人工神经网络在注塑机控制中的应用

以注塑制品的张力系数为研究对象，将人工神经网络中的反向传播算法(简称BP模型)应用于张力系数的预测。并将其结果与用多元回归分析方法计算的结果进行比较，以分析人工神经网络在注塑机控制应用中的优越性及其应用前景。     

人工神经网络在林业中的应用研究与展望

4简要介绍了人工神经网络的基本概念，论述和分析了人工神经网络在森林规划与管理、水土流失的等级及林地等方面的分类，森林预测模型、林业工程的过程控制等方面的应用，并进一步研究了人工神经网络在精确林业及树木特征数据库的建立等方面的应用。     

人工神经网络在机械设计中的应用

本文通过对机械设计专家系统和人工神经网络的讨论，研究了人工神经网络和专家系统技术在机械设计智能系统中的综合应用问题，并提出了人工神经网络在机械设计中的总体应用方案，为进一步研究打下了基础。     

基于样本数据的线切割加工工艺模型研究

分析了线切割加工工艺的特点,介绍了人工神经网络的定义及要点,研究了BP网络在线切割工艺中的应用,利用人工神经网络技术,建立网络模型,实现与Internet的集成.     

基于人工神经网络的浮游植物密度预测模型研究

基于人工神经网络在赤潮预测中应用的基础，利用遗传算法对其网络结构进行优化之后，用于预测浮游植物密度有良好的成效，对赤潮预警系统的研究有实际的应用价值。文章介绍了浮游植物密度预测的人工神经网络模型和经过优化后的人工神经网络预测计算。     

工程类几何模型分类与聚类的现状与趋势

总结了目前工程类几何模型分类与聚类问题的研究进展。首先分析了基于k最近邻方法和基于支持向量机的模型分类技术,回顾了传统聚类技术和人工神经网络在模型聚类方面的应用现状;然后探讨了聚类和降维技术在模型分类与聚类的过程可视化和结果可视化方面的应用。最后,通过对已有研究成果的比较分析,预测了工程类几何模型的分类与聚类的研究方向。     

应用人工神经网络监测切削颤振

文章提出了应用人工神经网络模型进行颤振预报的方法，结合实测振动数据，找出了足以反映振动状态变化的特征量，并给邮了人工神经网络模型的程序实现方法。最后对人工神经网络方法的性能和改进作了探讨，指出了人工神经网络是一种较为可靠预报模型。     

应用人工神经网路监测切削颤振

文章提出了应用人工神经网络模型进行颤振预报的方法。结合实测振动数据，找出了足以反映振动状态变化的特征量，并给出了人工神经网络模型的程序实现方法。最后对人工神经网络方法的性能和改进作了探讨，指出人工神经网络是一种较为可靠的预报模型。     

借助热重分析数据评估润滑油使用性能的人工神经网络方法

研究了将人工神经网络方法应用于热重分析数据评估润滑油某些使用性能的可行性，给出了组建最优人工神经网络的规则。应用５－３－３结构ＢＰ模型人工神经网络评估润滑油的使用性能，取得了满意的结果，表明将人工神经网络应用于润滑油使用性能的评估是可行的。     

人工神经网络在密封材料性能预测中的应用

文章简要地介绍了人工神经网络的特点,并与传统方法作了简单的比较,总结了人工神经网络在材料性能和配方研究方面的应用,讨论了人工神经网络用于无石棉密封材料配方优化和制品性能预测的优点及其应用状况.     

用人工神经网络预测切削力

借助ANN模型的函数逼近功能对切削力进行快速预测进行了研究。通过基于LM算法的人工神经网络模型对加工参数及刀具凡何参数对切削力的影响进行了预测，其结果与金属切削的实验数据有很好的吻合性。其研究成果可为加工过程中刀具及加工参数的选择提供依据。     

A comparative study of the ANN and RSM modeling approaches for predicting burr size in drilling

This paper describes the comparison of the burr size predictive models based on artificial neural networks (ANN) and response surface methodology (RSM). The models were developed based on three-level full factorial design of experiments conducted on AISI 316L stainless steel work material with cutting speed, feed, and point angle as the process parameters. The ANN predictive models of burr height and burr thickness were developed using a multilayer feed forward neural network, trained using an error back propagation learning algorithm (EBPA), which is based on the generalized delta rule. The performance of the developed ANN models were compared with the second-order RSM mathematical models of burr height and thickness. The comparison clearly indicates that the ANN models provide more accurate prediction compared to the RSM models. The details of experimentation, model development, testing, and performance comparison are presented in the paper.     

