[pt] As RNA Temporais, em função de sua estrutura, consideram o tempo na sua operação, incorporando memória de curto prazo distribuída na rede em todos os neurônios escondidos e em alguns dos casos nos neurônios de saída. Esta classe de redes é utilizada para representar melhor a natureza temporal dos sistemas dinâmicos. Em contraste, a RNA estática tem uma estrutura apropriada para tarefas de reconhecimento de padrões, classificação e outras de natureza estática ou estacionária tendo sido utilizada com sucesso em diversas aplicações. O objetivo desta tese, portanto foi estudar a teoria e avaliar o desempenho das Redes Neurais Temporais em comparação com as Redes Neurais Estáticas, em aplicações de sistemas dinâmicos. O desenvolvimento desta pesquisa envolveu 3 etapas principais: pesquisa bibliográfica das metodologias desenvolvidas para RNA Temporais; seleção e implementação de modelos para a avaliação destas redes; e estudo de casos. A pesquisa bibliográfica permitiu compila e classificar os principais trabalhos sobre RNA Temporais. Tipicamente, estas redes podem ser classificadas em dois grupos: Redes com Atraso no Tempo e Redes Recorrentes. Para a análise de desempenho, selecionou-se uma redee de cada grupo para implementação. Do primeiro grupo foi selecionada a Rede FIR, onde as sinapses são filtros FIR (Finite-duration Impulse Response) que representam a natureza temporal do problema. A rede FIR foi selecionada por englobar praticamente, todos os outros métodos de sua classe e apresentar um modelo matemático mais formal. Do segundo grupo, considerou-se a rede recorrente de Elman que apresenta realimentação global de cada um dos neurônios escondidos para todos eles. No estudo de casos testou-se o desempenho das redes selecionadas em duas linhas de aplicação: previsão de séries temporais e processamento digital de sinais. No caso de previsão de séries temporais, foram utilizadas séries de consumo de energia elétrica, comparando-se os resultados com os encontrados na literatura a partir de métodos de Holt-Winters, Box & Jenkins e RNA estáticas. No caso da aplicação das RNA em processamento digital de sinais, utilizou-se a filtragem de ruído em sinais de voz onde foram feitas comparações com os resultados apresentados pelo filtro neural convencional, que é uma rede feed-forward multicamada com o algoritmo de retropropagação para o aprendizado. Este trabalho demonstrou na prática que as RNA temporais conseguem capturar as características dos processos temporais de forma mais eficiente que as RNA Estatísticas e outros métodos tradicionais, podendo aprender diretamente o comportamento não estacionário das séries temporais. Os resultados demonstraram que a rede neural FIR e a rede Elman aprendem melhor a complexidade dos sinais de voz. / [en] This dissertation investigates the development of Artificial Neural Network (ANN) in the solution of problems where the patterns presented to the network have a temporary relationship to each other, such as time series forecast and voice processing. Temporary ANN considers the time in its operation, incorporating memory of short period distributed in the network in all the hidden neurons and in the output neurons in some cases. This class of network in better used to represent the temporary nature of the dynamic systems. In contrast, Static ANN has a structure adapted for tasks of pattern recognition, classification and another static or stationary problems, achieving great success in several applications. Considered an universal approximator, Static ANN has also been used in applications of dynamic systems, through some artifices in the input of the network and through statistical data pre- processings. The objective of this work is, therefore to study the theory and evaluate the performance of Temporal ANN, in comparison with Static ANN, in applications of dynamics systems. The development of this research involved 3 main stages: bibliographical research of the methodologies developed for Temporal ANN; selection and implementation of the models for the evaluation of these networks; and case studies. The bibliographical research allowed to compile and to classify the main on Temporal ANN, Typically, these network was selected, where the synapses are filters FIR (Finite-duration Impulse Response) that represent the temporary nature of the problem. The FIR network has been selected since it includes practically all other methods of its class, presenting a more formal mathematical model. On the second group, the Elman recurrent network was considered, that presents global feedback of each neuron in the hidden layer to all other neurons in this layer. In the case studies the network selected have been tested in two application: forecast of time series and digital signal processing. In the case of forecast, result of electric energy consumption time series prediction were compared with the result found in the literature such as Holt-Winters, Box & Jenkins and Static ANN methods. In the case of the application of processing where the comparisons were made with the results presented by the standard neural filter, made of a multilayer feed-forward network with the back propagation learning algorithm. This work showed in practice that Temporal ANN captures the characteristics of the temporary processes in a more efficient way that Static ANN and other methods, being able to learn the non stationary behavior of the temporary series directly. The results showed that the FIR neural network and de Elman network learned better the complexity of the voice signals.

