Prediction of fracture parameters of concrete by Artificial Neural Networks

Modelling of material behaviour generally involves the development of a mathematical model derived from observations and experimental data. An alternative way discussed in this paper is Artificial Neural Network (ANN)-based modelling which is a subfield of artificial intelligence. The main benefit in using an ANN approach is that the network is built directly from experimental data using the self-organising capabilities of the ANN. In this paper the Two-Parameter Model (TPM) in the fracture of cementitious materials is modelled with a back-propagation ANN. The results of an ANN-based TPM look viable and very promising.     

评估知识生产绩效的BP神经网络模型

把知识生产划分为批量型知识生产和订单型知识生产两类,绘制了它们的流程,并系统地比较了它们与传统制造行业供应链的差异。针对知识生产的流程更加复杂多变,供应链更新更快且对象更难于量化的特点,提出了批量型和订单型两类知识生产的绩效评估指标体系,在此基础上通过BP神经网络模型,得出了它们的绩优度,为知识生产绩效的评估提供了参考和依据。     

递归神经网络的模糊随机学习算法

提出了一种新的基于模糊逻辑的Alopex学习算法（FLA）。FLA算法利用模糊逻辑推理实时获得适应于学习过程的适当的算法修正值，克服了Alopex算法中修正值固定不变的缺点，使得随机学习过程在速度、精度和稳定性之间获得平衡。将该算法应用于神经网络的训练，可以无需神经网络的梯度信息和结构信息，因此可以用于具有各种结构特性的递归神经网络对动态系统的学习过程。实验结果表明了FLA算法的有效性。     

Artificial Neural Networks: Comparison of Two Programs for Modeling a Process of Nanoparticle Preparation

ABSTRACT

Artificial Neural Networks (ANNs) were used to predict nanoparticle size and micropore surface area of polylactic acid nanoparticles, prepared by a double emulsion method. Different batches were prepared while varying polymer and surfactant concentration, as well as homogenization pressure. Two commercial ANNs programs were evaluated: Neuroshell® Predictor, a black-box software adopting both neural and genetic strategies, and Neurosolutions®, allowing a step-by-step building of the network. Results were compared to those obtained by statistical method. Predictions from ANNs were more accurate than those calculated using non-linear regression. Neuroshell® Predictor allowed quantification of the relative importance of the inputs. Furthermore, by varying the network topology and parameters using Neurosolutions®, it was possible to obtain output values which were closer to experimental values. Therefore, ANNs represent a promising tool for the analysis of processes involving preparation of polymeric carriers and for prediction of their physical properties.     

Mechanical Property Prediction of Commercially Pure Titanium Welds with Artificial Neural Network

Factors that affect weld mechanical properties of commercially pure titanium have been investigated using artificial neural networks. Input data were obtained from mechanical testing of single-pass, autogenous welds, and neural network models were used to predict the ultimate tensile strength, yield strength, elongation, reduction of area, Vickers hardness and Rockwell B hardness. The results show that both oxygen and nitrogen have the most significant effects on the strength while hydrogen has the least effect over the range investigated. Predictions of the mechanical properties are shown and agree well with those obtained using the 'oxygen equivalent' (OE) equations.     

Sentence Similarity Measurement with Convolutional Neural Networks Using Semantic and Syntactic Features

Calculating the semantic similarity of two sentences is an extremely challenging problem. We propose a solution based on convolutional neural networks (CNN) using semantic and syntactic features of sentences. The similarity score between two sentences is computed as follows. First, given a sentence, two matrices are constructed accordingly, which are called the syntax model input matrix and the semantic model input matrix; one records some syntax features, and the other records some semantic features. By experimenting with different arrangements of representing the syntactic and semantic features of the sentences in the matrices, we adopt the most effective way of constructing the matrices. Second, these two matrices are given to two neural networks, which are called the sentence model and the semantic model, respectively. The convolution process of the neural networks of the two models is carried out in multiple perspectives. The outputs of the two models are combined as a vector, which is the representation of the sentence. Third, given the representation vectors of two sentences, the similarity score of these representations is computed by a layer in the CNN. Experiment results show that our algorithm (SSCNN) surpasses the performance MPCPP, which noticeably the best recent work of using CNN for sentence similarity computation. Comparing with MPCNN, the convolution computation in SSCNN is considerably simpler. Based on the results of this work, we suggest that by further utilization of semantic and syntactic features, the performance of sentence similarity measurements has considerable potentials to be improved in the future.     