Prediction of quality responses in micro-EDM process using an adaptive neuro-fuzzy inference system (ANFIS) model

In the present trend of technological development, micro-machining is gaining popularity in the miniaturization of industrial products. In this work, a hybrid process of micro-wire electrical discharge grinding and micro-electrical discharge machining (EDM) is used in order to minimize inaccuracies due to clamping and damage during transfer of electrodes. The adaptive neuro-fuzzy inference system (ANFIS) and back propagation (BP)-based artificial neural network (ANN) models have been developed for the prediction of multiple quality responses in micro-EDM operations. Feed rate, capacitance, gap voltage, and threshold values were taken as the input parameters and metal removal rate, surface roughness and tool wear ratio as the output parameters. The results obtained from the ANFIS and the BP-based ANN models were compared with observed values. It is found that the predicted values of the responses are in good agreement with the experimental values and it is also observed that the ANFIS model outperforms BP-based ANN.     

Hybrid FE/ANN and LPR approach for the inverse identification of material parameters from cutting tests

Accuracy of numerical models based in finite elements (FE), extensively used for simulation of cutting processes, depends strongly on the identification of proper material parameters. Experimental identification of the constitutive law parameters for simulation of cutting processes involves unsolved problems such as the complex testing techniques or the difficulty to reproduce the stress triaxiality state during cutting. This work proposes a methodology for the inverse identification of the material parameters from cutting test. Two hybrid approaches are compared. One of them based on FE and artificial neural networks (ANN). The other one based on FE and local polynomial regression (LPR). Firstly, a FE model is validated with experimental data. Then, ANN and LPR are trained with FE simulations. Finally, the estimated ANN and LPR models are used for the inverse identification of material parameters. This identification is solved as an optimization problem. The FE/LPR approach shows good performance, outperforming the FE/ANN approach.     

Development of RSM- and ANN-based models to predict and analyze the effects of process parameters of laser-hardened commercially pure titanium on heat input and tensile strength

Laser transformation hardening (LTH) is an innovative and advanced laser surface modification technique as compared to conventional transformation hardening processes and has been employed in aerospace, marine, chemical applications, heat exchangers, cryogenic vessels, components for chemical processing and desalination equipment, condenser tubing, airframe skin, and nonstructural components which introduces the advantageous residual stresses into the surface, improving the mechanical properties like wear, resistance to corrosion, tensile strength, and fatigue strength. In the present study, LTH of commercially pure titanium, nearer to ASTM grade 3 of chemical composition was investigated using continuous wave 2 kW, Nd: YAG laser. The effect of laser process variables such as laser power, scanning speed, and focused position was investigated using response surface methodology (RSM) and artificial neural network (ANN) keeping argon gas flow rate of 10 lpm as fixed input parameter. This paper describes the comparison of the heat input (HI) and ultimate tensile strength (σ) (simply called as tensile strength) predictive models based on ANN and RSM. The paper also presents the effect of laser process variables on the HI and ultimate σ. The research work also emphasizes on the effect of HI on σ. The experiments were conducted based on a three-factor, three-level Box–Behnken surface statistical design. Quadratic polynomial equations were developed for proper process parametric study for its optimal performance characteristics. The experimental results under optimum conditions were compared with the simulated values obtained from the RSM and ANN model. Adequacy of the developed models was tested by analysis of variance technique. A multilayer feed-forward neural network with a Levenberg–Marquardt back-propagation algorithm was adopted to develop the relationships between the laser hardening process parameters, HI, and ultimate σ. The performance of the developed ANN models were compared with the second-order RSM mathematical models of HI and σ. There was good agreement between the experimental and simulated values of RSM and ANN. The comparison clearly indicates that the ANN models provide more accurate prediction compared to the RSM models. It has been found that there is a trend of increased tensile strength with the decrease of hardening heat input and a trend of increased tensile strength with the increase of hardening cooling rate. As heat input decreases, there will be a faster cooling rate. Considering the effect of HI on ultimate σ, it was found that the lower the heat input, the faster cooling rate. The details of experimentation, model development, testing, validation of models, effect of laser process variables on heat input and ultimate σ, effect of HI on σ, and performance comparison of RSM and ANN models are presented in the paper. The results of Box–Behnken design of RSM and ANN models also indicate that the proposed models predict the responses adequately within the limits of input parameters being used. It is suggested that regression equations can be used to find optimum conditions for HI and σ of laser-hardened commercially pure titanium material.     