[pt] O dióxido de titânio é amplamente utilizado pela indústria e pesquisa como fotocatalisador, cuja principal desvantagem ainda é sua aplicação sob luz visível. Propriedades como quantidade de fases, tamanho do cristalito, área de superfície específica, volume de poros e valor da banda proibida (Eg) são explorados por métodos de síntes e para aprimorar a performance do TiO2. No entanto, elas são ajustadas empiracamente. O presente trabalho foi realizado a fim de descrever uma relação analítica entre essas propriedades para a fotocatálise, usando Redes Neurais Artificiais (RNAs) como ferramente estatística. Afim de ter o banco de dados mais representativo, foram usados 53 artigos. O Eg foi considerado a medida a qual avalia a performance fotocatalítica, sendo o parâmetro de saída da rede. Dois blocos A e B, distintos pelas variáveis de entrada, foram arranjados em grupos para investigar a influência das variáveis em pares, com 257 e 220 fotocatalisadores para cada, respectivamente. Exploraram-se diferentes algoritmos de treinamento (baseados em Retropropagação), tipos de redes (Feedforward, Cascade forward e Elman), funções de transferência, número de neurônios e redemulticamadas. Avaliaram-se os modelos pela Soma dos Erros Quadráticos (SSE),pelo coeficiente de correlação de regressão (R2) tanto para o treinamento e quanto para o teste, pelo comportamento de predição do banco de dados e pelo diagrama de regressão dos valores preditos pelos observados. Os resultados do bloco A sugerem que as variáveis não aparentam ter uma relação. Os modelos de múltiplas camadas no bloco B revelaram um aumento no desempenho. O resultado de maior coeficiente teve topologia de 4-4-6-1, correspondendo a camada de entrada, primeira camada oculta, segunda camada oculta e camda de saída, respectivamente. Obteve-se R2 de 84 por cento para o treinamento e 50 por cento para o teste, com SSE de 2.24.Esse resultado sugere que a rede não é capaz de prever o Eg, mas ela pode ser aprimorada. Os parâmetros estruturais devem ser revisados, de acordo com padrões de caracterizações e dados estatísticos. Consequentemente, o modelo pode ser bem ajustado, otimizado e usado na melhoria da fotocatálise. / [en] Titanium dioxide has been widely applied by industry and scientific research as a photocatalyst,whose main drawback still has been the application under visible light.Properties such as phases amount,crystallite size, specific surface area, pore volume, and band gap value (Eg)have been explored by synthesis methods to improve TiO2 s performance. However, they are empirically adjusted.The present work was carried out to describe an analytical relation between those properties for photocatalysis, using Artificial Neural Networks (ANNs) as a statistical tool. Aiming the most representative set, 53 literature papers were used for the database. Eg was considered the measurement which evaluates the photocatalytic performance, namely the network s out put variable. Two blocks A and B, which are distinguished by input variables, were arranged into groups to investigate the variables pair influences, using 257 and 220 photocatalysts vectors for each,respectively. Modelling attempts examined different training algorithms(based on Back- propagation), types of networks (Feedforward, Cascade forward and Elman), transfer functions, number of hidden neurons, and multilayer network.The developedmodelswereevaluatedbythesumofsquarederror(SSE),the correlation coefficient(R2) of regression for both training and test data, the prediction behaviour of the dataset,and the regression diagram of predicted and observed values. The block A results suggest the variables do not have an apparent relationship. Multilayers models on block B revealed an increase of network identification performance. The result with the highest coefficient showed 4-4-6-1 topology; corresponding, respectively, to input, first hidden, second hidden and output layers.It had R2 of 84 percent for training and to 50 percent fortest, with SSE of 2.24.This result suggests this network is not able topredict the Eg, but it can be improved. The structural properties should be reviewed, according to standards of characterization and statistical data. Hence, the model could be well fitted, optimized, and used for photocatalysis improvement.