CFD modelling of moisture evaporation in an industrial dispersed system

A methodology is presented here for the mathematical modelling of moisture evaporation in a dispersed system in an industrial tower. An empirical model using characteristic drying curves was applied to calculate moisture evaporation from a droplet and particle. A simple method was developed to calculate the agglomeration process of the dispersed phase in the drying towers, using transient functions between the initial and final particle size distributions, at the nozzle and in the final product. The developed model and simulation results were validated on the basis of industrial spray tower experiments. High instability of the airflow due to the geometry of the dryer and the construction of the air inlets was observed. The general methodology applied within this CFD model is universal, and can be applied to the scaling-up of installations for dewatering in dispersed systems in order to determine configurations of feeding systems and control the product quality and safety of the process.     

Neural network constitutive modelling for non‐linear characterization of anisotropic materials

This paper presents a new technique of neural network constitutive modelling for non‐linear characterization of anisotropic materials. The proposed technique, based on a recently developed energy‐based characterization framework, derives the variations of the external work applied to and the strain energy induced in a specimen. The error between the variations of the energies is subsequently applied to correct the neural network properties by using a modified backpropagation algorithm. Unlike the conventional techniques for neural network constitutive modelling, the proposed technique develops models by quantifying the deformation of the specimen on a continuum basis. This allows the neural network constitutive models to be constructed from a single load test of one specimen. Numerical examples first examine the effect of specimen geometries and loading conditions. The effect of noise in the experimental measurements is subsequently investigated while having the applicability for non‐linear constitutive behaviour shown thereafter. The application for anisotropic materials is finally demonstrated by modelling a unidirectional lamina based on the measurements of a biaxial load test on a balanced laminate. Copyright © 2010 John Wiley & Sons, Ltd.     

Integrated inventory models with fuzzy annual demand and fuzzy production rate in a supply chain

The successful implementation of Just-in-time (JIT) production in today's supply chain environment requires a new spirit of cooperation between the buyer and the vendor. An integrated inventory model with such a consideration is based on the total cost optimization under a common stock policy and business formula. However, the supposition of known annual demand in most related literature may not be realistic. This paper proposes the inclusion of fuzzy annual demand and/or the production rate, and then employs the signed distance, a ranking method for fuzzy numbers, to find the estimate of the common total cost in the fuzzy sense, and subsequently derives the corresponding optimal buyer's quantity and the integer number of lots in which the items are delivered from the vendor to the purchaser. Numerical examples are provided to illustrate the results of proposed models.     

模糊神经网络在精密三轴转台机械台体故障诊断中的应用研究

针对三轴转台机械台体故障,提出了一种基于模糊神经网络的故障诊断方法,给出了模糊神经网络的结构和学习算法,并阐述了模糊故障诊断原理和故障判别方法将振动信号和电流噪声信号结构用于机械台体的故障诊断,测试结果表明,该方法是有效的。     

Production control in a complex production system using approximate dynamic programming

Development of an efficient production and inventory control policy for a production system with multiple working stations, intermediate components and end products is difficult. In particular, uncertain demand and large changeover times at the work stations cause significant problems. In this paper, we consider an assembly line for dishwashers which require multiple types of wire racks that must be fabricated and coated at different work centres before supplying the assembly lines. An approximate dynamic programming (ADP) method is proposed to address the complexities associated with such a system. In addition, an Artificial Neural Network model is designed to approximate state values of the system, thus helping the system to make decisions at particular states. A near optimal production and inventory control policy is developed through an ADP algorithm. The proposed method can be extended to any similar system.     

