Short-term electricity price forecasting is now crucial practice in deregulated electricity markets, as it forms the basis for maximizing the profits of the market participants. In this thesis, short-term electricity prices are forecast using three different predictor schemes, Artificial Neural Networks (ANNs), Support Vector Machine (SVM) and a hybrid scheme, respectively. ANNs are the very popular and successful tools for practical forecasting. In this thesis, a hidden-layered feed-forward neural network with back-propagation has been adopted for detailed comparison with other forecasting models. SVM is a newly developed technique that has many attractive features and good performance in terms of prediction. In order to overcome the limitations of individual forecasting models, a hybrid technique that combines Fuzzy-C-Means (FCM) clustering and SVM regression algorithms is proposed to forecast the half-hour electricity prices in the UK electricity markets. According to the value of their power prices, thousands of the training data are classified by the unsupervised learning method of FCM clustering. SVM regression model is then applied to each cluster by taking advantage of the aggregated data information, which reduces the noise for each training program. In order to demonstrate the predictive capability of the proposed model, ANNs and SVM models are presented and compared with the hybrid technique based on the same training and testing data sets in the case studies by using real electricity market data. The data was obtained upon request from APX Power UK for the year 2007. Mean Absolute Percentage Error (MAPE) is used to analyze the forecasting errors of different models and the results presented clearly show that the proposed hybrid technique considerably improves the electricity price forecasting.

The fiber reinforced polymer (FRP) bars have been widely used in pre-stressing applications and reinforcing of the civil structures. High strength-to-weight ratio and high resistance to the corrosion make the FRP bars a good replacement for steel reinforcing bars in civil engineering applications. According to the CAN/CSA-S806-12 standard, the maximum recommended stress in FRP bars under service loads should not exceed 25% and 65% of the ultimate strength for glass FRP (GFRP) and carbon FRP (CFRP), respectively. These stress values are set to prevent creep failure in FRP bars. However, for in-service applications, there are few physical indicators that these values have been reached or exceeded. In this work analysis of acoustic emission (AE) signals is used. Two new techniques based on pattern recognition and frequency entropy of the isolated acoustic emission (AE) signal are presented for monitoring damage progression and prediction of failure in FRPs. / May 2017

We present two Swarm Intelligence based approaches for data clustering. The first algorithm, Fuzzy Ants, presented in this thesis clusters data without the initial knowledge of the number of clusters. It is a two stage algorithm. In the first stage the ants cluster data to initially create raw clusters which are refined using the Fuzzy C Means algorithm. Initially, the ants move the individual objects to form heaps. The centroids of these heaps are redefined by the Fuzzy C Means algorithm. In the second stage the objects obtained from the Fuzzy C Means algorithm are hardened according to the maximum membership criteria to form new heaps. These new heaps are then moved by the ants. The final clusters formed are refined by using the Fuzzy C Means algorithm. Results from experiments with 13 datasets show that the partitions produced are competitive with those from FCM. The second algorithm, Fuzzy ant clustering with centroids, is also a two stage algorithm, it requires an initial knowledge of the number of clusters in the data. In the first stage of the algorithm ants move the cluster centers in feature space. The cluster centers found by the ants are evaluated using a reformulated Fuzzy C Means criterion. In the second stage the best cluster centers found are used as the initial cluster centers for the Fuzzy C Means algorithm. Results on 18 datasets show that the partitions found by FCM using the ant initialization are better than those from randomly initialized FCM. Hard C Means was also used in the second stage and the partitions from the ant algorithm are better than from randomly initialized Hard C Means. The Fuzzy Ants algorithm is a novel method to find the number of clusters in the data and also provides good initializations for the FCM and HCM algorithms. We performed sensitivity analysis on the controlling parameters and found the Fuzzy Ants algorithm to be very sensitive to the Tcreateforheap parameter. The FCM and HCM algorithms, with random initializations can get stuck in a bad extrema, the Fuzzy ant clustering with centroids algorithm successfully avoids these bad extremas.

