This thesis reports the work of using simulated annealing to design more efficient fuzzy logic systems to model problems with associated uncertainties. Simulated annealing is used within this work as a method for learning the best configurations of type-1 and type-2 fuzzy logic systems to maximise their modelling ability. Therefore, it presents the combination of simulated annealing with three models, type-1 fuzzy logic systems, interval type-2 fuzzy logic systems and general type-2 fuzzy logic systems to model four bench-mark problems including real-world problems. These problems are: noise-free Mackey-Glass time series forecasting, noisy Mackey-Glass time series forecasting and two real world problems which are: the estimation of the low voltage electrical line length in rural towns and the estimation of the medium voltage electrical line maintenance cost. The type-1 and type-2 fuzzy logic systems models are compared in their abilities to model uncertainties associated with these problems. Also, issues related to this combination between simulated annealing and fuzzy logic systems including type-2 fuzzy logic systems are discussed. The thesis contributes to knowledge by presenting novel contributions. The first is a novel approach to design interval type-2 fuzzy logic systems using the simulated annealing algorithm. Another novelty is related to the first automatic design of general type-2 fuzzy logic system using the vertical slice representation and a novel method to overcome some parametrisation difficulties when learning general type-2 fuzzy logic systems. The work shows that interval type-2 fuzzy logic systems added more abilities to modelling information and handling uncertainties than type-1 fuzzy logic systems but with a cost of more computations and time. For general type-2 fuzzy logic systems, the clear conclusion that learning the third dimension can add more abilities to modelling is an important advance in type-2 fuzzy logic systems research and should open the doors for more promising research and practical works on using general type-2 fuzzy logic systems to modelling applications despite the more computations associated with it.

There are two main defined classes of uncertainty namely: fuzziness and ambiguity, where ambiguity is ‘one-to-many’ relationship between syntax and semantic of a proposition. This definition seems that it ignores ‘many-to-many’ relationship ambiguity type of uncertainty. In this thesis, we shall use complex-uncertainty to term many-to-many relationship ambiguity type of uncertainty. This research proposes a new approach for handling the complex ambiguity type of uncertainty that may exist in data, for software engineering of predictive Data Mining (DM) classification models. The proposed approach is based on Relative-Fuzzy Logic (RFL), a novel type of fuzzy logic. RFL defines a new formulation of the problem of ambiguity type of uncertainty in terms of States Of Proposition (SOP). RFL describes its membership (semantic) value by using the new definition of Domain of Proposition (DOP), which is based on the relativity principle as defined by possible-worlds logic. To achieve the goal of proposing RFL, a question is needed to be answered, which is: how these two approaches; i.e. fuzzy logic and possible-world, can be mixed to produce a new membership value set (and later logic) that able to handle fuzziness and multiple viewpoints at the same time? Achieving such goal comes via providing possible world logic the ability to quantifying multiple viewpoints and also model fuzziness in each of these multiple viewpoints and expressing that in a new set of membership value. Furthermore, a new architecture of Hierarchical Neural Network (HNN) called ML/RFL-Based Net has been developed in this research, along with a new learning algorithm and new recalling algorithm. The architecture, learning algorithm and recalling algorithm of ML/RFL-Based Net follow the principles of RFL. This new type of HNN is considered to be a RFL computation machine. The ability of the Relative Fuzzy-based DM prediction model to tackle the problem of complex ambiguity type of uncertainty has been tested. Special-purpose Integrated Development Environment (IDE) software, which generates a DM prediction model for speech recognition, has been developed in this research too, which is called RFL4ASR. This special purpose IDE is an extension of the definition of the traditional IDE. Using multiple sets of TIMIT speech data, the prediction model of type ML/RFL-Based Net has classification accuracy of 69.2308%. This accuracy is higher than the best achievements of WEKA data mining machines given the same speech data.

The aim of this research project is to develop an innovative framework to implement lean principles in Product-Service System (PSS) with the capability of assessing the leanness level of the services offering process. The framework comprises three implementation phases namely: assessment of the current state, developing a future state, and stabilising the new way of operations. Additionally, the framework covers the enablers, factors, and appropriate lean tools required for the successful implementation of lean practices in Product-Service System (PSS), as well as, the challenges that may obstacle the implementation process. The proposed framework has integrated an assessment model that provides a quantifiable measure of the leanness level of Product-Service System (PSS). Five main enablers and thirty three factors emerging from these main enablers deemed to be critical for the successful implementation of lean practices in Product-Service System (PSS). Moreover, a series of eight inhibitors appeared to block the implementation process. The Product-Service System leanness assessment model was developed upon three main levels, namely: enablers, criteria, and attributes. The first level contains five enablers. These enablers are supplier relationship, management leanness, workforce leanness, process excellence, and customer relationship. In the second level there are twenty one criteria such as: supplier delivery, culture of management and process optimisation. Finally, the third level consists of seventy three attributes. By using multi-grade fuzzy approach the PSS leanness index was computed and areas for further improvement were identified. A combination of research methodology approaches has been employed in this research. Firstly, an extensive literature review related to lean and PSS was conducted. Secondly, the qualitative approach and the case study were selected as an appropriate methodology for this research, using semi-structured and structured interview techniques to gather the required data from experts who are involved in lean projects in their companies. Finally, validation of the results was carried out using real life industrial case studies and experts judgment. Case studies demonstrate that the framework provides guidelines for manufacturing companies that aim to implement lean principles in Product-Service System (PSS). The framework enables manufacturing companies to better satisfy their customers’ needs through responding quickly to their changing demands; to improve the service offering process through reducing the creation of wastes and non-value added activities; and to improve competitiveness through increasing customers’ value. Additionally, the PSS leanness index is useful for improving the service offering process. The index provides manufacturing companies with a real insight into the leanness level of their service offering, as well as, it provides managers with a quantifiable measure of how lean their PSS is. The index identifies the gap between the current state and the future state and this helps in determining areas for further improvement.

