A neural network approach for prediction of discharge in straight compound open channel flow

Most natural rivers and streams consist of two stage channels known as main channel and flood plains. Accurate prediction of discharge in compound open channels is extremely important from river engineering point of view. It helps the practitioners to provide essential information regarding flood mitigation, construction of hydraulic structures and prediction of sediment load so as to plan for effective preventive measures. Discharge determination models such as the single channel method (SCM), the divided channel method (DCM), the coherence method (COHM) and the exchange discharge method (EDM) are widely used; however, they are insufficient to predict discharge accurately. Therefore, an attempt has been made in this work to predict the total discharge in compound channels with an artificial neural network (ANN) and compare with the above models. The mean absolute percentage error with artificial neural networks is found to be consistently low as compared to other models.     

Predicting drill wear using an artificial neural network

The present work deals with drill wear monitoring using an artificial neural network. A back propagation neural network (BPNN) has been used to predict the flank wear of high-speed steel (HSS) drill bits for drilling holes on copper work-piece. Experiments have been carried out over a wide range of cutting conditions and the effect of various process parameter like feedrate, spindle speed, and drill diameter on thrust force and torque has been studied. The data thus obtained from the experiments have been used to train a BPNN for wear prediction. The performance of the trained neural network has been tested with the experimental data, and has been found to be satisfactory.     

Predicting the impact of vegetations in open channels with different distributaries’ operations on water surface profile using artificial neural networks

Most of the open water irrigation channels in Egypt suffer from the infestation of aquatic weeds, especially the submerged ones that cause numerous hydraulic problems for the open channels themselves and their water distributaries such as increasing water losses, obstructing water flow, and reducing channels’ water distribution efficiencies. Accurate simulation and prediction of flow behavior in such channels is very essential for water distribution decision makers. Artificial neural networks (ANN) have proven to be very successful in the simulation of several physical phenomena, in general, and in the water research field in particular. Therefore, the current study aims towards introducing the utilization of ANN in simulating the impact of vegetation in main open channel, which supplies water to different distributaries, on the water surface profile in this main channel. Specifically, the study, presented in the current paper utilizes ANN technique for the development of various models to simulate the impact of different submerged weeds’densities, different flow discharges, and different distributaries operation scheduling on the water surface profile in an experimental main open channel that supplies water to different distributaries. In the investigated experiment, the submerged weeds were simulated as branched flexible elements. The investigated experiment was considered as an example for implementing the same methodology and technique in a real open channel system. The results showed that the ANN technique is very successful in simulating the flow behavior of the pre-mentioned open channel experiment with the existence of the submerged weeds. In addition, the developed ANN models were capable of predicting the open channel flow behavior in all the submerged weeds’cases that were considered in the ANN development process     

Factors affecting algal blooms in a man-made lake and prediction using an artificial neural network

It is difficult to predict when, where, and how long algal blooms will occur in a water body. The objectives of this study were to determine the factors affecting algal bloom and predict chlorophyll-a (Chl-a) levels in the reservoir formed by damming a river using an artificial neural network (ANN). The automatic water quality monitoring data [water temperature, pH, dissolved oxygen (DO), electric conductivity, total organic carbon (TOC), Chl-a, total nitrogen (T-N), and total phosphorus (T-P)], weather data (precipitation, temperature, insolation, and duration of sunshine) and hydrologic data (water level, discharges, and inflows) in the man-made Lake Juam during 2008–2010 were used to develop a model to predict Chl-a as an indirect measure of the abundance of algae. The ANN was trained using the collected data during 2008–2010 and the accuracy of the model was verified using the data collected in 2011. It was found that Chl-a concentration, TOC, pH and atmospheric and water temperatures were the most important parameters in predicting Chl-a concentrations. The Chl-a prediction was most influenced by the parameters showing the algal activities such as Chl-a, TOC and pH. Due to the relatively long hydraulic retention time of ∼131 days, the inflow and outflow did not affect the prediction much. Likewise, atmospheric and water temperatures did not respond to the change of the Chl-a concentration due to the lake’s relatively slow response to the temperature. Most importantly, T-N and T-P were not the major factors in predicting Chl-a levels at Lake Juam. The ANN trained with the time series data successfully predicted the Chl-a concentration and provided information regarding the principal factors affecting algal bloom at Lake Juam.     