Orientador: Luís Roberto Almeida Gabriel Filho / Coorientador: Luiz Fernando Sommaggio Coletta / Coorientador: Camila Pires Cremasco Gabriel / Resumo: O presente trabalho buscou identificar padrões na produção de cana de por meio da utilização de Inteligência Artificial. Para tanto, foi realizada coleta de informações de fontes secundárias, com dados estatísticos fornecidos por órgãos públicos sobre a área cultivada e a produção de cana de açúcar, índices como pluviométricos e de temperatura e o tipo de solo dos municípios do estado de São Paulo, no ano de 2017, por meio de pesquisa documental. Com a utilização dos métodos Floresta dos Caminhos Ótimos (OPF), K-means e Fuzzy C-means (FCM) buscou-se identificar clusters, ou padrões, que representem essas características produtivas. Além disso, o trabalho testou a utilização do algoritmo OPF como ferramenta de apoio à decisão no setor agroindustrial e fez a comparação do método com os agrupadores de padrões K-means e FCM. Após o processamento dos dados foi possível identificar padrões na produção de cana de açúcar pelos três algoritmos, sendo que o OPF proporcionou resultados muito parecidos com o K-means e FCM, confirmando a eficiência do método. Além disso, foi possível identificar, no ano de 2017, um padrão de produção com municípios com alta produtividade, grandes áreas destinadas a produção de cana de açúcar e produção da cultura, com temperatura média alta e índices pluviométricos baixos. Os municípios que possuem pequenas áreas com plantação de cana de açúcar possuem uma variabilidade muito grande em resultados de produtividade. O padrão de município com baixa produtivi... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: The present work sought to identify patterns in sugarcane production through the use of Artificial Intelligence. For this purpose, information was collected from secondary sources, with statistical data provided by public agencies on cultivated area and sugarcane production, rainfall and temperature indices, and the soil type of the municipalities of the State of São Paulo, in the year 2017, through documentary research. Using Optimum-Path Forest (OPF), K-means and Fuzzy C-means (FCM) methods, the aim was to identify clusters, or patterns, that represent these productive characteristics. In addition, the work tested the use of OPF algorithm as a decision support tool in the agribusiness sector and compared the method with the K-means and FCM standards groupers. After data processing, it was possible to identify patterns in sugarcane production by the three algorithms, and OPF provided results very similar to K-means and FCM, confirming the efficiency of the method. In addition, it was possible to identify, in the year 2017, a production pattern of municipalities with high productivity, large areas destined to the production of sugar cane and crop production, with high average temperature and low rainfall. Municipalities that have small areas with sugar cane plantation have a very large variability in productivity results. The municipal pattern with low productivity is accompanied by very low average temperature, very high rainfall rates and soils of the type Cambisols, Neosols and Spodosols. The soil type pattern that provided the highest productivity for the municipalities was the Oxisol. / Mestre

The aim of strategic group analysis is to find out if clusters of firms that have a similar strategic position exist within an industry or not. In this thesis, by using a conceptual framework that reflects the strategic context, contents and process of construction companies and utilising alternative clustering methods such as traditional cluster analysis, self-organizing maps, and fuzzy C-means technique, a strategic group analysis was conducted for the Turkish construction industry. Results demonstrate that there are three strategic groups among which significant performance differences exist. Self-organising maps provide a visual representation of group composition and help identification of hybrid structures. Fuzzy C-means technique reveals the membership degrees of a firm to each strategic group. It is recommended that real strategic group structure can only be identified by using alternative cluster analysis methods. The positive effect of differentiation strategy on achieving competitive advantage is widely acknowledged in the literature and proved to be valid for the Turkish construction industry as a result of strategic group analysis. In this study, a framework is proposed to model the differentiation process in construction. The relationships between the modes and drivers of differentiation are analyzed by structural equation modeling. The results demonstrate that construction companies can either differentiate on quality or productivity. Project management related factors extensively influence productivity differentiation whereas they influence quality differentiation indirectly. Corporate management related factors only affect quality differentiation. Moreover, resources influence productivity differentiation directly whereas they have an indirect effect on quality differentiation.