Due to obstruction in the national plan of urbanization in Tripoli (Libya) and population growth, serious problems have emerged in the form of random settlements, overcrowding and poor infrastructure. After more than two decades of inertia, the government has created a national plan in order to resolve the problems, hence it has enforced the demolition of some zones and modified other (irregularly built) ones, however the process is extremely costly. This research introduces a solution through cellular automata (CA) model to predict growth trends; size of residential, industrial and utilities areas; and to project future population. The model is implemented using digitized land use maps of Tripoli to indicate each areas as group of cells to predict their growth. The model incorporates two types of fuzzy rules bases, the first of which is based on the inputs population and area, and the second of which is based on the three inputs of population, area and density. The population prediction is performed using three scenarios, namely decreasing, fixed and increasing growth rates, such that all possibilities of growth are covered. In addition, the residential area prediction is performed based on two cases: normal density and low density. The former is introduced since new areas tend to have more open spaces and bigger houses. Furthermore, the model considers the growth of the industrial areas to be slower than that of residential areas. The model is developed and validated for the period of 1980 to 2010. The prediction is performed for thirty years from 2010 to 2040. In addition to the CA model, a regression model is developed and tested on the three growth scenarios for the same period (30 years). The prediction results are very close for 2040 in terms of population. The model incorporates the introduction of public services areas that are distributed equally on the growth areas, which occupy about 15-20% of the total area. This model can help the government to develop areas in a proper way and controls the expansion to have well layout and planned of the city, improving people's standard of living sustainably, while protecting the environment with better planning.

published_or_final_version / Electrical and Electronic Engineering / Master / Master of Philosophy

No description available.

Haptic applications often employ devices with many degrees of freedom in order to allow the user to have natural movement during human-machine interaction. From the development point of view, the complexity in mechanical dynamics imposes a lot of challenges in modelling the behaviour of the device. Traditional system identification methods for nonlinear systems are often computationally expensive. Moreover, current research on using neural network approaches disconnect the physical device dynamics with the identification process. This thesis proposes a different approach to system identification of complex haptic devices when analytical models are formulated. It organizes the unknowns to be identified based on the governing dynamic equations of the device and reduces the cost of computation. All the experimental work is done with the Freedom 6S, a haptic device with input and feedback in positions and velocities for all 6 degrees of freedom . <br /><br /> Once a symbolic model is developed, a subset of the overall dynamic equations describing selected joint(s) of the haptic robot can be obtained. The advantage of being able to describe the selected joint(s) is that when other non-selected joints are physically fixed or locked up, it mathematically simplifies the subset dynamic equation. Hence, a reduced set of unknowns (e. g. mass, centroid location, inertia, friction, etc) resulting from the simplified subset equation describes the dynamic of the selected joint(s) at a given mechanical orientation of the robot. By studying the subset equations describing the joints, a locking sequence of joints can be determined to minimize the number of unknowns to be determined at a time. All the unknowns of the system can be systematically determined by locking selected joint(s) of the device following this locking sequence. Two system identification methods are proposed: Method of Isolated Joint and Method of Coupling Joints. Simulation results confirm that the latter approach is able to successfully identify the system unknowns of Freedom 6S. Both open-loop experimental tests and close-loop verification comparison between the measured and simulated results are presented. <br /><br /> Once the haptic device is modelled, fuzzy logic is used to address chattering phenomenon common to strong virtual effects. In this work, a virtual wall is used to demonstrate this approach. The fuzzy controller design is discussed and experimental comparison between the performance of using a proportional-derivative gain controller and the designed fuzzy controller is presented. The fuzzy controller is able to outperform the traditional controller, eliminating the need for hardware upgrades for improved haptic performance. Summary of results and conclusions are included along with suggested future work to be done.

