• Advances in environmental research
• /
• 제5권3호
• /
• pp.153-167
• /
• 2016

We applied multilayer perceptron (MLP) and radial basis function (RBF) neural network in upstream and downstream water quality stations of the Karaj Reservoir in Iran. For both neural networks, inputs were pH, turbidity, temperature, chlorophyll-a, biochemical oxygen demand (BOD) and nitrate, and the output was dissolved oxygen (DO). We used an MLP neural network with two hidden layers, for upstream station 15 and 33 neurons in the first and second layers respectively, and for the downstream station, 16 and 21 neurons in the first and second hidden layer were used which had minimum amount of errors. For learning process 6-fold cross validation were applied to avoid over fitting. The best results acquired from RBF model, in which the mean bias error (MBE) and root mean squared error (RMSE) were 0.063 and 0.10 for the upstream station. The MBE and RSME were 0.0126 and 0.099 for the downstream station. The coefficient of determination ($R^2$) between the observed data and the predicted data for upstream and downstream stations in the MLP was 0.801 and 0.904, respectively, and in the RBF network were 0.962 and 0.97, respectively. The MLP neural network had acceptable results; however, the results of RBF network were more accurate. A sensitivity analysis for the MLP neural network indicated that temperature was the first parameter, pH the second and nitrate was the last factor affecting the prediction of DO concentrations. The results proved the workability and accuracy of the RBF model in the prediction of the DO.

• Advances in aircraft and spacecraft science
• /
• 제6권6호
• /
• pp.529-550
• /
• 2019

The effect of multiple process parameters on a set of continuous response variables is, especially in experimental designs, difficult and intricate to determine. Due to the complexity in aeroacoustic and vibroacoustic studies, the often-performed simple one-factor-at-a-time method turns out to be the least effective approach. In contrast, the statistical Design of Experiments is a technique used with the objective to maximize the obtained information while keeping the experimental effort at a minimum. The presented work aims at giving insights on Design of Experiments applied to aeroacoustic and vibroacoustic problems while comparing different experimental designs and approximation models. For this purpose, an experimental rig of a ducted low-pressure fan is developed that allows gathering data of both, aerodynamic and aeroacoustic nature while analysing three independent process parameters. The experimental designs used to sample the design space are a Central Composite design and a Box-Behnken design, both used to model a response surface regression, and Latin Hypercube sampling to model an Artificial Neural network. The results indicate that Latin Hypercube sampling extracts information that is more diverse and, in combination with an Artificial Neural network, outperforms the quadratic response surface regressions. It is shown that the Latin Hypercube sampling, initially developed for computer-aided experiments, can also be used as an experimental design. To further increase the benefit of the presented approach, spectral information of every experimental test point is extracted and Artificial Neural networks are chosen for modelling the spectral information since they show to be the most universal approximators.

• 대한의용생체공학회:학술대회논문집
• /
• 대한의용생체공학회 1995년도 추계학술대회
• /
• pp.109-112
• /
• 1995

This paper is a study on the detection of the significant point in ECG signal. ECG signal consists of two components; one is high frequency component to be detected and the other is low frequency component to be removed. AR model is appropriate for modelling and removing the low frequency component. AR model coefficients are updated by artificial neural network algorithm. We can remove the background noise(low frequency) by passing through the AR filter. The remaining signals which include high frequency noise are sent to the matched filter to pass only the signal which we want to extract. The template used in matched filter is updated adaptively.

• Membrane and Water Treatment
• /
• 제9권2호
• /
• pp.115-121
• /
• 2018

Artificial neural network (ANN) simulation is used to predict the dynamic change of permeate flux during wheat starch industry wastewater microfiltration with and without static turbulence promoter. The experimental program spans range of a sedimentation times from 2 to 4 h, for feed flow rates 50 to 150 L/h, at transmembrane pressures covering the range of $1{\times}10^5$ to $3{\times}10^5Pa$. ANN predictions of the wastewater microfiltration are compared with experimental results obtained using two different set of microfiltration experiments, with and without static turbulence promoter. The effects of the training algorithm, neural network architectures on the ANN performance are discussed. For the most of the cases considered, the ANN proved to be an adequate interpolation tool, where an excellent prediction was obtained using automated Bayesian regularization as training algorithm. The optimal ANN architecture was determined as 4-10-1 with hyperbolic tangent sigmoid transfer function transfer function for hidden and output layers. The error distributions of data revealed that experimental results are in very good agreement with computed ones with only 2% data points had absolute relative error greater than 20% for the microfiltration without static turbulence promoter whereas for the microfiltration with static turbulence promoter it was 1%. The contribution of filtration time variable to flux values provided by ANNs was determined in an important level at the range of 52-66% due to increased membrane fouling by the time. In the case of microfiltration with static turbulence promoter, relative importance of transmembrane pressure and feed flow rate increased for about 30%.

• 한국항해학회지
• /
• 제23권3호
• /
• pp.35-44
• /
• 1999

This research aims to seek active control of ball-beam position stability by resorting to neural networks whose layers are given bias weights. The controller consists of an LQR (linear quadratic regulator) controller and a neural networks controller in parallel. The latter is used to improve the responses of the established LQR control system, especially when controlling the system with nonlinear factors or modelling errors. For the learning of this control system, the feedback-error learning algorithm is utilized here. While the neural networks controller learns repetitive trajectories on line, feedback errors are back-propagated through neural networks. Convergence is made when the neural networks controller reversely learns and controls the plant. The goals of teaming are to expand the working range of the adaptive control system and to bridge errors owing to nonlinearity by adjusting parameters against the external disturbances and change of the nonlinear plant. The motion equation of the ball-beam system is derived from Newton's law. As the system is strongly nonlinear, lots of researchers have depended on classical systems to control it. Its applications of position control are seen in planes, ships, automobiles and so on. However, the research based on artificial control is quite recent. The current paper compares and analyzes simulation results by way of the LQR controller and the neural network controller in order to prove the efficiency of the neural networks control algorithm against any nonlinear system.

