Heart signals allow for a comprehensive analysis of the heart. Electrocardiography (ECG or EKG) uses electrodes to measure the electrical activity of the heart. Extracting ECG signals is a non-invasive process that opens the door to new possibilities for the application of advanced signal processing and data analysis techniques in the diagnosis of heart diseases. With the help of today’s large database of ECG signals, a computationally intelligent system can learn and take the place of a cardiologist. Detection of various abnormalities in the patient’s heart to identify various heart diseases can be made through an Adaptive Neuro-Fuzzy Inference System (ANFIS) preprocessed by subtractive clustering. Six types of heartbeats are classified: normal sinus rhythm, premature ventricular contraction (PVC), atrial premature contraction (APC), left bundle branch block (LBBB), right bundle branch block (RBBB), and paced beats. The goal is to detect important characteristics of an ECG signal to determine if the patient’s heartbeat is normal or irregular. The results from three trials indicate an average accuracy of 98.10%, average sensitivity of 94.99%, and average specificity of 98.87%. These results are comparable to two artificial neural network (ANN) algorithms: gradient descent and Levenberg Marquardt, as well as the ANFIS preprocessed by grid partitioning.

Stress and overload are strongly associated with unsafe behaviour, which motivated various studies to detect them automatically in workplaces. This study aims to advance safety research by developing a data-driven stress and overload detection method. An unsupervised deep learning-based anomaly detection method is developed to detect stress. The proposed method performs with convolutional neural network encoder-decoder and long short-term memory equipped with an attention layer. Data from a field experiment with 18 participants was used to train and test the developed method. The field experiment was designed to include a pre-defined sequence of activities triggering mental and physical stress, while a wristband biosensor was used to collect physiological signals. The collected contextual and physiological data were pre-processed and then resampled into correlation matrices of 14 features. Correlation matrices are used as an input to the unsupervised Deep Learning (DL) based anomaly detection method. The developed method is validated, offering accuracy and F-measures close to 0.98. The technique employed captures the input data attributes correlation, promoting higher interpretability of the DL method for easier comprehension. Over-reliance on uncertain absolute truth, the need for a high number of training samples, and the requirement of a threshold for detecting anomalies are identified as shortcomings of the proposed method. To overcome these shortcomings, an Adaptive Neuro-Fuzzy Inference System (ANFIS) was designed and developed. While the ANFIS method did not improve the overall accuracy, it outperformed the DL-based method in detecting anomalies precisely. The overall performance of the ANFIS method is better than the DL-based method for the anomalous class, and the method results in lower false alarms. However, the DL-based method is suitable for circumstances where false alarms are tolerated. / Dissertation / Doctor of Philosophy (PhD)

Linear repetitive projects, which are resource-driven in nature, are characterized by a series of repetitive activities in which the resources share the same space either in sequential or parallel manner. The frequent movement of resources over limited shared space needs to be well-planned to avoid potential issues during the execution of linear projects. As such, schedules developed for these projects needs not only to take into account all the logical, project-dependent and precedence constraints of activities but also to incorporate the space and time constraints that co-exist for the movement of thei8r resources. Negligence in incorporating spatial and temporal constraints in developing and improving schedules of linear projects increases the risk of delays and workspace congestions that can substantially hinder the performance of the activity resources. The study presented here proposes and develops an uncertainty-aware scheduling and control framework for linear projects to address the needs mentioned above. For this purpose, first, a new type of float was introduced as the Space-Time Float. The Space-Time Float is an envelope for all possible movement patterns that a linear activity or its associated resources can take considering the time and space constraints of that activity. The next endeavor in the development of the uncertainty-aware linear scheduling and control framework was to augment the current linear scheduling methods by presenting an uncertainty-aware optimization method to optimize the duration of linear projects while minimizing their potential congestions. A constraint satisfaction approach was used for the two-tier optimization of duration and congestion, and a fuzzy inference system was incorporated to assess the inherent uncertainty in linear activities. A new type of buffer, Uncertainty-Aware Productivity Buffer is also introduced to account for the uncertainties inherent in project activities. Spatial progress of activities needs not only to be considered in the planning phase but also to be closely monitored during construction. The framework presented in this study also applies to the monitoring and control of linear projects. While most of the current methods still do not accommodate real-time bi-directional control of linear projects, this framework is based on the Cyber-Physical Systems (CPS) architecture and bi-directional communication of data. To this end, a CPS-based application for Earned Value (EV) monitoring and control of road and highway projects is presented. Different steps of the generated framework are validated through various literature and field-based case studies. The results demonstrate the effectiveness of the presented method in planning and control of unforeseen variations from the planned schedules of linear projects. As such, the present study contributes and adds to the current body of knowledge of linear projects by presenting an efficient scheduling and control framework that takes into account logical, spatio-temporal and project-based constraints of linear activities. / Thesis / Doctor of Philosophy (PhD)