Up to now various recommendation systems have been proposed for web based applications such as e-commerce and information retrieval where a large amount of product or information is available. Basically, the task of the recommendation systems in those applications, for example the e-commerce, is to find and recommend the most relevant items to users/customers. In this domain, the most prominent approaches are collaborative filtering and content-based filtering. Sometimes these approaches are called as user profiling as well. In this work, a context-aware recommendation system is proposed for mobile environment, which also can be considered as an extension of those recommendation systems proposed for web-based information retrieval and e-commerce applications. In the web-based information retrieval and e-commerce applications, for example in an online book store (e-commerce), the users&amp / #8217 / actions are independent of their instant context (location, time&amp / #8230 / etc). But as for mobile environment, the users&amp / #8217 / actions are strictly dependent on their instant context. These dependencies give raise to need of filtering items/actions with respect to the users&amp / #8217 / instant context. In this thesis, an approach coupling approaches from two different domains, one is the mobile environment and other is the web, is proposed. Hence, it will be possible to separate whole approach into two phases: context-aware prediction and user profiling. In the first phase, combination of two methods called fuzzy c-means clustering and learning automata will be used to predict the mobile user&amp / #8217 / s motions in context space beforehand. This provides elimination of a large amount of items placed in the context space. In the second phase, hierarchical fuzzy clustering for users profiling will be used to determine the best recommendation among the remaining items.

Submitted by Roberto Luvisutto Martins (rlmartins.agro@hotmail.com) on 2018-05-25T15:17:52Z No. of bitstreams: 1 Dissertação Roberto Luvisutto Martins.pdf: 1862150 bytes, checksum: 7a9e99015cb69cd2a388498ceb3cc6ba (MD5) / Approved for entry into archive by Maria Lucia Martins Frederico null (mlucia@fca.unesp.br) on 2018-05-25T17:14:31Z (GMT) No. of bitstreams: 1 martins_rl_me_botfca.pdf: 1862150 bytes, checksum: 7a9e99015cb69cd2a388498ceb3cc6ba (MD5) / Made available in DSpace on 2018-05-25T17:14:31Z (GMT). No. of bitstreams: 1 martins_rl_me_botfca.pdf: 1862150 bytes, checksum: 7a9e99015cb69cd2a388498ceb3cc6ba (MD5) Previous issue date: 2018-02-02 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / O trabalho teve como objetivo delimitar unidades de monitoramento da umidade do solo por meio da aplicação da análise geoestatística e multivariada em dados de textura do solo de um pomar de videira de vinho cv. Chardonnay / Paulsen 1103, de 1,1 ha, localizado em Espirito Santo do Pinhal, São Paulo. Em 54 pontos georreferenciados, foram coletadas 108 amostras de solo, nas camadas de 0-0,20 e 0,20-0,40 m, para determinação da textura do solo (areia, silte e argila) e estimativa do conteúdo de água disponível (AD) e água prontamente disponível (APD). Os dados de textura do solo, AD e APD foram submetidos à análise estatística descritiva e ao teste de Kolmogorov–Smirnov. Posteriormente, os dados de textura do solo foram submetidos à análise geoestatística e interpolação por krigagem ordinária, para determinação das unidades de monitoramento da umidade do solo por meio das análises de componentes principais e de agrupamento fuzzy c-means. O melhor número de unidades de foi determinado a partir de dois índices, sendo quatro o resultado obtido, para ambas as camadas, as quais apresentaram diferença significativa tanto para as três frações granulométricas do solo quanto para AD e APD. Unidades de monitoramento com os valores maiores de argila e menores de silte contribuíram para os menores valores de AD e APD. A utilização restrita dos dados de textura do solo permite o delineamento de unidades de monitoramento da umidade do solo com elevada representatividade do conteúdo de água disponível do solo. / The objective of this study was to delimit soil water monitoring zones through the application of geostatistical and multivariate analysis in soil texture data of a 1.1 ha wine vine orchard cv. Chardonnay / Paulsen 1103, located at Espirito Santo do Pinhal, State of São Paulo, Brazil. In 54 georeferenced points, 108 soil samples were collected in the 0-0.20 m and 0.20-0.40 m layers to determine the percentage of soil granulometric fractions (sand, silt and clay), available soil water content (AW) and readily available soil water content (RAW). Soil fractions, AW and RAW data were submitted to descriptive statistics and Kolmogorov– Smirnov test. Soil granulometric fraction data were submitted to geostatistical analysis to characterize the data spatial distribution and their interpolation. Then, principal component and fuzzy c-means cluster analyses were applied for the determination of the best number of soil water monitoring zones based on two indexes. Four zones for both soil layers were determined, and they showed significant difference for both soil granulometric fractions as for AW and RAW. Zones with the highest values of clay fraction and smallest values for silt fraction gave the lowest values of AW and RAW. The use of soil texture data allows the design of soil water monitoring units with high representativeness of the available soil water content.