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.     

Prediction of tool wear using regression and ANN models in end-milling operation

Tool wear prediction plays an important role in industry for higher productivity and product quality. Flank wear of cutting tools is often selected as the tool life criterion as it determines the diametric accuracy of machining, its stability and reliability. This paper focuses on two different models, namely, regression mathematical and artificial neural network (ANN) models for predicting tool wear. In the present work, flank wear is taken as the response (output) variable measured during milling, while cutting speed, feed and depth of cut are taken as input parameters. The Design of Experiments (DOE) technique is developed for three factors at five levels to conduct experiments. Experiments have been conducted for measuring tool wear based on the DOE technique in a universal milling machine on AISI 1020 steel using a carbide cutter. The experimental values are used in Six Sigma software for finding the coefficients to develop the regression model. The experimentally measured values are also used to train the feed forward back propagation artificial neural network (ANN) for prediction of tool wear. Predicted values of response by both models, i.e. regression and ANN are compared with the experimental values. The predictive neural network model was found to be capable of better predictions of tool flank wear within the trained range.     

Artificial neural network and regression models for performance prediction of abrasive waterjet in rock cutting

An experimental study is carried out for modeling the rock cutting performance of abrasive waterjet. Kerf angle (KA) is considered as a performance criteria and modeled using artificial neural network (ANN) and regression analysis based on operating variables. Three operating variables, including traverse speed, standoff distance, and abrasive mass flow rate, are studied for obtaining different results for the KA. Data belonging to the trials are used for construction of ANN and regression models. The developed models are then tested using a test data set which is not utilized during construction of models. Additionally, the regression model is validated using various statistical approaches. The results of regression analysis are also used to determine the significant operating variables affecting the KA. Furthermore, the performances of derived models are compared for showing the accuracy levels in prediction of the KA. As a result, it is concluded that both ANN and regression models can give adequate prediction for the KA with an acceptable accuracy level. The compared results reveal also that the corresponding ANN model is more reliable than the regression model. On the other hand, the standoff distance and traverse speed are statistically determined as dominant operating variables on the KA, respectively.     

熔融挤出堆积成形MEM─250系统

在分析熔融堆积成形过程的基础上，提出精密控制组织材料的序是保证成形质量的关键，分析影响序的主要因素，提出相应的控制措施。介绍MEM─250系统的组成、特点及应用等情况。     

A smart force platform using artificial neural networks

The human body may interact with structures and these interactions are developed through the application of contact forces, for instance when walking. The aim of this paper is to propose a new methodology using Artificial Neural Network (ANN) for calibrating a force platform in order to reduce the uncertainties in the values of estimated vertical Ground Reaction Force and the positioning of the applied force in the human gait. Force platforms have been used to evaluate the pattern of human applied forces and to fit models for the interaction between pedestrians and structures. Linear relation assumptions between input and output are common in traditional Least Mean Square methods used in calibration. Some discrepancies due to nonlinearities in the experimental setup (looseness, wear, support settlements, electromagnetic noise, etc.) may harm the overall fitting. Literature has shown that nonlinear models, like ANN, can better handle this. During the calibration, the input data to the ANN were the reference voltages applied to the Wheatstone bridge, while the output data were the values of the standard weights applied in the force platform in defined sites. Supervised training based on k-fold cross validation was used to check the ANN generalization. The use of ANN shows significant improvements for the measured variables, leading to better results for predicted values with low uncertainty when compared to the results of a simple traditional calibration using Least Mean Squares.