Finite Element (FE) analysis is a powerful numerical technique widely utilized to simulate the real-world response of complex engineering structures. With the advancements in adaptive optimization frameworks, multi-fidelity (coupled shell-solid) FE models are increasingly sought during the early design stages where a large design space is being explored. This is because multi-fidelity models have the potential to provide accurate solutions at a much lower computational cost. However, the time and effort required to create accurate and optimal multi-fidelity models with acceptable meshes for highly complex structures is still significant and is a major bottleneck in the FE modeling process. Additionally, there is a significant level of subjectivity involved in the decision-making about the multi-fidelity element topology due to a high dependence on the analyst's experience and expertise, which often leads to disagreements between analysts regarding the optimal modeling approach and heavy losses due to schedule delays. Moreover, this analyst-to-analyst variability can also result in significantly different final engineering designs. Thus, there is a greater need to accelerate the FE modeling process by automating the development of robust and adaptable multi-fidelity models as well as eliminating the subjectivity and art involved in the development of multi-fidelity models. This dissertation presents techniques and frameworks for accelerating the finite element modeling process of multi-fidelity models. A framework for the automated development of multi-fidelity models with adaptable 2-D/3-D topology using the parameterized full-fidelity and structural fidelity models is presented. Additionally, issues related to the automated meshing of highly complex assemblies is discussed and a strategic volume decomposition technique blueprint is proposed for achieving robust hexahedral meshes in complicated assembly models. A comparison of the full-solid, full-shell, and different multi-fidelity models of a highly complex stiffened thin-walled pressure vessel under external and internal tank pressure is presented. Results reveal that automation of multi-fidelity model generation in an integrated fashion including the geometry creation, meshing and post-processing can result in considerable reduction in cost and efforts. Secondly, the issue of analyst-to-analyst variability is addressed using a Decision Tree (DT) based Fuzzy Inference System (FIS) for recommending optimal 2D-3D element topology for a multi-fidelity model. Specifically, the FIS takes the structural geometry and desired accuracy as inputs (for a range of load cases) and infers the optimal 2D-3D topology distribution. Once developed, the FIS can provide real-time optimal choices along with interpretability that provides confidence to the analyst regarding the modeling choices. The proposed techniques and frameworks can be generalized to more complex problems including non-linear finite element models and as well as adaptable mesh generation schemes. / Doctor of Philosophy / Structural analysis is the process of determining the response (mainly, deformation and stresses) of a structure under specified loads and external conditions. This is often performed using computational modeling of the structure to approximate its response in real-life conditions. The Finite Element Method (FEM) is a powerful and widely used numerical technique utilized in engineering applications to evaluate the physical performance of structures in several engineering disciplines, including aerospace and ocean engineering. As optimum designs are increasing sought in industries, the need to develop computationally efficient models becomes necessary to explore a large design space. As such, optimal multi-fidelity models are preferred that utilize higher fidelity computational domain in the critical areas and a lower fidelity domain in less critical areas to provide an optimal trade-off between accuracy and efficiency. However, the development of such optimal models involves a high level of expertise in making a-priori and a-posteriori optimal modeling decisions. Such experience based variability between analysts is often a major cause of schedule delays and considerable differences in final engineering designs. A combination of automated model development and optimization along with an expert system that relieves the analyst of the need for experience and expertise in making software and theoretical assumptions for the model can result in a powerful and cost-effective computational modeling process that accelerates technological advancements. This dissertation proposes techniques for automating robust development of complex multi-fidelity models. Along with these techniques, a data-driven expert system framework is proposed that makes optimal multi-fidelity modeling choices based on the structural configuration and desired accuracy level.