Classification decision model for a Hong Kong clothing manufacturing plant locations using an artificial neural network

Clothing manufacturers’ direct investment and joint ventures in developing regions have seen to grow rapidly in the past few decades. Manufacturers face difficulties during the decision-making process in the selection of a plant location due to vague and subjective considerations. Selecting a plant location relies mostly on subjective intuition and assessment as variables to be considered in the decision making process. But these variables cannot always be represented in terms of objective value, such as country risk and community facilities. Though several optimization techniques have been developed to assist decision makers in searching for the optimal sites, it is difficult to rank the sites which display a small difference of scores. Classification is thus more reasonable and realistic. This paper investigates two recent types of classification techniques, namely unsupervised and supervised artificial neural networks, on the site selection problem of clothing manufacturing plants. The limitations of adaptive resonance theory in unsupervised artificial neural networks will be demonstrated. A comparison of the performance of the three types of supervised artificial neural networks – including back propagation, learning vector quantization and probabilistic neural network – is used and the proposed classification decision model will be presented. The experimental results indicate that the supervised artificial neural network is a proven and effective classifier in which a probabilistic neural network performs better than the others in this site selection problem.     

Friction materials development by using DOE/RSM and artificial neural network

A group of non-asbestos organic based friction materials containing 16 ingredients were investigated in this work using the techniques of design of experiment (2^k DOE), response surface methodology (RSM), and artificial neural network (ANN). The ingredients effects on three friction characteristics including 1st fading rate, 2nd fading rate, and speed sensitivity were studied by 2^k DOE. Five ingredients of phenolic resin, synthetic graphite, potassium titanate, mineral fiber, and calcium silicate were found to be statistically significant for these responses and should be studied further. In the meantime, an artificial neural network with Elman recurrent configuration was trained and tested using the data generated from dynamometer tests in 2^k DOE experiments. Concerning the confounding of two-ingredient interaction effects and main effects, response surface methodology was employed to optimize the friction material formulation. The well trained and tested Elman artificial neural network was then used to predict the friction characteristics of the trials generated by RSM. Based on the ANN prediction and RSM analysis, an optimization of material formulation was obtained and validated by experiments.     

测站水位的改进型神经网络模型预测

神经网络作为一种非机理性模型,在非线性问题方面被广泛应用,由于水位与上下游位置存在明显的非线性关系,因此选用神经网络并结合现有的工具软件进行测站水位的预测。在全面分析神经网络特点的基础上,通过某径流上有测点的水位资料建立了下游测点的水位预测神经网络模型,详细给出了模型建立的具体步骤。工程实例表明该模型具有预报精度高等优点,可以为类似工程提供参考。     

Prediction of calcium concentration in human blood serum using an artificial neural network

A predictive method, based on artificial neural network (ANN) has been developed to study absorbance and pH effects on the equilibrium of blood serum. This strategy has been used to analyze serum samples and to predict the calcium concentration in blood serum. A dedicated data acquisition system is designed and fabricated using a LPC2106 microcontroller with light emitting diode (LED) as source and photodiode as sensor to measure absorbance and to calculate the calcium concentration. A multilayer neural network with back propagation (BP) training algorithm is used to simulate different concentration of calcium (Ca²⁺) as a function of absorbance and pH, to correlate and predict calcium concentration. The computed calcium concentration by neural network is quite satisfactory with correlations R² = 0.998 and 0.995, standard errors of 0.0127 and 0.0122 in validation and testing stages respectively. Statistical analysis are carried out to check the accuracy and precision of the proposed ANN model and validation of results produce a relative error of about 3%. These results suggest that ANN can be efficiently applied and is in good agreement with values obtained with the current clinical spectrophotometric methods. Hence, ANN can be used as a complementary tool for studying metal ion complexion, with special attention to the blood serum analysis.     

Application of hybrid neural network in discharge coefficient prediction of triangular labyrinth weir

Discharge coefficient (C_d) is an important parameter of triangular labyrinth weir. It is of great significance to predict the discharge coefficient accurately. In this research, in order to more accurately predict the C_d, in view of the traditional BP neural network is easy to fall into the local minimum in the training process, genetic algorithm (GA) and particle swarm optimization (PSO) are employed to optimize the traditional BP neural network's initial weights and thresholds. Nonlinear regression analysis (NLR) is also added to compare with these intelligent methods and four discharge coefficient prediction models are built, namely the NLR, the BPNN, the GA-BPNN and the PSO-BPNN. After the completion of the model construction, in order to objectively evaluate the performance of these models, the prediction results of these models are compared with the experiment results, and the determination coefficient (R²), the mean absolute error (MAE) and the root mean square error (RMSE) are introduced as the performance indicators to quantify the model performance. The results show that the accuracy and stability of the NLR are much worse than that of the intelligent models. The prediction results of the GA-BPNN and the PSO-BPNN are quite accurate with a higher decision coefficient than the BPNN. Moreover, the MAEs and the RMSEs of the GA-BPNN and the PSO-BPNN were significantly reduced by 25 and 40% compared with BPNN, respectively, and the maximum prediction errors were 4.4% and 2.6%, severally. Meanwhile, the width of error uncertainty band of GA-BPNN and PSO-BPNN is narrower than BPNN. By comparing GA-BPNN and PSO-BPNN with the discharge coefficient prediction models of triangular labyrinth weir in previous literatures, it is found that the mean absolute percentage error (MAPE) values of GA-BPNN and PSO-BPNN are 1.504% and 1.225% respectively, which are lower than other existing models. At the same time, the other performance indexes are better than most existing models, indicating that the genetic algorithm and PSO algorithm are more effective than the traditional BP algorithm in adjusting BP neural network parameters, easier to find the global optimal value, and improve the prediction accuracy and applicability of the model.     

