基于MIC和MPA-KELM的脱硫出口SO2浓度预测

建立脱硫出口SO₂浓度预测模型是实现脱硫系统经济运行的基础。针对这一问题,提出了基于最大信息系数(MIC)的变量选择方法和基于海洋捕食算法(MPA)优化核极限学习机(KELM)的脱硫出口SO₂浓度预测模型。首先,采用机理分析法筛选影响出口SO₂浓度的变量,提出循环浆液综合流量表达方法,便于描述浆液循环泵组合的影响特性;在此基础上,通过基于最大信息系数的变量选择算法确定模型输入变量;运用MPA对KELM的正则系数C和核参数S进行寻优,建立MPA-KELM的出口SO₂浓度预测模型;最后,利用电厂运行数据进行仿真实验。实验结果表明,所建立出口SO₂浓度预测模型的均方误差、平均绝对百分比误差分别为1.236 66 mg/m³和4.987 6%,预测精度高,能够为脱硫系统出口SO₂的现场优化控制提供技术支持。     

Optimization of hybrid laser–TIG welding of 316LN stainless steel using genetic algorithm

Determination of optimum hybrid laser–TIG welding process variables for achieving the maximum depth of penetration (DOP) in type 316LN stainless steel has been carried out using a genetic algorithm (GA). Nd:YAG pulsed laser and the TIG heat source were coupled at the weld pool to carry out hybrid welding. Design of experiments approach was used to generate the experimental design matrix. Bead-on-plate welds were carried out based on the design matrix. The input variables considered were laser power, pulse frequency, pulse duration, and TIG current. The response variable considered was the DOP. Multiple-regression model was developed correlating the process variables with the DOP using the generated data. The regression model was used for evaluating the objective function in GA. GA-based model was developed and it produced a set of solutions. Tournament and roulette wheel selection methods were used during the execution of GA. It was found that both the selection methods identified similar welding process parameters for achieving the maximum DOP. Excellent agreement was observed between the target DOP and the DOP values obtained in the validation experiments during hybrid laser–TIG welding.     

Design of artificial neural networks using a genetic algorithm to predict collection efficiency in venturi scrubbers

In this study, a new approach for the auto-design of neural networks, based on a genetic algorithm (GA), has been used to predict collection efficiency in venturi scrubbers. The experimental input data, including particle diameter, throat gas velocity, liquid to gas flow rate ratio, throat hydraulic diameter, pressure drop across the venturi scrubber and collection efficiency as an output, have been used to create a GA-artificial neural network (ANN) model. The testing results from the model are in good agreement with the experimental data. Comparison of the results of the GA optimized ANN model with the results from the trial-and-error calibrated ANN model indicates that the GA-ANN model is more efficient. Finally, the effects of operating parameters such as liquid to gas flow rate ratio, throat gas velocity, and particle diameter on collection efficiency were determined.     

Optimization of chemical composition for TC11 titanium alloy based on artificial neural network and genetic algorithm

It is quite difficult for materials to develop the quantitative model of chemical elements and mechanical properties, because the relationship between them presents the multivariable and non-linear. In this work, the combined approach of artificial neural network (ANN) and genetic algorithm (GA) was employed to synthesize the optimum chemical composition for satisfying mechanical properties for TC11 titanium alloy based on the large amount of experimental data. The chemical elements (Al, Mo, Zr, Si, Fe, C, O, N and H) were chosen as input parameters of the ANN model, and the output parameters are mechanical properties, including ultimate tensile strength, yield strength, elongation and reduction of area. The fitness function for GA was obtained from trained ANN model. It is found that the percentage errors between experimental and predicted are all within 5%, which suggested that the ANN model has excellent generalization capability. The results strongly indicated that the proposed optimization model offers an optimal chemical composition for TC11 titanium alloy, which implies it is a novel and effective approach for optimizing materials chemical composition.     