Este trabalho apresenta duas metodologias baseadas em lógica fuzzy para automatizar o diagnóstico de defeito em equipamentos mecânicos, além de fazer uma comparação de seu desempenho utilizando um caso experimental. As duas metodologias estudadas são: o sistema de inferência fuzzy e o algoritmo baseado em Fuzzy C-Means. O alarme estatístico é uma metodologia existente atualmente na indústria com este objetivo e que será utilizado neste trabalho para comparação de desempenho. Para realizar os testes, foram desenvolvidos programas que permitiram criar alarmes e sistemas fuzzy utilizando um banco de dados experimental. De modo diferente ao que são feitos normalmente, os sistemas fuzzy de diagnóstico testados neste trabalho foram construídos automaticamente utilizando informações do banco de dados experimentais composto por sinais de vibração, que representam a condição normal e diversos tipos de defeitos em mancais de rolamentos. Os parâmetros escalares característicos necessários para a entrada nos sistemas fuzzy foram obtidos através do processamento dos sinais de vibração de mancais de rolamentos. Nas análises realizadas neste trabalho, foi estudada a influência de diversos características de criação do sistema fuzzy. Como exemplo, pode-se citar como principal influência, a complexidade do banco de dados a ser analisado pelo sistema fuzzy. Por fim, além de apresentar uma comparação de performance entre as metodologias fuzzy apresentadas no trabalho, com o alarme estatístico, são discutidas as características de cada uma destas metodologias. Destacam-se como principais contribuições deste trabalho, a obtenção de uma metodologia utilizada para criar de maneira automática o sistema de inferência fuzzy e as modificações realizadas no algoritmo Fuzzy C-Means para aperfeiçoar o desempenho em classificação de defeitos. / This works describes two proposed methodologies for the automatic diagnoses in mechanical equipment: the fuzzy system inference and a Fuzzy C-Means based algorithm. Their performances are evaluated in an experimental case and, afterwards, also compared by the statistical alarm, a diagnostic methodology very used in industries at present. In order to do the tests, a developed computer algorithm allowed creating alarms and fuzzy systems by the use of an experimental database. These tested diagnostic systems were automatically built using information from the mentioned database that was composed by samples of vibration signals, representing several types of rolling bearing defects and the bearing normal condition. The fuzzy systems input scalar parameters were obtained by signal processing. The influence of some of the building fuzzy systems parameters in the system performance was also studied, which allow establishing, for example, that the database complexity is an important factor in the fuzzy system performance. Finally, this work discusses the main characteristics of each one of the described methodologies. The most important contribution of this work is the proposition of a methodology for creating fuzzy system automatically as well as the analysis of the fuzzy C-Means as a tool for system diagnoses.