This thesis focuses on soft sensor development based on fuzzy logic used for real time online monitoring of anaerobic digestion to improve methane output and for robust fermentation. Important process parameter indicators such as pH, biogas production, daily difference in pH and daily difference in biogas production were used to infer alkalinity, a reliable indicator of process stability. Additionally, a fuzzy logic and a rule-based controller were developed and tested with single stage anaerobic digesters operating with cow slurry and cellulose. Alkalinity predictions from the fuzzy logic algorithm were used by both controllers to regulate the organic loading rate that aimed to optimise the biogas process. The predictive performance of a software sensor determining alkalinity that was designed using fuzzy logic and subtractive clustering and was validated against multiple linear regression models that were developed (Partner N° 2, Rothamsted Research 2010) for the same purpose. More accurate alkalinity predictions were achieved by utilizing a fuzzy software sensor designed with less amount of data compared to a multiple linear regression model whose design was based on a larger database. Those models were utilised to control the organic loading rate of a twostage, semi-continuously fed stirred reactor system. Three 5l reactors without support media and three 5l reactors with different support media (burst cell reticulated polyurethane foam coarse, burst cell reticulated polyurethane foam medium and sponge) were operated with cow slurry for a period of seven weeks and twenty weeks respectively. Reactors with support media were proven to be more stable than the reactors without support media but did not exhibit higher gas productivity. Biomass support media were found to influence digester recovery positively by reducing the recovery period. Optimum process parameter ranges were identified for reactors with and without support media. Increased biogas production was found to occur when the loading rates were 3-3.5g VS/l/d and 4-5g VS/l/d respectively. Optimum pH ranges were identified between 7.1-7.3 and 6.9-7.2 for reactors with and without support media respectively, whereas all reactors became unstable at ph<6.9. Alkalinity levels for system stability appeared to be above 3500 mg/l of HCO3 - for reactors without media and 3480 mg/l of HCO3 - for reactors with support media. Biogas production was maximized when alkalinity was 3 between 3500-4500 mg/l of HCO3 - for reactors without support media and 3480- 4300 mg/l of HCO3 - for reactors with support media. Two fuzzy logic models predicting alkalinity based on the operation of the three 5l reactors with support media were developed (FIS I, FIS II). The FIS II design was based on a larger database than FIS I. FIS II performance when applied to the reactor where sponge was used as the support media was characterized by quite good MAE and bias values of 466.53 mg/l of HCO3- and an acceptable value for R2= 0.498. The NMSE was close to 0 with a value of 0.03 and a slightly higher FB= 0.154 than desired. The fuzzy system robustness was tested by adding NaHCO3 to the reactor with the burst cell reticulated polyurethane foam medium and by diluting the reactor where sponge was used as the support media with water. FIS I and FIS II were able to follow the system output closely in the first case, but not in the second. FIS II functionality as an alkalinity predictor was tested through the application on a 28l cylindrical reactor with sponge as the biomass support media treating cow manure. If data that was recorded when severe temperature fluctuations occurred (that highly impact digester performance), are excluded, FIS II performance can be characterized as good by having R2= 0.54 and MAE=Bias= 587 mg/l of HCO3-. Predicted alkalinity values followed observed alkalinity values closely during the days that followed NaHCO3 addition and water dilution. In a second experiment a rulebased and a Mamdani fuzzy logic controller were developed to regulate the organic loading rate based on alkalinity predictions from FIS II. They were tested through the operation of five 6.5l reactors with biomass support media treating cellulose. The performance indices of MAE=763.57 mg/l of HCO3-, Bias= 398.39 mg/l of HCO3-, R2= 0.38 and IA= 0.73 indicate a pretty good correlation between predicted and observed values. However, although both controllers managed to keep alkalinity within the desired levels suggested for stability (>3480 mg/l of HCO3-), the reactors did not reach a stable state suggesting that different loading rates should be applied for biogas systems treating cellulose.

The performance of a quadrotor can be significantly disturbed in presence of wind. In this paper, a simple-to-implement attitude controller is proposed to render a robust and accurate trajectory tracking in presence of disturbance and model uncertainties. The attitude controller design is based on Quantitative Feedback Theory (QFT). A fuzzy logic controller is further employed to provide satisfactory position trajectory tracking for the quadrotor. The performances of the controllers, in terms of disturbance rejection and trajectory tracking are experimentally studied. Finally, a flight scenario is performed to compare the performances of the designed QFT-Fuzzy control scheme with the ArduCopter controller. / May 2016

This thesis develops methodologies for continuous estimation of hydrological variables which infill missing daily rainfall data and the forecast of weekly streamflows from a watershed. Several mathematical programming formulations were developed and used to estimate missing historical rainfall data. Functional relationships were created between radar precipitation and known rain gauge data then are used to estimate the missing data. Streamflow predictions models require highly non-linear mathematical models to capture the complex physical characteristics of a watershed. An artificial neural network model was developed for streamflow prediction. There are no set methods of creating a neural network and the selection of architecture and inputs to a neural network affects the performance. This thesis addresses this issue with automated input and network architecture selection through optimization. MATLABª scripts are developed and used to test many combinations and select a model through optimization. / by Ricardo Eric Brown. / Thesis (M.S.C.S.)--Florida Atlantic University, 2012. / Includes bibliography. / Electronic reproduction. Boca Raton, Fla., 2012. Mode of access: World Wide Web.