Made available in DSpace on 2016-08-10T10:40:27Z (GMT). No. of bitstreams: 1 ANA PAULA DE SOUSA.pdf: 965882 bytes, checksum: a3647999f994441f4537855527b52292 (MD5) Previous issue date: 2012-03-09 / The objective of this study was to compare statistical methods and artificial intelligence to the problem of time series forecasting using Holt-Winters, Box-Jenkins and the Elman neural network. The models were used to predict one step ahead of the price of ethanol in the state of Goias and compared using measures of specific errors. At the end, the results indicated that all three techniques were competitive in terms of predicting one step ahead especially the statistical models appeared to be the most suitable methods in terms of balance between performance and complexity. / O objetivo deste trabalho foi comparar os métodos de estatística e de inteligência artificial para o problema da previsão de séries temporais através de Holt-Winters, Box- Jenkins e a rede neural de Elman. Os modelos foram utilizados para previsão um passo a frente dos preços do etanol no estado de Goiás e comparados através medidas de erros específicas. Ao final, os resultados indicaram que todos os métodos se mostraram competitivos em termos de predição um passo à frente, destacando-se os modelos estatísticos como os mais adequados em termos de parcimônia entre desempenho e complexidade.

Detta arbete utvärderar ANN-arkitekturerna feedforward, Elman och ESCN då de används för att styra en bil i en enkel 2D-simulering. Nätverken tränas av en evolutionär algoritm som använder nätverkens vikter som genom för dess individer. Syftet med arbetet är att se om arkitekturerna presterar olika bra. Simuleringens komplexitet, i form av halka och sladd, samt banans svårighetsgrad varieras för att se vilka arkitekturer som klarar vilka komplexa problem bäst och var de eventuellt brister. Ett program utvecklades som testade de olika fallen och resultatet visade att Elman presterade sämst, speciellt då komplexiteten ökade, och ESCN presterade lite bättre än feedforward. Varför Elman presterade sämre fick inget svar i detta arbete, och ESCN använde sitt minne på ett sätt som skulle kunna vara värt att titta vidare på. Framtida arbete skulle kunna vara att ta reda på orsakerna till de ovanliga beteendena som uppstod samt att genomföra mer utförliga tester.

[pt] A humanidade está vivenciando a Quarta Revolução Industrial, caracterizada pela implementação global da internet, utilização de inteligência artificial e automatização dos processos. Este último é de grande importância para indústria química, uma vez que seu desenvolvimento possibilitou um aumento significativo da quantidade de dados armazenados diariamente, o que gerou uma demanda para análise desses dados. Este enorme fluxo de informações tornou o sistema cada vez mais complexo com uma aleatoriedade de falhas no processo que se identificadas poderiam ajudar a melhorar o processo e evitar acidentes. Uma solução ainda pouco comum na indústria, porém com grande potencial para identificar estas falhas de processo com excelência, é a emergente inteligência artificial. Para lidar com esta questão, o presente trabalho realiza a detecção e identificação de falhas em processos industriais através da modelagem de redes neurais artificias. O banco de dados foi obtido através do uso do benchmark de processo Tennessee Eastman, implementado no Software Matlab 2017b, o qual foi projetado para simular uma planta química completa. A enorme quantidade de dados gerados pelo processo tornou possível a simulação em um contexto de Big Data. Para modelagem dos dados, foram tanto aplicadas redes neurais tradicionais feedforward, quanto redes recorrentes: Rede de Elman e Echo State Network. Os resultados apontaram que as redes feedforward e de Elman obtiveram melhores desempenhos analisados pelo coeficiente de determinação (R2). Assim, o primeiro modelo obteve melhor topologia com 37x60x70x1, algoritmo de treinamento trainlm, funções de ativação tansig para as duas camadas intermediárias e camada de saída ativada pela purelin com R2 de 88,69 por cento. O modelo da rede de Elman apresentou sua melhor topologia com 37x45x55x1, algoritmo de treinamento trainlm, funções de ativação tansig para as duas camadas intermediárias e camada de saída ativada pela função purelin com R2 de 83,63 por cento. Foi concluido que as redes analisadas podem ser usadas em controle preditivo de falhas em processos industriais, podendo ser aplicadas em plantas químicas no futuro. / [en] Humanity