A hybridized artificial neural network and imperialist competitive algorithm optimization approach for prediction of soil compaction in soil bin facility

We were inspired to furnish information concerning the promising applicability of a hybrid approach involving artificial neural networks (ANNs), with manifold network functions, and a meta-heuristic optimization algorithm for prediction of soil compaction indices. The employed network functions were the prevailed feed-forward network and the novel cascade-forward network algorithms to accommodate multivariate inputs of wheel load, tire inflation pressure, number of passage, slippage, and velocity each at three different levels for estimating the study objectives of soil compaction (i.e. penetration resistance and soil sinkage). The experimentations were carried out in a soil bin facility utilizing a single wheel-tester. Each ANN trials was developed merely and then by merging with the recently introduced evolutionary optimization technique of imperialist competitive algorithm (ICA). The results were compared on the basis of a modified performance function (MSEREG) and coefficient of determination (R²). Our results elucidated that hybrid ICA–ANN further succeeded to denote lower modeling error amongst which, cascade-forward network optimized by ICA managed to yield the highest quality solutions.     

Flow velocity estimation using a fish-shaped lateral line probe with product-moment correlation features and a neural network

Environmental studies on fish require measurements of highly turbulent flows in both the laboratory and in the field. A fish-shaped bioinspired flow measuring device is applied in conjunction with data processing workflow which leverages the interactions between the body and the surrounding flow field for velocity estimation in turbulent flows. Our objective is to develop a robust velocity estimation methodology relevant for studies of fish behavior using a bioinspired fish-shaped artificial lateral line probe (LLP). We show that the device is capable of covering the range of flow velocities from 0 to 1.5 m/s. Three different sets of experiments performed in a closed flow tunnel, a model vertical slot fishway and laboratory open channel flume were collected and combined to provide time-averaged flow velocity and LLP measurements under fully turbulent flow conditions. Based on the experimental results, a signal processing workflow using Pearson product-moment correlation coefficient (PCC) features in conjunction with an artificial neural network (ANN) is presented. Using PCC features provides a simple data fusion methodology exploiting the use of the LLP's as a simultaneous collocated sensing array. In this work we show that (1) the PCC-ANN workflow provides the first LLP velocity estimator without repeated calibration across the full span of 0–1.5 m/s, (2) using all pressure sensors results in the best performance with R²=0.917, but requires a PCC feature matrix of 55 dimensions and (3) a stepwise reduction of the PCC feature matrix allows for the use of as few as 11 dimensions, and results in R²=0.911, indicating that a modest reduction in LLP velocity estimation performance can be gained by a large reduction in dimensionality. A surprising finding was that after stepwise reduction, the best performing sensor pair combinations were not the expected pitot-like anteroposterior couples spanning from nose to body. Instead, it was found that optimal velocity estimation using the LLP exploited a network of sensor pairs. It is shown that the LLP can be implemented similar to an ADV for highly turbulent flows over the range of 0–1.5 m/s, and in addition provides body-centric pressure distributions which may aid in the interpretation of fish hydrodynamic preferences in future environmental studies.     

压电振动系统的神经网络建模和预测研究

本文应用MATALB／XPC实时仿真工具测量了贴有压电元件的复合材料薄壁结构的振动响应。并对其进行神经网络的离线建模和预测。比较了几种网络的优缺点。选择了引进外部反馈的前向BP网络作为非线性系统建模的方法，有望推广用于智能结构的健康监测和振动主动控制。     

Prediction of contraction scour using ANN and GA

This paper emphasizes on the application of soft computing tools such as artificial neural network (ANN) and genetic algorithm (GA) in the prediction of scour depth within channel contractions. The experimental data of earlier investigators are used in developing the models and ANN and GA Toolboxes of MATLAB software are utilized for the purpose. The multilayered perceptron (MLP) neural networks with feed-forward back-propagation training algorithms were designed to predict the scour depth. The mean squared error and correlation coefficient are used to check the performance of networks. It is found that the ANN architecture 4-16-1 having trained with Levenberg-Marquardt ‘trainlm’ function had best performance having mean squared error of 0.001 and correlation coefficient of 0.998. In addition, the suitability of ‘trainlm’ method over other training methods is also discussed. The scour depths predicted by ANN model were compared with those computed by the two analytical models (with and without sidewall correction for contracted zone) and an empirical model proposed by Dey and Raikar [1]. In addition, heuristic search technique called genetic algorithm is used to develop the predictor for maximum scour depth within channel contraction. The population size for GA was 500 members with total generations of 1000, crossover fraction of 0.8 and Gaussian operator for mutation. It is promising to observe that the GA model predicts the maximum scour depth equally well as that of empirical model of Dey and Raikar [1]. Hence, both ANN and GA models can be satisfactorily used to predict the scour depth within channel contractions.     