Development of machine learning‐based real time scheduling systems: using ensemble based on wrapper feature selection approach

There are two items that significantly enhance the generalisation ability (i.e. classification accuracy) of machine learning‐based classifiers: feature selection (including parameter optimisation) and an ensemble of the classifiers. Accordingly, the objective in this study is to develop an ensemble of classifiers based on a genetic algorithm (GA) wrapper feature selection approach for real time scheduling (RTS). The proposed approach can better enhance the generalisation ability of the RTS knowledge base (i.e. classifier) in comparison with three classical machine learning‐based classifier RTS systems, including the GA‐based wrapper feature selection mechanism, in terms of the prediction accuracy of 10‐fold cross validation as measured according to all the performance criteria. The proposed ensemble classifier RTS also provides better system performance than the three machine learning‐based RTS systems, including the GA‐based wrapper feature selection mechanism and heuristic dispatching rules, under all the performance criteria, over a long period in a flexible manufacturing system (FMS) case study.     

基于改进粒子群算法的径向基人工神经网络淀粉基发泡复合材料性能预测

以乙烯-醋酸乙烯酯(EVA)和淀粉质量比、甘油质量分数和NaHCO_3质量分数为输入,以拉伸强度和回弹率为输出,建立基于种群熵多样性评估和收敛、发散策略的粒子群改进算法的径向基人工神经网络(RBF ANN)的淀粉基发泡复合材料性能预测模型。结果表明,该模型的预测效果较好,预测均方差和相关系数分别为0.0160和0.9890。预测发现,淀粉基发泡复合材料的拉伸强度随甘油含量的增加而缓慢降低,随NaHCO_3含量的增加先减少后增加;回弹率随甘油含量的增加而递增,随NaHCO_3含量的增加而先增加后减少。     

An Early Stopping-Based Artificial Neural Network Model for Atmospheric Corrosion Prediction of Carbon Steel

The optimization of network topologies to retain the generalization ability by deciding when to stop overtraining an artificial neural network (ANN) is an existing vital challenge in ANN prediction works. The larger the dataset the ANN is trained with, the better generalization the prediction can give. In this paper, a large dataset of atmospheric corrosion data of carbon steel compiled from several resources is used to train and test a multilayer backpropagation ANN model as well as two conventional corrosion prediction models (linear and Klinesmith models). Unlike previous related works, a grid searchbased hyperparameter tuning is performed to develop multiple hyperparameter combinations (network topologies) to train multiple ANNs with mini-batch stochastic gradient descent optimization algorithm to facilitate the training of a large dataset. After that, one selection strategy for the optimal hyperparameter combination is applied by an early stopping method to guarantee the generalization ability of the optimal network model. The correlation coefficients (R) of the ANN model can explain about 80% (more than 75%) of the variance of atmospheric corrosion of carbon steel, and the root mean square errors (RMSE) of three models show that the ANN model gives a better performance than the other two models with acceptable generalization. The influence of input parameters on the output is highlighted by using the fuzzy curve analysis method. The result reveals that TOW, Cl- and SO2 are the most important atmospheric chemical variables, which have a well-known nonlinear relationship with atmospheric corrosion.     

Computational intelligence modelling of pharmaceutical tabletting processes using bio-inspired optimization algorithms

In pharmaceutical development, it is very useful to exploit the knowledge of the causal relationship between product quality and critical material attributes (CMA) in developing new formulations and products, and optimizing manufacturing processes. With the big data captured in the pharmaceutical industry, computational intelligence (CI) models could potentially be used to identify critical quality attributes (CQA), CMA and critical process parameters (CPP). The objective of this study was to develop computational intelligence models for pharmaceutical tabletting processes, for which bio-inspired feature selection algorithms were developed and implemented for optimisation while artificial neural network (ANN) was employed to predict the tablet characteristics such as porosity and tensile strength. Various pharmaceutical excipients (MCC PH 101, MCC PH 102, MCC DG, Mannitol Pearlitol 200SD, Lactose, and binary mixtures) were considered. Granules were also produced with dry granulation using roll compaction. The feed powders and granules were then compressed at various compression pressures to produce tablets with different porosities, and the corresponding tensile strengths were measured. For the CI modelling, the efficiency of seven bio-inspired optimization algorithms were explored: grey wolf optimization (GWO), bat optimization (BAT), cuckoo search (CS), flower pollination algorithm (FPA), genetic algorithm (GA), particle swarm optimization (PSO), and social spider optimization (SSO). Two-thirds of the experimental dataset was randomly chosen as the training set, and the remaining was used to validate the model prediction. The model efficiency was evaluated in terms of the average reduction (representing the fraction of selected input variables) and the mean square error (MSE). It was found that the CI models can well predict the tablet characteristics (i.e. porosity and tensile strength). It was also shown that the GWO algorithm was the most accurate in predicting porosity. While the most accurate prediction for the tensile strength was achieved using the SSO algorithm. In terms of the average reduction, the GA algorithm resulted in the highest reduction of inputs (i.e. 60%) for predicting both the porosity and the tensile strength.     

