Search

Finding alternative power sources has been an important topic of study worldwide. It is vital to find substitutes for finite fossil fuels. Such substitutes may be termed renewable energy sources and infinite supplies. Such limitless sources are derived from ambient energy like wind energy, solar energy, sea waves energy; on the other hand, smart cities megaprojects have been receiving enormous amounts of funding to transition our lives into smart lives. Smart cities heavily rely on smart devices and electronics, which utilize small amounts of energy to run. Using batteries as the power source for such smart devices imposes environmental and labor cost issues. Moreover, in many cases, smart devices are in hard-to-access places, making accessibility for disposal and replacement difficult. Finally, battery waste harms the environment. To overcome these issues, vibration-based energy harvesters have been proposed and implemented. Vibration-based energy harvesters convert the dynamic or kinetic energy which is generated due to the motion of an object into electric energy. Energy transduction mechanisms can be delivered based on piezoelectric, electromagnetic, or electrostatic methods; the piezoelectric method is generally preferred to the other methods, particularly if the frequency fluctuations are considerable. In response, piezoelectric vibration-based energy harvesters (PVEHs), have been modeled and analyzed widely. However, there are two challenges with PVEH: the maximum amount of extractable voltage and the effective (operational) frequency bandwidth are often insufficient. In this dissertation, a new type of integrated multiple system comprised of a cantilever and spring-oscillator is proposed to improve and develop the performance of the energy harvester in terms of extractable voltage and effective frequency bandwidth. The new energy harvester model is proposed to supply sufficient energy to power low-power electronic devices like RFID components. Due to the temperature fluctuations, the thermal effect over the performance of the harvester is initially studied. To alter the resonance frequency of the harvester structure, a rotating element system is considered and analyzed. In the analytical-numerical analysis, Hamilton’s principle along with Galerkin’s decomposition approach are adopted to derive the governing equations of the harvester motion and corresponding electric circuit. It is observed that integration of the spring-oscillator subsystem alters the boundary condition of the cantilever and subsequently reforms the resulting characteristic equation into a more complicated nonlinear transcendental equation. To find the resonance frequencies, this equation is solved numerically in MATLAB. It is observed that the inertial effects of the oscillator rendered to the cantilever via the restoring force effects of the spring significantly alter vibrational features of the harvester. Finally, the voltage frequency response function is analytically and numerically derived in a closed-from expression. Variations in parameter values enable the designer to mutate resonance frequencies and mode shape functions as desired. This is particularly important, since the generated energy from a PVEH is significant only if the excitation frequency coming from an external source matches the resonance (natural) frequency of the harvester structure. In subsequent sections of this work, the oscillator mass and spring stiffness are considered as the design parameters to maximize the harvestable voltage and effective frequency bandwidth, respectively. For the optimization, a genetic algorithm is adopted to find the optimal values. Since the voltage frequency response function cannot be implemented in a computer algorithm script, a suitable function approximator (regressor) is designed using fuzzy logic and neural networks. The voltage function requires manual assistance to find the resonance frequency and cannot be done automatically using computer algorithms. Specifically, to apply the numerical root-solver, one needs to manually provide the solver with an initial guess. Such an estimation is accomplished using a plot of the characteristic equation along with human visual inference. Thus, the entire process cannot be automated. Moreover, the voltage function encompasses several coefficients making the process computationally expensive. Thus, training a supervised machine learning regressor is essential. The trained regressor using adaptive-neuro-fuzzy-inference-system (ANFIS) is utilized in the genetic optimization procedure. The optimization problem is implemented, first to find the maximum voltage and second to find the maximum widened effective frequency bandwidth, which yields the optimal oscillator mass value along with the optimal spring stiffness value. As there is often no control over the external excitation frequency, it is helpful to design an adaptive energy harvester. This means that, considering a specific given value of the excitation frequency, energy harvester system parameters (oscillator mass and spring stiffness) need to be adjusted so that the resulting natural (resonance) frequency of the system aligns with the given excitation frequency. To do so, the given excitation frequency value is considered as the input and the system parameters are assumed as outputs which are estimated via the neural network fuzzy logic regressor. Finally, an experimental setup is implemented for a simple pure cantilever energy harvester triggered by impact excitations. Unlike the theoretical section, the experimental excitation is considered to be an impact excitation, which is a random process. The rationale for this is that, in the real world, the external source is a random trigger. Harmonic base excitations used in the theoretical chapters are to assess the performance of the energy harvester per standard criteria. To evaluate the performance of a proposed energy harvester model, the input excitation type consists of harmonic base triggers. In summary, this dissertation discusses several case studies and addresses key issues in the design of optimized piezoelectric vibration-based energy harvesters (PVEHs). First, an advanced model of the integrated systems is presented with equation derivations. Second, the proposed model is decomposed and analyzed in terms of mechanical and electrical frequency response functions. To do so, analytic-numeric methods are adopted. Later, influential parameters of the integrated system are detected. Then the proposed model is optimized with respect to the two vital criteria of maximum amount of extractable voltage and widened effective (operational) frequency bandwidth. Corresponding design (influential) parameters are found using neural network fuzzy logic along with genetic optimization algorithms, i.e., a soft computing method. The accuracy of the trained integrated algorithms is verified using the analytical-numerical closed-form expression of the voltage function. Then, an adaptive piezoelectric vibration-based energy harvester (PVEH) is designed. This final design pertains to the cases where the excitation (driving) frequency is given and constant, so the desired goal is to match the natural frequency of the system with the given driving frequency. In this response, a regressor using neural network fuzzy logic is designed to find the proper design parameters. Finally, the experimental setup is implemented and tested to report the maximum voltage harvested in each test execution., © 2023 Alireza Babaei