• 전기학회논문지P
• /
• 제58권3호
• /
• pp.249-256
• /
• 2009

The widely used control theory based design of PI family controllers fails to perform satisfactorily under parameter variation, nonlinear or load disturbance. In high performance applications, it is useful to automatically extract the complex relation that represent the drive behaviour. The use of learning through example algorithms can be a powerful tool for automatic modelling variable speed drives. They can automatically extract a functional relationship representative of the drive behavior. These methods present some advantages over the classical ones since they do not rely on the precise knowledge of mathematical models and parameters. The paper proposes high performance speed and current control of synchronous reluctance motor(SynRM) drive using adaptive learning mechanism-fuzzy neural network (ALM-FNN) and fuzzy logic control (FLC) controller. The proposed controller is developed to ensure accurate speed and current control of SynRM drive under system disturbances and estimation of speed using artificial neural network(ANN) controller. Also, this paper proposes the analysis results to verify the effectiveness of the ALM-FNN, FLC and ANN controller.

• 한국조명전기설비학회:학술대회논문집
• /
• 한국조명전기설비학회 2009년도 춘계학술대회 논문집
• /
• pp.416-419
• /
• 2009

The widely used control theory based design of PI family controllers fails to perform satisfactorily under-parameter variation, nonlinear or load disturbance. In high performance applications, it is useful to automatically extract the complex relation that represent the drive behaviour. The use of loaming through example algorithms can be a powerful tool for automatic modelling variable speed drives. They can automatically extract a functional relationship representative of the drive behavior. These methods present some advantages over the classical ones since they do not rely on the precise knowledge of mathematical models and parameters. The paper proposes high performance speed and current control of synchronous reluctance motor(SynRM) drive using adaptive loaming mechanism-fuzzy neural network (ALM-FNN) and fuzzy logic control(FLC) controller. The proposed controller is developed to ensure accurate speed and current control of SynRM drive under system disturbances and estimation of speed using artificial neural network(ANN) controller. Also, this paper proposes the analysis results to verify the effectiveness of the ALM-FNN and ANN controller.

• 한국지능시스템학회논문지
• /
• 제14권4호
• /
• pp.390-395
• /
• 2004

반도체 공정은 많은 단위 공정으로 이루어진 복잡하고 동적인 공정이다. 그 중 에칭공정은 반도체 생산에서 중요한 공정중 하나이다. 본 논문에서는 데이터 마이닝과 지식 획득을 통한 의사지원시스템으로 생산성과 수율을 높일 수 있는 시스템을 구성하고자 하였다. 제안된 방법은 퍼지 논리와 신경망으로 구성되는데, 신경망으로 에칭공정의 품질을 나타내는 품질에 대한 결과를 예측하고, 예측된 결과를 퍼지 추론 시스템으로 분류하는 과정으로 수행된다. 퍼지 논리에 사용된 규칙은 전문가의 지식에 기반 하여 도출되거나 데이터로부터 도출된다. 본 시스템을 통해 공정의 최적 조건을 찾아 효율을 높이는 것이 본 연구의 주요 목표이다.

• Journal of Electrical Engineering and Technology
• /
• 제2권2호
• /
• pp.241-248
• /
• 2007

The accuracy of all the schemes that belong to vector controlled induction machine drives is strongly affected by parameter variations. The aim of this paper is to examine iron losses and magnetic saturation effect in sensorless vector control of induction machines. At first, an approach to induction machine modelling and vector control scheme, which account for both iron loss and saturation, is presented. Then, a model reference adaptive system (MRAS) based speed estimator is developed. The speed estimation is modified in such a way that iron losses and the variation in the saturation level are compensated. Thus by substituting an artificial neural network flux estimator into the MRAS speed estimator. Experimental results are presented to verify the effectiveness of the proposed approach.

• 환경영향평가
• /
• 제20권5호
• /
• pp.705-716
• /
• 2011

For the efficient discovery of knowledge and information from the observed systems, data mining techniques can be an useful tool for the prediction of water quality at intake station in rivers. Deterioration of water quality can be caused at intake station in dry season due to insufficient flow. This demands additional outflow from dam since some extent of deterioration can be attenuated by dam reservoir operation to control outflow considering predicted water quality. A seasonal occurrence of high ammonia nitrogen ($NH_3$-N) concentrations has hampered chemical treatment processes of a water plant in Geum river. Monthly flow allocation from upstream dam is important for downstream $NH_3$-N control. In this study, prediction models of water quality based on multiple regression (MR), artificial neural network and data mining methods were developed to understand water quality variation and to support dam operations through providing predicted $NH_3$-N concentrations at intake station. The models were calibrated with eight years of monthly data and verified with another two years of independent data. In those models, the $NH_3$-N concentration for next time step is dependent on dam outflow, river water quality such as alkalinity, temperature, and $NH_3$-N of previous time step. The model performances are compared and evaluated by error analysis and statistical characteristics like correlation and determination coefficients between the observed and the predicted water quality. It is expected that these data mining techniques can present more efficient data-driven tools in modelling stage and it is found that those models can be applied well to predict water quality in stream river systems.