Comparison Between Genetic Algorithm and Genetic Programming Approach for Modeling the Stress Distribution

This article compares genetic algorithm (GA) and genetic programming (GP) for system modeling in metal forming. As an example, the radial stress distribution in a cold-formed specimen (steel X6Cr13) was predicted by GA and GP. First, cylindrical workpieces were forward extruded and analyzed by the visioplasticity method. After each extrusion, the values of independent variables (radial position of measured stress node, axial position of measured stress node, and coefficient of friction) were collected. These variables influence the value of the dependent variable, radial stress. On the basis of training data, different prediction models for radial stress distribution were developed independently by GA and GP. The obtained models were tested with the testing data. The research has shown that both approaches are suitable for system modeling. However, if the relations between input and output variables are complex, the models developed by the GP approach are much more accurate.     

纳滤膜对水中酚类物质的截留率及其定量结构关系的研究

分别选用NF90,NF270和NF-纳滤膜测定了21种酚类物质的截留率,研究了纳滤膜对酚类物质的截留机理.试验结果表明,酚类物质的截留率受到酚类取代基位置和膜特性的影响:截留率从大到小的次序为邻位取代>间位取代>对位取代;孔径越小、荷电量越大的纳滤膜截留率越大.此外,在MATLAB6.5环境下,利用遗传算法(GA)结合偏最小二乘回归(PLS)和人工神经网络(ANN)建立了酚类化合物的截留率与其定量结构关系的模型,两种方法建立的模型的相关系数可分别达到0.886 9和0.993 7.进而结合定量构效关系模型,从有机物分子结构参数的角度分析了影响酚类物质截留率的因素,并对几种物质的截留率进行了有效的预测.从预测结果来看,GA-ANN模型的预测精度要好于GA-PLS模型.     

Optimization of Welding Process Parameters for 9Cr-1Mo Steel Using RSM and GA

Optimization of A-TIG welding process parameters for 9Cr-1Mo steel has been carried out using response surface methodology (RSM) and genetic algorithm (GA). RSM has been used to obtain the design matrix for generating data on the influence of process parameters on the response variables. A second-order response surface model was developed for predicting the response for the set of given input variables. Then, numerical and graphical optimization was performed using RSM to obtain the target depth of penetration (DOP) and heat-affected zone (HAZ) width using desirability approach. Multiple regression models were developed based on the generated data, and then the models were used in GA to determine the optimum process parameters for achieving the target DOP and HAZ width. GA-based models employed two different selection processes. Both the RSM- and GA-based models suggested a number of solutions in terms of process parameters, and the identified solutions were validated by experiments. GA-based model employing tournament selection has been found to be a more accurate method for determining the optimum A-TIG welding process parameters.     

Application of Artificial Intelligent Tools to Modeling of Glucosamine Preparation from Exoskeleton of Shrimp

The objective of this study was to forecast and optimize the glucosamine production yield from chitin (obtained from Persian Gulf shrimp) by means of genetic algorithm (GA), particle swarm optimization (PSO), and artificial neural networks (ANNs) as tools of artificial intelligence methods. Three factors (acid concentration, acid solution to chitin ratio, and reaction time) were used as the input parameters of the models investigated. According to the obtained results, the production yield of glucosamine hydrochloride depends linearly on acid concentration, acid solution to solid ratio, and time and also the cross-product of acid concentration and time and the cross-product of solids to acid solution ratio and time. The production yield significantly increased with an increase of acid concentration, acid solution ratio, and reaction time. The production yield is inversely related to the cross-product of acid concentration and time. It means that at high acid concentrations, the longer reaction times give lower production yields. The results revealed that the average percent error (PE) for prediction of production yield by GA, PSO, and ANN are 6.84, 7.11, and 5.49%, respectively. Considering the low PE, it might be concluded that these models have a good predictive power in the studied range of variables and they have the ability of generalization to unknown cases.     

