the computational power and improve the optimal controller's robustness. Furthermore, transfer learning is adopted to improve the performance of the radial basis function based HJB solutions with limited experimental data. Except for reducing the dimension of the system, a new neural networks structure - separable Gaussian neural networks is proposed to reduce the computation cost by making use of the factorizable property of the Gaussian activation functions in RBFNN. Numerical simulation results show that the separable gaussian functions reduce the computation cost and the required parameters in neural networks to solve the partial differential equations.

Model uncertainty and disturbances also play an important role in control design. The Luenberger, extended state observers and neural networks are adopted to estimate the unmeasurable system states and approximate unmodelled system dynamics. The minimum number of sensors to estimate the unknown dynamics are investigated with the help of the special structure of the extended state observer. With the unknown dynamics estimated, the recursive least squares algorithm is adopted to identify the parameters in the unknown dynamics if the persistent excitation condition is satisfied. Numerical examples show that the extended state observer performs well in estimating the unknown dynamics with a limited number of sensors. Except for using the observers, neural networks are also adopted to approximate the unknown dynamics of the system. Here, we adopted the Delta robot as an engineering application to illustrate the performance of a neural networks model with sliding mode control. By leveraging the machine learning techniques, the model uncertainty issue can be well-solved during the control design.