is experiencing the 4th Industrial Revolution, characterized by the global implementation of the internet, use of artificial intelligence and automation of processes. The last one is of great importance for the chemical industry, since its development allowed a significant increase in the amount of data stored daily, which generated a demand for the analysis of this data. This enormous flow of information made the system more and more complex with a randomness of process faults that if identified could help improve the process and prevent accidents. A solution not yet common in industry, but with great potential to identify these process faults with excellence, is the emergent artificial intelligence. To deal with this issue, the present work performs fault detection and diagnosis in industrial processes through artificial neural networks modeling. The database was obtained using the benchmark of processes Tennessee Eastman, implemented in Matlab 2017b Software, which is designed to simulate a complete chemical plant. The huge amount of data generated by the process made it possible to simulate in a Big Data context. For data modeling, were applied both traditional feedforward neural networks as well as recurrent networks: Elman Network and Echo State Network. The results indicated that the feedforward and Elman networks obtained better performances analyzed by the determination coefficient (R2). Thus, the first model obtained the best topology with 37x60x70x1, trainlm as training algorithm, tansig as activation functions for the two intermediate layers and output layer activated by the purelin function with R2 of 88.69 percent. The Elman network model presented its best topology with 37x45x55x1, trainlm as training algorithm, tansig as activation functions for the two intermediate layers and output layer activated by purelin function with R2 of 83.63 percent. It was concluded that the analyzed networks can be used in predictive control of fault in industrial processes and can be applied in chemical plants in the future.

銅價在近年來不斷的創下歷史新高，由於台灣蓬勃的電子、半導體、工具機產業皆需要銅，因此銅進口量位居全球第五(ICSG,2009)，使得台灣企業的生產成本受國際銅價的波動影響甚鉅，全球有70%的銅價是按照英國倫敦金屬交易所(London Metal Exchange, LME)的牌價進行貿易，因此本研究欲建置預測模式以預測銅價未來趨勢。 　　本研究之資料來源為2003年1月2日至2011年7月14日的LME三月期銅價，並依文獻探討選取LME的銅庫存、三月期鋁價、三月期鉛價、三月期鎳價、三月期鋅價、三月期錫價，以及金價、銀價、石油價格、美國生產者物價指數、美國消費者物價指數、聯邦資金利率作為影響因素的分析資料。時間序列分析、類神經網路已被廣泛的用於預測股市及期貨，本研究先藉由向量自我迴歸模型篩選出有影響力的變數，同時建置GARCH時間序列預測模型與具有遞迴的Elman類神經網路預測模型，再整合兩者建置GARCH-Elman類神經網路預測模型。 本研究之向量自我迴歸模型顯示銅價與金、鋁、銅庫存前第1期；自身前第2期；鎳、錫前第3期；鋅前第4期的變動有負向的影響；受到石油前第2期的變動有正向的影響，這其中以銅的自我解釋變異最高，銅庫存最低，推測其影響已有效率地反映到銅價上。也驗證預測模型必須考量總體經濟變數，且變數先經向量自我迴歸模型的篩選能因減少雜訊而提升類神經網路的預測能力。依此建置的GARCH模型有33.81%的累積報酬率、Elman類神經網路38.11%、整合兩者的GARCH-Elman類神經網路56.46%，皆優於實際銅價指數的累積報酬率。對銅有需求的企業者，能更為準確的預測漲跌趨勢，依此判斷如何跟原物料供應商簽訂合約的價格與期間，使其免於價格趨勢的誤判而提高生產成本，並提出五點建議供未來研究者參考。 / The recent copper price in London Metal Exchange (LME) has breaking the historical high. Taiwan’s booming electronics, semiconductor and machine tool industry causing copper import volume ranked fifth in the world (ICSG, 2009). Because of 70% of copper worldwide trade in accordance with the price of the London Metal Exchange, this study using time series and neural networks to build the LME copper price forecast model. 　　This study considering copper, copper stocks, aluminum, lead, nickel, zinc, tin, gold, silver, oil ,federal funds rate, CPI and PPI during the period of 2003/1/2 to 2011/7/14. Time series model and neural networks have been widely used for forecasting the stock market and futures. In this study, using Vector Autoregressive (VAR) model screened influential variables, building GARCH model and Elman neural network to forecast the LME copper price; and further, integrating this two models to build GARCH-Elman neural network prediction model. 　　This study’s VAR models show that the copper has negative effect with gold, aluminum, copper stocks, nickel, tin, zinc and itself. And has positive impact with oil prices. The highest of explained variance is copper. Copper stocks are lowest, speculating that its impact has been efficiently reflecting on the price of copper. Verifying the prediction model must consider the macroeconomics variables. Using VAR model screened influential variables can reduce noise to enhance the predictive ability of the neural network. This study’s GARCH model has 33.81% of the cumulative rate of return, Elman neural network has 38.11% and the GARCH-Elman neural network has 56.46%. All of them are better than the actual price of copper.