Este trabalho aborda o problema de modelagem de séries de vazões afluentes aos aproveitamentos hidrelétricos. A previsão de vazão natural fluvial é realizada semanalmente para 158 usinas hidrelétricas do Sistema Interligado Nacional (SIN), pois trata-se de insumo fundamental para o planejamento e operação do sistema elétrico brasileiro. Diversos modelos são utilizados na determinação destas previsões, entre os quais podem ser citados os modelos físicos, os estatísticos e aqueles que aplicam sistemas inteligentes. Apesar de contínuos aprimoramentos terem sido incorporados ao processo de previsão de vazão, existem alguns aproveitamentos hidrelétricos para os quais os resultados de estimação têm apresentado grandes desvios. Neste contexto, com a motivação de se obter uma resposta acurada, investigam-se os sistemas fuzzy como modelos concorrentes aplicados à previsão de vazões semanais. O objetivo do trabalho é reduzir os erros de estimação para uma usina piloto, incorporando à previsão de vazão os dados de precipitação. Para a construção da série histórica de precipitação média da bacia hidrográfica, fez-se uma exaustiva pesquisa por estações pluviométricas, seguida por tratamento de dados de medição e método de interpolação. Ao final do trabalho, é apresentada uma análise comparativa entre os resultados obtidos com o Modelo Autorregressivo Periódico (PAR) e o sistema de inferência fuzzy. Com base no desempenho observado, superior ao modelo autorregressivo, comprova-se a adequação do modelo proposto para a modelagem do processo hidrológico. / This work addresses the modelling problem of hydropower plants reservoir streamflow series. The natural streamflow forecasting for 157 hydroelectric power plants of the National Interconnected System - NIS is updated on a weekly basis, which is an essential input for the planning and operation of the Brazilian Electric Power System. Several models are used to determine this prediction, such as physicals, statisticals and the ones that use intelligent systems. Despite the improvements to natural streamflow forecasting, substantial deviation has been found for the expected results of some hydropower plants. Highlighted the importance of this variable, fuzzy systems applied to weekly streamflows forecasts will be investigated as alternative models, in order to obtain better results. The purpose of this work is to reduce the estimation errors for a pilot hydropower plant, incorporating precipitation data into the forecast. Therefore, an exhaustive research to acquire data from hydrometeorological stations was conducted. After being treated, a variable selection method was applied to the data, defining the most relevant input variables for the prediction model. At the end, a comparative analysis shows that the fuzzy model presents a better performance than the periodic autoregressive model used by ONS to plan the operation of the electric power system.

In recent years, response analysis of complex structures under impact loads has attracted a great deal of attention. For example, a collision or an accident that produces impact loads that exceed the design load can cause severe damage on the structural components. Although the AASHTO specification is used for impact-resistant bridge design, it has many limitations. The AASHTO specification does not incorporate complex and uncertain factors. Thus, a well-designed structure that can survive a collision under specific conditions in one region may be severely damaged if it were impacted by a different vessel, or if it were located elsewhere with different in-situ conditions. With these limitations in mind, we propose different solutions that use smart control technology to mitigate impact hazard on structures. However, it is challenging to develop an accurate mathematical model of the integrated structure-smart control systems. The reason is due to the complicated nonlinear behavior of the integrated nonlinear systems and uncertainties of high impact forces. In this context, novel algorithms are developed for identification, control and monitoring of nonlinear responses of smart structures under high impact forces. To evaluate the proposed approaches, a smart aluminum and two smart reinforced concrete beam structures were designed, manufactured, and tested in the High Impact Engineering Laboratory of Civil and Environmental Engineering at WPI. High-speed impact force and structural responses such as strain, deflection and acceleration were measured in the experimental tests. It has been demonstrated from the analytical and experimental study that: 1) the proposed system identification model predicts nonlinear behavior of smart structures under a variety of high impact forces, 2) the developed structural health monitoring algorithm is effective in identifying damage in time-varying nonlinear dynamic systems under ambient excitations, and 3) the proposed controller is effective in mitigating high impact responses of the smart structures.

"Maintaining an efficient and reliable infrastructure requires continuous monitoring and control. In order to accomplish these tasks, algorithms are needed to process large sets of data and for modeling based on these processed data sets. For this reason, computationally efficient and accurate modeling algorithms along with data compression techniques and optimal yet practical control methods are in demand. These tools can help model structures and improve their performance. In this thesis, these two aspects are addressed separately. A principal component analysis based adaptive neuro-fuzzy inference system is proposed for fast and accurate modeling of time-dependent behavior of a structure integrated with a smart damper. Since a smart damper can only dissipate energy from structures, a challenge is to evaluate the dissipativity of optimal control methods for smart dampers to decide if the optimal controller can be realized using the smart damper. Therefore, a generalized deterministic definition for dissipativity is proposed and a commonly used controller, LQR is proved to be dissipative. Examples are provided to illustrate the effectiveness of the proposed modeling algorithm and evaluating the dissipativity of LQR control method. These examples illustrate the effectiveness of the proposed modeling algorithm and dissipativity of LQR controller."