Integration of human behavioural aspects in a dynamic model for a manufacturing system

The computational simulation of human intelligent behaviour has been one of the main research topics in the artificial intelligence (AI) domain. Therefore, a great number of behavioural models were proposed considering emotional, cognitive and psychological factors to simulate the human behaviour in different domains such as military or manufacturing systems. In addition to psychological factors, the social state of a group of workers plays a critical role in rational decision-making, perception, human interaction, and human intelligence. Thus, it is judicious to analyse the workers’ behaviour at work and to integrate their needs and requirements in manufacturing systems models in order to improve the simulation accuracy. In this context, this paper suggests a graphical and a mathematical representation model of workers’ behaviours as well as the ties that can exist among them. The model is also extended to consider inter-worker social relations that can influence the individual performance.     

Synthesis of porous carbon spheres derived from lignin through a facile method for high performance supercapacitors

Porous carbon spheres (PCS) derived from lignin have been prepared through a facile method and fabricated as electrodes for electric double-layer capacitors. Spherical shaped mixtures of lignosulfonate and crystalized KOH are formed by spray drying of a solution of lignosulfonate and KOH. Activation by KOH is performed at high temperatures along with lignosulfonate carbonization. With an appropriate pore structure, the obtained PCS have a specific surface area of 1372.87 m² g^?1 and show a capacitance of 340 F g^?1 in 3 M KOH at a current density of 0.5 A g^?1. Moreover, a symmetric supercapacitor fabricated using the PCS as electrodes show a maximum capacitance of 68.5 F g^?1, and an energy density of 9.7 W h kg^?1 at a power density of 250 W kg^?1. The capacity retention is more than 94.5% after 5000 galvanostatic charge-discharge cycles. The excellent characteristics seem to be ascribed to the pore structures of PCS that have a large specific surface area and a low electrical resistance.     

基于模糊神经网络技术的定量秤研究

以传送带、料门给料的动态称重过程为对象,提出了一种新型动态定量称重控制方法。该方法从过程对象的实际出发,基于多元复合控制思想,称重策略采取分段控制并引入模糊神经网络控制技术,调节量采取给料门和传送带两个自由度协调。研究结果表明:该方法能够比较理想地解决动态定量称重过程中速度与准确度的矛盾。动态称重设定值为1000g时,该装置定量误差为±0 5%,称重速度<8s 次。     

Establishment of a Fault Prognosis Model Using Wavelet Neural Networks and Its Engineering Application

Fault diagnosis is confronted with two problems; how to “measure“ the growth of a fault and how to predict the remaining useful lifetime of such a failing component or machine. This paper attempts to solve these two problems by proposing a model of fault prognosis using wavelet basis neural network. Gaussian radial basis functions and Mexican hat wavelet frames are used as scaling functions and wavelets, respectively. The centers of the basis functions are calculated using a dyadic expansion scheme and a k-means clustering algorithm.     

Modeling of the microstructure variables in the isothermal compression of TC11 alloy using fuzzy neural networks

The grain size and volume fraction of prior α phase in high temperature deformation appear highly nonlinear and fuzzy characteristic. The approach to model the grain size and volume fraction of prior α phase and to train the model structure is presented in terms of the fuzzy set and artificial neural networks method using BP learning algorithm. The experimental data of teacher's samples are prepared from the grain size and volume fraction of prior α phase after isothermal compression of TC11 alloy at the deformation temperatures ranging from 1023 to 1323 K with an interval of 20 K, the strain rates ranging from 0.001 to 10.0 s⁻¹, and the height reductions ranging from 50 to 70%. The predicted grain size and volume fraction are in a good agreement with the experimental results in the isothermal compression of TC11 alloy.     