Att estimera slutpriset av en bostadsförsäljning är en komplex uppgift där mäklartexter som beskriver bostäder är en vital del av försäljningen. Denna rapport undersöker om det går att använda mäklartexter för att generera mer träffsäkra estimeringar med maskininlärningsmodeller. Två olika maskininlärningsmodeller implementerades som resultat av en litteraturstudie och utvärderades mot Boolis existerande OLS-modell. De implementerade modellerna är OLS-BERT och Elman regressionsnätverk. OLS-BERT visade en generell förbättring jämfört med Boolis OLS-modell, i synnerhet av F-statistik där mätvärdet sjönk med 99,8 procent. P-värdet i T-statistik för “vista” (utsikten) har sjunkit från 37,7 till 1 procent. Elman regressionsnätverket sänkte Boolis OLS-modells MAPE från 58,5 till 6,62 procent. Modellerna utvärderades med åtta olika mått varav de för studiens viktigaste är MAPE, MAE, F-statistik och T-statistik. Genom att bryta ut attribut ur mäklartexter kan modellen förklara signifikansen hos indata, samt få något mer träffsäkra estimeringar av slutpriset av en bostadsförsäljning. Resultaten visar att det är en intressant metod som med fördel kan vidare utforskas. / Estimating the price of home sales is a complex task, where broker texts describing the housing is a vital part of the sales. This study explore the possibility to use broker texts to generate more accurate estimations using machine learning models. Two different machine learning models were implemented as a result of a literature study and evaluated against Booli’s existing OLS-model. The implemented machine learning models are OLS-BERT and an Elman regression network. OLS-BERT showed a general improvement compared to Booli’s OLS-model, in particular the F-statistic were 99.8 percent lower than Booli’s OLS-model. The p-value in T-statistic for “vista” was 37.7 percent with Booli’s OLS-model and 1 percent with OLS-BERT. The Elman regression network lowered the MAPE of Booli’s OLS-model from 58.5 to 6.62 percent. All models were evaluated using eight different measures, of which the most important for this study is MAPE, MAE, F-statistic, and T-statistic. The conclusion is that by mining attributes from broker texts the models can explain the significance of the input and generate somewhat more accurate estimations of the home sales price of sale. The results show that this is an interesting method that should be further explored.

A solar and diesel-wind hybrid power control systems is proposed in the thesis. The system consists of solar power, wind power, diesel-engine, a static synchronous compensator and an intelligent power controller. MATLAB/Simulink was used to build the dynamic model and simulate the solar and diesel-wind hybrid power system. A static synchronous compensator was used to supply reactive power and regulate the voltage of the hybrid system. To achieve a fast and stable response for the real power control, an intelligent controller was proposed, which consists of the Radial Basis Function Network (RBFN) and the Elman Neural Network (ENN) for maximum power point tracking (MPPT). The pitch angle control of wind power uses ENN controller, and the output is fed to the wind turbine to achieve the MPPT. The solar system uses RBFN, and the output signal is used to control the DC / DC boost converters to achieve the MPPT.