As we all know the classical set theory has a deep-rooted influence in the traditional mathematics. According to the two-valued logic, an element can belong to a set or cannot. In the former case, the element’s membership degree will be assigned to one, whereas in the latter case it takes the zero value. With other words, a feeling of imprecision or fuzziness in the two-valued logic does not exist. With the rapid development of science and technology, more and more scientists have gradually come to realize the vital importance of the multi-valued logic. Thus, in 1965, Professor Lotfi A. Zadeh from Berkeley University put forward the concept of a fuzzy set. In less than 60 years, people became more and more familiar with fuzzy set theory. The theory of fuzzy sets has been turned to be a favor applied to many fields. The study aims to apply some classical and extensional methods of fuzzy set theory in life expectancy and treatment prognoses for cancer patients. The research is based on real-life problems encountered in clinical works by physicians. From the introductory items of the fuzzy set theory to the medical applications, a collection of detailed analysis of fuzzy set theory and its extensions are presented in the thesis. Concretely speaking, the Mamdani fuzzy control systems and the Sugeno controller have been applied to predict the survival length of gastric cancer patients. In order to keep the gastric cancer patients, already examined, away from the unnecessary suffering from surgical operation, the fuzzy c-means clustering analysis has been adopted to investigate the possibilities for operation contra to nonoperation. Furthermore, the approach of point set approximation has been adopted to estimate the operation possibilities against to nonoperation for an arbitrary gastric cancer patient. In addition, in the domain of multi-expert decision-making, the probabilistic model, the model of 2-tuple linguistic representations and the hesitant fuzzy linguistic term sets (HFLTS) have been utilized to select the most consensual treatment scheme(s) for two separate prostate cancer patients. The obtained results have supplied the physicians with reliable and helpful information. Therefore, the research work can be seen as the mathematical complements to the physicians’ queries.