Este trabalho aborda o problema de modelagem de séries de vazões afluentes aos aproveitamentos hidrelétricos. A previsão de vazão natural fluvial é realizada semanalmente para 158 usinas hidrelétricas do Sistema Interligado Nacional (SIN), pois trata-se de insumo fundamental para o planejamento e operação do sistema elétrico brasileiro. Diversos modelos são utilizados na determinação destas previsões, entre os quais podem ser citados os modelos físicos, os estatísticos e aqueles que aplicam sistemas inteligentes. Apesar de contínuos aprimoramentos terem sido incorporados ao processo de previsão de vazão, existem alguns aproveitamentos hidrelétricos para os quais os resultados de estimação têm apresentado grandes desvios. Neste contexto, com a motivação de se obter uma resposta acurada, investigam-se os sistemas fuzzy como modelos concorrentes aplicados à previsão de vazões semanais. O objetivo do trabalho é reduzir os erros de estimação para uma usina piloto, incorporando à previsão de vazão os dados de precipitação. Para a construção da série histórica de precipitação média da bacia hidrográfica, fez-se uma exaustiva pesquisa por estações pluviométricas, seguida por tratamento de dados de medição e método de interpolação. Ao final do trabalho, é apresentada uma análise comparativa entre os resultados obtidos com o Modelo Autorregressivo Periódico (PAR) e o sistema de inferência fuzzy. Com base no desempenho observado, superior ao modelo autorregressivo, comprova-se a adequação do modelo proposto para a modelagem do processo hidrológico. / This work addresses the modelling problem of hydropower plants reservoir streamflow series. The natural streamflow forecasting for 157 hydroelectric power plants of the National Interconnected System - NIS is updated on a weekly basis, which is an essential input for the planning and operation of the Brazilian Electric Power System. Several models are used to determine this prediction, such as physicals, statisticals and the ones that use intelligent systems. Despite the improvements to natural streamflow forecasting, substantial deviation has been found for the expected results of some hydropower plants. Highlighted the importance of this variable, fuzzy systems applied to weekly streamflows forecasts will be investigated as alternative models, in order to obtain better results. The purpose of this work is to reduce the estimation errors for a pilot hydropower plant, incorporating precipitation data into the forecast. Therefore, an exhaustive research to acquire data from hydrometeorological stations was conducted. After being treated, a variable selection method was applied to the data, defining the most relevant input variables for the prediction model. At the end, a comparative analysis shows that the fuzzy model presents a better performance than the periodic autoregressive model used by ONS to plan the operation of the electric power system.

Esta dissertação testa e compara dois tipos de modelagem para previsão de uma mesma série temporal. Foi observada uma série temporal de distribuição de energia elétrica e, como estudo de caso, optou-se pela região metropolitana do Estado da Bahia. Foram testadas as combinações de três variáveis exógenas em cada modelo: a quantidade de clientes ligados na rede de distribuição de energia elétrica, a temperatura ambiente e a precipitação de chuvas. O modelo linear de previsão de séries temporais utilizado foi um SARIMAX. A modelagem de inteligência computacional utilizada para a previsão da série temporal foi um sistema de Inferência Fuzzy. Na busca de um melhor desempenho, foram feitos testes de quais variáveis exógenas melhor influenciam no comportamento da energia distribuída em cada modelo. Segundo a avaliação dos testes, o sistema Fuzzy de previsão foi o que obteve o menor erro. Porém dentre os menores erros, os resultados dos testes também indicaram diferentes variáveis exógenas para cada modelo de previsão. / This dissertation tests and compares two types of predicting models to the same time series. A time series of electricity distribution was observed and, as a case study, were opted for the metropolitan region of Bahia State. Three exogenous variables were tested in each model: the number of customers connected to the electricity distribution network, the temperature and the precipitation of rain. The linear model time series forecasting used was a SARIMAX. The modelling of computational intelligence used to predict the time series was a Fuzzy Inference System. For better performance, in each model was tested all the exogenous variables to fit the influence in the energy distributed. According to the evaluation of the tests, the Fuzzy forecasting system presented the lowest error. But among the smallest errors, the results of the tests also indicated different exogenous variables for each forecast model.