本研究考慮重要且對台股大盤指數走勢有連動影響的因素，主要納入對台股有領頭作用的美國三大股市，那斯達克(NASDAQ)指數、道瓊工業(Dow Jones)指數、標準普爾500(S&P500)指數;其他對台股緊密連動效果的國際股票市場，香港恆生指數、上海證券綜合指數、深圳證券綜合指數、日經225指數;以及納入左右國際經濟表現的國際原油價格走勢，美國西德州原油、中東杜拜原油和歐洲北海布蘭特原油;在宏觀經濟因素方面則考量失業率、消費者物價指數、匯率、無風險利率、美國製造業重要指標的存貨/銷貨比率、影響貨幣數量甚鉅的M1B;在技術分析方面則納入多種重要的指標，心理線 (PSY) 指標、相對強弱(RSI) 指標、威廉(WMS%R) 指標、未成熟隨機(RSV) 指標、K-D隨機指標、移動平均線(MA)、乖離率(BIAS)、包寧傑%b和包寧傑帶狀寬度(BandWidth%);所有考量因素共計35項，因為納入重要因子比較多，所以完備性較高。 本研究先採用的贏者全拿(Winner-Take-All) 競爭學習策略的自組織映射網路(Self-Organizing Feature Maps, SOM)，藉由將相似資料歸屬到已身的神經元萃取出關聯分類且以計算距離來衡量神經元的離散特徵，對於探索大量且高維度的非線性複雜特徵俱有優良的因素相依性投射效果，將有利於提高預測模式精準度。在線性擬合部分則結合倒傳遞(Back-Propagation, BP)、Elman反饋式和徑向基底函數類網路(Radial-Basis-Function Network, RBF)模式為指數預測輸出，並對台股加權指數隔日收盤指數進行預測和評量。而在傳統的Elman反饋式網路只在隱藏層存在反饋機制，本研究則在輸入層和隱藏層皆建立反饋機制，將儲存在輸入層和隱藏層的過去時間資訊回饋給網路未來參考。在徑向基底函數網路方面，一般選取中心聚類點採用隨機選取方式，若能有效降低中心點個數，可降低網路複雜度，本研究導入垂直最小平方法以求取誤差最小的方式強化非監督式學習選取中心點的能力，以達到網路快速收斂，提昇網路學習品質。 研究資料為台股指數交易收盤價，日期自2001/1/2，至2011/10/31共2676筆資料。訓練資料自2001/1/2至2009/12/31，共2223筆;實證測試資料自2010/1/4至2011/10/31，計453個日數。主要評估指標採用平均相對誤差(AMRE)和平均絕對誤差 (AAE)。在考慮因子較多的狀況下，實證結果顯示，在先透過SOM進行因子聚類分析之後，預測因子被分成四個組別，分別再透過BP、Elman recurrent和RBF方法進行線性擬合，平均表現方面，以RBF模式下的四個群組因子表現最佳，其中RBF模式之下的群組4，其AMRE可達到0.63%，最差的AMRE則是群組1，約為1.05%;而Elman recurrent模式下的四組群組因子之ARME則介於1.01%和1.47%之間;其中預測效果表現最差則是BP模式的預測結果。顯示RBF具有絕佳的股價預測能力。最後，在未來研究建議可以運用本文獻所探討之其他數種類神經網路模式進行股價預測。 / In this study, we considering the impact factors for TWSE index tendency, mainly aimed at the three major American stock markets, NASDAQ index, Dow Jones index, S&P 500, which leading the Taiwan stock market trend; the other international stock markets, such as the Hong Kong Hang-Seng Index, Shanghai Stock Exchange Composite Index, Shenzhen Stock Exchange Composite Index, NIKKEI 225 index, which have close relationship with Taiwan stock market; we also adopt the international oil price trend, such as the West Texas Intermediate Crude Oil in American, the Dubai crude oil in Middle Eastern, North Sea Brent crude oil in European, which affects international economic performance widely; On the side of macroeconomic factors, we considering the Unemployed rate, Consumer Price Index, exchange rate, riskless rate, the Inventory to Sales ratio which it is important index of American manufacturing industry, and the M1b factor which did greatly affect to currency amounts; In the part of Technical Analysis index, we adopt several important indices, such as the Psychology Line Index (PSY), Relative Strength Index (RSI), the Wechsler Memory Scale—Revised Index (WMS%R), Row Stochastic Value Index (RSV), K-D Stochastics Index, Moving Average Line (MA), BIAS, Bollinger %b (%b), Bollinger Band Width (Band Width%);All factors total of 35 which we have considered the important factor is numerous, so the integrity is high. In this study, at first we adopt the Self-Organizing Feature Maps Network which based on the Winner-Take-All competition learning strategy, Similar information by the attribution to the body of the neuron has been extracted related categories and to calculate the distance to measure the discrete characteristics of neurons, it has excellent projection effect by exploring large and complex high-dimensional non-linear characteristics for all the dependency factors , would help to improve the accuracy of prediction models, would be able to help to improve the accuracy of prediction models. The part of the curve fitting combine with the back-propagation (Back-Propagation, BP), Elman recurrent model and radial basis function network (Radial-Basis-Function Network, RBF) model for the index prediction outputs, forecast and assessment the next close price of Taiwan stocks weighted index. In the traditional Elman recurrent network exists only one feedback mechanism in the hidden layer, in this study in the input and hidden layer feedback mechanisms are established, the previous information