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.

Made available in DSpace on 2017-07-21T18:53:09Z (GMT). No. of bitstreams: 1 Maria Isabel Stets.pdf: 1158333 bytes, checksum: c7c568acfaf2daf456026f86016faf6b (MD5) Previous issue date: 2008-03-12 / The effluent generated in abattoirs has a high concentration of organic matter and need treatment before being discarded in the water bodies. Anaerobic biological filters are considered a good technique for the treatment of industrial effluents; however, choosing the correct support media is extremely important to ensure the success of the reactor. In the same way, the study of microbial diversity can optimize the performance and the operation of these systems. In this work it was evaluated the efficiency of the treatment of the effluent from slaughterhouse, using three upflow anaerobic biological filters, and it was compared the microbial diversity of these filters. The reactor called A, B and C were built in chloride polivinile and filled with rings of polypropylene, polyurethane foam, or pieces of clay brick, respectively. These reactors were operated at room temperature and the used Hydraulic Retention Times of (HRTs) were of 30, 20, 18, 14, 10, 8, 6, 5, 4, 3, 2, 1.5 and 1 day. To monitor the efficiency of the process, the pH, alkalinity, acidity, Chemistry Oxygen Demand (COD), and total solids content of volatile solids, nitrogen and phosphorus substrate and effluents of each reactor were analyzed. The values of pH decreased with the reduction of HRT and the ratio volatile acidity / alkalinity remained below 0.13 during the majority of the analyzed period. In the one day’s HRT, the reactor C was more effective reaching 80.76 of removal. There was a high rate of nitrogen removal probably due to its low concentration in the substrate. High phosphorus removals also occurred: there was a decrease of phosphorus removal when the HRT declines in reactors A and B, the same did not occur in the reactor C. The three reactors behaved similarly regarding to the collecting sites and there was no difference between these sites when COD removal, volatile acidity / alkalinity ratio and pH were analyzed. To evaluate the effect of different media support in the microbial community, the reactors DNA biomass was extracted, quantified and used as template in an amplification reaction of 16S rDNA gene from bacteria and metanogenic. The amplified fragments of approximately 1500 bp were treated with HinfI, RsaI and HaeIII restriction enzymes to obtain the reactors ARDRA (Amplified Ribossomal DNA Restriction Analysis) profiles. The indices of richness, richness modified, Shannon-Weaver diversity index and similarity dendogram, determined from ARDRA, grouped the reactors A and B, and left the reactor C isolated due to its dissimilarity. The Hierarchical Cluster Analysis (HCA) also grouped the reactors that way, however, that group has less than 0.1% of similarity, suggesting a low interference in the reactors media support in microbial communities. / O efluente gerado em abatedouros possui uma elevada concentração de matéria orgânica e necessita de tratamento antes de ser descartado nos corpos hídricos. Os filtros biológicos anaeróbios são considerados uma boa técnica para tratamento de efluentes industriais, porém, a escolha correta do meio suporte é extremamente importante para garantir o sucesso do reator. Da mesma forma, o estudo da diversidade microbiana pode otimizar o desempenho e a operação desses sistemas. Neste trabalho, foi realizada a avaliação da eficiência do tratamento de efluente de abatedouro, utilizando-se três filtros biológicos anaeróbios de fluxo ascendente, bem como a comparação da diversidade microbiana desses filtros. Os reatores denominados A, B e C foram construídos em cloreto de polivinila e preenchidos com anéis de polipropileno, espuma de poliuretano ou pedaços de tijolo de argila, respectivamente. Esses reatores foram operados à temperatura ambiente e os Tempos de Retenção Hidráulica (TRHs) utilizados foram de 30; 20; 18; 14; 10; 8; 6; 5; 4; 3; 2; 1,5 e 1 dia. Para monitorar a eficiência do processo, foram analisados no substrato e efluente de cada reator pH, alcalinidade, acidez, Demanda Química de Oxigênio (DQO), teor de sólidos totais e sólidos voláteis, nitrogênio efósforo. Os valores de pH do efluente dos reatores decresceram com a redução do TRH. A relação acidez volátil/alcalinidade ficou abaixo de 0,13 durante a maior parte do período analisado. No TRH de 1 dia o reator C foi o mais eficiente alcançando 80,76 de remoção de DQO. Observou-se uma elevada taxa de remoção de nitrogênio em todos os reatores provavelmente devido à sua baixa quantidade no substrato. Também ocorreram elevadas remoções de fósforo, mas houve decréscimo dessa remoção com a diminuição do TRH nos reatores A e B, o mesmo não ocorrendo no reator C. Quando se analisaram as amostras retiradas em quatro diferentes alturas dos reatores, observou-se que os três se comportaram de forma similar quanto aos pontos de coleta e não houve diferença entre esses pontos quanto à remoção de DQO, relação acidez volátil/alcalinidade e pH. Para avaliar o efeito dos diferentes meios suporte na comunidade microbiana, o DNA da biomassa dos reatores foi extraído, quantificado e usado como molde numa reação de amplificação do gene 16S rDNA de bactérias e archaeas metanogênicas. Os fragmentos de aproximadamente 1500 pb amplificados foram tratados com as enzimas de restrição HinfI, RsaI e HaeIII para obtenção dos perfis de ARDRA (Análise da Restrição de rDNA Amplificado) dos reatores. Os índices de riqueza, riqueza modificada, diversidade de Shannon-Weaver e dendograma de similaridade, determinados a partir da ARDRA, agruparam os reatores A e B, ficando o reator C isolado devido à sua dessemelhança. A Análise de Agrupamento Hierárquico (HCA) também agrupou os reatores dessa forma, entretanto, esse agrupamento possui menos que 0,1% de similaridade, sugerindo uma baixa interferência do meio suporte nas comunidades microbianas.