Classification performance comprehensive evaluation of an air classifier based on fuzzy analytic hierarchy process

The classification performance evaluation goal for an air classifier is usually limited to one of the classification performance indices including cut size, classification precision, Newton classification efficiency and degree of dispersion. This method hardly evaluates these performance indices of an air classifier comprehensively and suitably. In order to evaluate the classification performance truly and synthetically, Fuzzy Analytic Hierarchy Process is used to calculate the weights of the classification performance indices after determining the hierarchical model in the present paper. The dimensionless transformation eliminates the effect of the different dimensions. Then, the comprehensive evaluation value of the classification performance for each experiment is obtained using the linear weighted method. The maximum value corresponds to the best classification performance among these evaluation values. In the present study, a turbo air classifier is used as the classification system and talc powders are used as materials. The best classification performance indices are a cut size of 16.5 μm, a classification precision of 0.59, a Newton classification efficiency of 57%, and a degree of dispersion of 2.13. The corresponding optimal operational parameter combinations are: the feeding speed is 40 kg·h^–1, the air inlet velocity is 5 m·s^–1 and the rotor cage's rotary speed is 1200·min^–1. This assessment method avoids the limitation of evaluating a single classification performance index and the incomplete information derived from single factor experiments. Furthermore, the method also provides quantitative evaluation criteria for the classification performance of an air classifier. In the proposed method, the classification performance indices can be selected and the precedence relation matrix of Fuzzy Analytic Hierarchy Process can be set flexibly according to production requirements.     

网络入侵防御系统设计与测试

该文主要研究云计算网络环境下的入侵检测与防御技术,在总结传统入侵检测技术的基础上,对云计算环境中的入侵检测系统进行比较全面的研究,开发以神经网络技术为基础的网络入侵防御系统。对于入侵检测模块,重点对数据捕获、行为规则匹配以及神经网络判别模块进行分析,并通过具体的测试检验其实现结果。     

Health Assessment of Concrete Dams by Overall Inverse Analyses and Neural Networks

In several existing dams alcali–silica reaction (ASR) during several decades of service life, or diffused micro-cracking (due to concrete ageing and/or past extreme loads, such as earthquakes) give rise to deterioration of concrete stiffness and to correlated reduction of its strength. An inverse methodology is presented herein apt to identify damage in concrete dams on the basis of hydrostatic loading, measurements by traditional monitoring instruments, such as pendulums and collimators, and artificial neural networks trained by means of finite-element simulations. The arch-gravity dam referred to in this study is sub-divided into homogeneous zones, to which a constant Young modulus is attributed as unknown parameter which quantifies possible damage. These elastic moduli are estimated on the basis of pseudo-experimental data and identification procedures. After a suitable ‘training’ process, artificial neural networks (ANNs) are employed for numerical solutions of the inverse problem, and their potentialities and limitations are examined to the present purposes. In particular, they turn out to be robust and practically useful in the presence of information which are scarce quantitatively (few available measurements) and/or qualitatively (large noise-to-signal ratio).     

Optimization of a machining economics model with fuzzy exponents and coefficients

The primary objective of a machining economics model is to determine the optimal cutting parameters that minimize production costs while satisfying some design constraints. This paper develops a solution method that can derive the fuzzy unit production cost of a fuzzy machining economic model when the exponents of decision variables in the objective function, the cost and the constraint coefficients are fuzzy numbers. A pair of two-level machining economics problems is formulated to calculate the upper and lower bounds of the fuzzy unit production cost at possibility level α. Based on the duality theorem and by using a variable substitution technique, the two-level machining economics problem is transformed into the one-level conventional geometric program. Solving the corresponding pair of geometric programs produces the interval of the unit production cost. The examples show that the interval of unit production cost contain more information when the parameters in machining economics problems are fuzzy numbers.