will be stored in the input and hidden layer and will be back to the network for future reference. In the radial basis function network, the general method is to selecting cluster center points by random selection, if we have the effectively way to reduce the number of the center points, which can reduces network complexity, in this study introduce the Orthogonal Least Squares method in order to obtain the smallest way to strengthen unsupervised learning center points selecting ability, in order to achieve convergence of the network fast, and improve network learning quality. Research data for the Trading close price of Taiwan Stock Index, the date since January 2, 2001 until September 30, 2011, total data number of 2656. since January 2, 2001 to December 31, 2009 a total number of 2223 trading close price as training data; empirical testing data, from January 4, 2010 to September 30, 2011, a total number of 433. The primary evaluation criteria adopt the Average Mean Relative Error (AMRE) and the Average Absolute Error (AAE). In the condition for consider more factors, the empirical results show that, by first through SOM for factor clustering analysis, the prediction factors were divided into four categories and then through BP, Elman recurrent and RBF methods for curve fitting, at the average performance , the four group factors of the RBF models get the best performance, the group 4 of the RBF model, the AMRE can reach 0.63%, the worst AMRE is group 1, about 1.05%; and the four groups of Elman recurrent model of ARME is between 1.01% and 1.47%; the worst prediction model is BP method. RBF has shown excellent predictive ability for stocks index. Finally, the proposal can be used in future studies of the literatures that we have explore several other methods of neural network model for stock trend forecasting.

An artificial Neural Network (ANN) system has been developed that can analyse aircraft flight data to provide a reconstruction of the aerodynamic loads experienced by the aircraft during flight, including manoeuvre, buffet and distributed loading. For this research data was taken from the International Follow-On Structural Test Project (IFOSTP) F/A-18 fatigue test conducted by the Royal Australian Air Force and Canadian Forces. This fatigue test involved the simultaneous application of both manouevre and buffet loads using airbag actuators and shakers. The applied loads were representative of the actual loads experienced by an FA/18 during flight tests. Following an evaluation of different ANN types an Ellman network with three linear layers was selected. The Elman back-propagation network was tested with various parameters and structures. The network was trained using the MATLAB 'traingdx' function with is a gradient descent with momentum and adaptive learning rate back-propagation algorithm. The ANN was able to provide a good approximation of the actual manoeuvre or buffet loads at the location where the training loads data were recorded even for input values which differ from the training input values. In further tests the ability to estimate distributed loading at locations not included in the training data was also demonstrated. The ANN was then modified to incorporate various methods for the calculation and prediction of output error and reliability Used in combination and in appropriate circumstances, the addition of these capabilities significantly increase the reliability, accuracy and therefore usefulness of the ANN system's ability to estimate aircraft loading.To demonstrate the ANN system's usefulness as a fatigue monitoring tool it was combined with a formulae for crack growth analysis. Results inficate the ANN system may be a useful fatigue monitoring tool enabling real time monitoring of aircraft critical components using existing strain gauge sensors.