Automated visual inspection is an emerging multi-disciplinary field with many challenges; it combines different aspects of computer vision, pattern recognition, automation, and control systems. There does not exist a large body of work dedicated to the design of generalized visual inspection systems; that is, those that might easily be made applicable to different product types. This is an important oversight, in that many improvements in design and implementation times, as well as costs, might be realized with a system that could easily be made to function in different production environments. <br /><br /> This thesis proposes a framework for generalizing and automating the design of the defect classification stage of an automated visual inspection system. It involves using an expandable set of features which are optimized along with the classifier operating on them in order to adapt to the application at hand. The particular implementation explored involves optimizing the feature set in disjoint sets logically grouped by feature type to keep search spaces reasonable. Operator input is kept at a minimum throughout this customization process, since it is limited only to those cases in which the existing feature library cannot adequately delineate the classes at hand, at which time new features (or pools) may have to be introduced by an engineer with experience in the domain. <br /><br /> Two novel methods are put forward which fit well within this framework: cluster-space and hybrid-space classifiers. They are compared in a series of tests against both standard benchmark classifiers, as well as mean and majority vote multi-classifiers, on feature sets comprised of just the logical feature subsets, as well as the entire feature sets formed by their union. The proposed classifiers as well as the benchmarks are optimized with both a progressive combinatorial approach and with an genetic algorithm. Experimentation was performed on true colour industrial lumber defect images, as well as binary hand-written digits. <br /><br /> Based on the experiments conducted in this work, it was found that the sequentially optimized multi hybrid-space methods are capable of matching the performances of the benchmark classifiers on the lumber data, with the exception of the mean-rule multi-classifiers, which dominated most experiments by approximately 3% in classification accuracy. The genetic algorithm optimized hybrid-space multi-classifier achieved best performance however; an accuracy of 79. 2%. <br /><br /> The numeral dataset results were less promising; the proposed methods could not equal benchmark performance. This is probably because the numeral feature-sets were much more conducive to good class separation, with standard benchmark accuracies approaching 95% not uncommon. This indicates that the cluster-space transform inherent to the proposed methods appear to be most useful in highly dependant or confusing feature-spaces, a hypothesis supported by the outstanding performance of the single hybrid-space classifier in the difficult texture feature subspace: 42. 6% accuracy, a 6% increase over the best benchmark performance. <br /><br /> The generalized framework proposed appears promising, because classifier performance over feature sets formed by the union of independently optimized feature subsets regularly met and exceeded those classifiers operating on feature sets formed by the optimization of the feature set in its entirety. This finding corroborates earlier work with similar results [3, 9], and is an aspect of pattern recognition that should be examined further.

Automated visual inspection is an emerging multi-disciplinary field with many challenges; it combines different aspects of computer vision, pattern recognition, automation, and control systems. There does not exist a large body of work dedicated to the design of generalized visual inspection systems; that is, those that might easily be made applicable to different product types. This is an important oversight, in that many improvements in design and implementation times, as well as costs, might be realized with a system that could easily be made to function in different production environments. <br /><br /> This thesis proposes a framework for generalizing and automating the design of the defect classification stage of an automated visual inspection system. It involves using an expandable set of features which are optimized along with the classifier operating on them in order to adapt to the application at hand. The particular implementation explored involves optimizing the feature set in disjoint sets logically grouped by feature type to keep search spaces reasonable. Operator input is kept at a minimum throughout this customization process, since it is limited only to those cases in which the existing feature library cannot adequately delineate the classes at hand, at which time new features (or pools) may have to be introduced by an engineer with experience in the domain. <br /><br /> Two novel methods are put forward which fit well within this framework: cluster-space and hybrid-space classifiers. They are compared in a series of tests against both standard benchmark classifiers, as well as mean and majority vote multi-classifiers, on feature sets comprised of just the logical feature subsets, as well as the entire feature sets formed by their union. The proposed classifiers as well as the benchmarks are optimized with both a progressive combinatorial approach and with an genetic algorithm. Experimentation was performed on true colour industrial lumber defect images, as well as binary hand-written digits. <br /><br /> Based on the experiments conducted in this work, it was found that the sequentially optimized multi hybrid-space methods are capable of matching the performances of the benchmark classifiers on the lumber data, with the exception of the mean-rule multi-classifiers, which dominated most experiments by approximately 3% in classification accuracy. The genetic algorithm optimized hybrid-space multi-classifier achieved best performance however; an accuracy of 79. 2%. <br /><br /> The numeral dataset results were less promising; the proposed methods could not equal benchmark performance. This is probably because the numeral feature-sets were much more conducive to good class separation, with standard benchmark accuracies approaching 95% not uncommon. This indicates that the cluster-space transform inherent to the proposed methods appear to be most useful in highly dependant or confusing feature-spaces, a hypothesis supported by the outstanding performance of the single hybrid-space classifier in the difficult texture feature subspace: 42. 6% accuracy, a 6% increase over the best benchmark performance. <br /><br /> The generalized framework proposed appears promising, because classifier performance over feature sets formed by the union of independently optimized feature subsets regularly met and exceeded those classifiers operating on feature sets formed by the optimization of the feature set in its entirety. This finding corroborates earlier work with similar results [3, 9], and is an aspect of pattern recognition that should be examined further.