基于集成学习的FRP加固混凝土梁抗弯承载力预测研究

为解决当前纤维增强复合材料(FRP)加固钢筋混凝土梁抗弯承载力预测中模型不统一、计算繁琐、精度有限等问题,建立了统一化的抗弯承载力预测模型。根据既有文献收集外贴式、端锚式和嵌入式3种FRP典型加固方式加固钢筋混凝土梁试验数据,确定影响加固梁承载力的关键因素,通过XGBoost(极限梯度提升树)算法训练回归各影响因素与加固后梁抗弯承载力间的非线性映射关系,得到统一化的FRP加固钢筋混凝土梁抗弯承载力预测模型。随后在测试样本集上对该模型的预测精度进行了验证,与基于支持向量回归(SVR)和人工神经网络(ANN)两种代表性机器学习算法得到的预测模型进行了横向对比,并分析了不同加固方式下的预测精度。研究结果表明：该文得到的基于XGBoost的抗弯承载力预测模型拟合优度R²=0.9417,可见整体精度较高,有良好的性能;相比基于传统机器学习算法SVR和ANN建立的预测模型,基于集成学习算法XGBoost的拟合优度分别提升了8.00%及6.70%,均方根误差减少了33.94%和30.72%,平均绝对误差减少了32.38%和30.51%,表明基于XGBoost的模型精度更高,远优...     

Artificial neural network application of modeling failure rate for Boeing 737 tires

This paper presents an application of artificial neural network (ANN) technique for conducting the reliability analysis of Boeing 737 tires. For this purpose, an ANN model utilizing the feed‐forward back‐propagation algorithm as a learning rule is developed. The inputs to the neural network are the flight operational time and the number of landings as independent variables and the output is the failure rate of the tires. Two years of data are used for failure rate prediction model and validation. Model validation, which reflects the suitability of the model for future predictions, is performed by comparing the predictions of the model with that of Weibull regression model. The results show that the failure rate predicted by the ANN is closer in agreement with the actual data than the failure rate predicted by the Weibull model. The present work also identifies some of the common tire failures and presents representative results based on the established model for the most frequently occurring tire failure. Copyright © 2010 John Wiley & Sons, Ltd.     

Simultaneous distribution and sizing optimization for stiffeners with an improved genetic algorithm with two-level approximation

This article presents an improved genetic algorithm with two-level approximation (GATA) to optimize the distribution and size of stiffeners simultaneously. A novel optimization model of stiffeners, including two kinds of design variables, is established. The first level approximation problem transforms the original implicit problem to an explicit problem which involves the topology and size variables. Then, a genetic algorithm (GA) addresses the mixed variables. The individuals in the GA are coded by topology variables, and when calculating an individual’s fitness, the second level approximation problem is embedded to optimize the size variables. Considering the stiffeners’ optimization, several aspects of the initial GATA are updated, including the relationship between two kinds of variables, the weight and its sensitivity calculation and the GA strategy, to optimize the stiffeners’ size and distribution simultaneously. Numerical examples show that the improved GATA is effective in optimizing the stiffened shells’ topology and size variables simultaneously.     

Optimal design approach for heating irregular-shaped objects in three-dimensional radiant furnaces using a hybrid genetic algorithm–artificial neural network method

In this article, a new hybrid method based on the combination of the genetic algorithm (GA) and artificial neural network (ANN) is developed to optimize the design of three-dimensional (3-D) radiant furnaces. A 3-D irregular shape design body (DB) heated inside a 3-D radiant furnace is considered as a case study. The uniform thermal conditions on the DB surfaces are obtained by minimizing an objective function. An ANN is developed to predict the objective function value which is trained through the data produced by applying the Monte Carlo method. The trained ANN is used in conjunction with the GA to find the optimal design variables. The results show that the computational time using the GA-ANN approach is significantly less than that of the conventional method. It is concluded that the integration of the ANN with GA is an efficient technique for optimization of the radiant furnaces.