基于改进型BP神经网络的氢原子钟钟差预测

原子钟的钟差预测是原子钟时标计算和原子钟驾驭的关键环节,良好的钟差预测可明显提高原子钟时标和原子钟驾驭的精度。为进一步提高氢原子钟的钟差预测精度,本文提出了一种改进型的BP神经网络算法,并用中国计量科学研究院守时实验室氢原子钟组的实际数据进行了验证。验证结果表明,本文提出的改进型BP神经网络钟差预测算法与目前采用的线性回归钟差预测算法、SVM钟差预测算法相比,显著地提高了氢原子钟钟差预测精度。该钟差预测算法的提出对提高原子钟时标和驾驭精度有很好的推动作用。     

供暖热网神经网络预测模型的研究

将人工神经网络应用于供热网实时预报,建立起可用于热网供暖预报的外时延反馈型BP网络模型,及内时延反馈型Elman网络。且利用实际热网数据对所建立的网络进行训练和检验,结果表明两种预报模型均具有较好的动态跟踪能力和预报特性。而Elman网络在节点结构上比外时延反馈型BP网络更简单,在确定网络节点结构上更快捷,更具有实际推广和应用价值。     

Estimation of biosurfactant yield produced by Klebseilla sp. FKOD36 bacteria using artificial neural network approach

Environmental and medium parameters estimation is an essential step in Bioprocess engineering. In the present study, artificial neural network (ANN) was employed in estimation of biosurfactants yield from bacterial strain Klebseilla sp. FKOD36, surface tension reduction as well emulsification index. The data obtained from experimental design were used in modelling and optimization of ANN method. Temperature, pH value, incubation period, carbon, nitrogen and hydrocarbon sources were used as input of ANN model independently in the prediction of biosurfactants yield, surface tension reduction and emulsification index. Using the optimized values of critical input elements of ANN, the experimental values of biosurfactant yield, emulsification index and surface tension showed close agreement with the model estimate. The most efficient ANN model assessment was 0.030 g/l for actual value 0.038 g/l of biosurfactant yield, 31.67% for actual value 31.68% of emulsification index, and 21.6 dyne/cm for actual value 21.5 dyne/cm of surface tension respectively.     

基于过程神经网络的航空发动机性能参数预测

针对传统方法难以对性能参数进行有效预测的问题,提出一种基于过程神经网络的性能参数预测方法。为解决反向传播学习算法收敛速度慢、易陷于局部极小点等问题,开发了一种基于正交基函数展开的Leven-berg-Marquardt学习算法。为提高过程神经网络的泛化能力,从提高训练样本的质量和规模入手,研究了实际测量数据的预处理方法,并提出一种基于样条函数拟合和相空间重构理论的训练样本集构造方法。最后,将该方法用于某型航空发动机性能参数的预测,获得了满意的结果。     

Optimum surface roughness in end milling Inconel 718 by coupling neural network model and genetic algorithm

In this study, optimum cutting parameters of Inconel 718 are determined to enable minimum surface roughness under the constraints of roughness and material removal rate. In doing this, advantages of statistical experimental design technique, experimental measurements, artificial neural network and genetic optimization method are exploited in an integrated manner. Cutting experiments are designed based on statistical three-level full factorial experimental design technique. A predictive model for surface roughness is created using a feed forward artificial neural network exploiting experimental data. Neural network model and analytical definition of material removal rate are employed in the construction of optimization problem. The optimization problem was solved by an effective genetic algorithm for variety of constraint limits. Additional experiments have been conducted to compare optimum values and their corresponding roughness and material removal rate values predicted from the genetic algorithm. Generally a good correlation is observed between the predicted optimum and the experimental measurements. The neural network model coupled with genetic algorithm can be effectively utilized to find the best or optimum cutting parameter values for a specific cutting condition in end milling Inconel 718.     

基于人工神经网络预测磨合磨损的最佳分形维数

基于BP人工神经网络的L-M算法,建立了磨合磨损的分形参数预测模型。将该模型用于销一盘磨合磨损试验。对最佳分形维数进行了准确预测。该模型收敛速度快、误差小,输出结果与实验结果有极好的吻合性。