Multi-layer perceptron-based data-driven multiscale modelling of granular materials with a novel Frobenius norm-based internal variable

One objective of developing machine learning(ML)-based material models is to integrate them with well-established numerical methods to solve boundary value problems(BVPs).In the family of ML models,recurrent neural networks(RNNs)have been extensively applied to capture history-dependent constitutive responses of granular materials,but these multiple-step-based neural networks are neither sufficiently efficient nor aligned with the standard finite element method(FEM).Single-step-based neural networks like the multi-layer perceptron(MLP)are an alternative to bypass the above issues but have to introduce some internal variables to encode complex loading histories.In this work,one novel Frobenius norm-based internal variable,together with the Fourier layer and residual architectureenhanced MLP model,is crafted to replicate the history-dependent constitutive features of representative volume element(RVE)for granular materials.The obtained ML models are then seamlessly embedded into the FEM to solve the BVP of a biaxial compression case and a rigid strip footing case.The obtained solutions are comparable to results from the FEM-DEM multiscale modelling but achieve significantly improved efficiency.The results demonstrate the applicability of the proposed internal variable in enabling MLP to capture highly nonlinear constitutive responses of granular materials.

Prediction of flyrock distance induced by mine blasting using a novel Harris Hawks optimization-based multi-layer perceptron neural network

In mining or construction projects,for exploitation of hard rock with high strength properties,blasting is frequently applied to breaking or moving them using high explosive energy.However,use of explosives may lead to the flyrock phenomenon.Flyrock can damage structures or nearby equipment in the surrounding areas and inflict harm to humans,especially workers in the working sites.Thus,prediction of flyrock is of high importance.In this investigation,examination and estimation/forecast of flyrock distance induced by blasting through the application of five artificial intelligent algorithms were carried out.One hundred and fifty-two blasting events in three open-pit granite mines in Johor,Malaysia,were monitored to collect field data.The collected data include blasting parameters and rock mass properties.Site-specific weathering index(WI),geological strength index(GSI) and rock quality designation(RQD)are rock mass properties.Multi-layer perceptron(MLP),random forest(RF),support vector machine(SVM),and hybrid models including Harris Hawks optimization-based MLP(known as HHO-MLP) and whale optimization algorithm-based MLP(known as WOA-MLP) were developed.The performance of various models was assessed through various performance indices,including a10-index,coefficient of determination(R^(2)),root mean squared error(RMSE),mean absolute percentage error(MAPE),variance accounted for(VAF),and root squared error(RSE).The a10-index values for MLP,RF,SVM,HHO-MLP and WOA-MLP are 0.953,0.933,0.937,0.991 and 0.972,respectively.R^(2) of HHO-MLP is 0.998,which achieved the best performance among all five machine learning(ML) models.

Prediction of Logistics Demand via Least Square Method and Multi-Layer Perceptron

To implement the prediction of the logistics demand capacity of a certain region,a comprehensive index system is constructed,which is composed of freight volume and other eight relevant economic indices,such as gross domestic product(GDP),consumer price index(CPI),total import and export volume,port's cargo throughput,total retail sales of consumer goods,total fixed asset investment,highway mileage,and resident population,to form the foundation for the model calculation.Based on the least square method(LSM)to fit the parameters,the study obtains an accurate mathematical model and predicts the changes of each index in the next five years.Using artificial intelligence software,the research establishes the logistics demand model of multi-layer perceptron(MLP)neural network,makes an empirical analysis on the logistics demand of Quanzhou City,and predicts its logistics demand in the next five years,which provides some references for formulating logistics planning and development strategy.

航空发动机滑油消耗率计算与预测方法

针对航空发动机滑油箱油量测量值易受多个参数影响导致滑油消耗率难以计算和预测的问题,提出了一种改进的滑油量数据提取规则和滑油消耗率预测方法。基于密度聚类算法(Density-based spatial clustering of applications with noise,DBSCAN)等方法对发动机数据进行了清洗,获取平稳飞行状态下滑油量数据。使用最小二乘法对滑油量进行拟合,得到了滑油消耗率,平均拟合优度达到了0.86。在此基础上,利用多层感知器(Multi-layer perception,MLP)建立了滑油消耗率与飞行状态参数之间的关系,预测结果与实际值的平均绝对百分比误差为1.15%。本文提出的方法能够满足实际工程需求,为评估航空发动机滑油系统的健康状况提供了可靠参考。

基于JSM和MLP改进发音错误检测的方法

针对发音错误检测的发音字典生成提出基于联合序列多阶模型(Joint-sequence multi-gram,JSM)和多层神经感知(Multi-layer perception,MLP)的方法.首先使用JSM模型对发音错误进行建模,将标准发音和错误发音组合为发音对,表示它们之间的对应关系,再使用N元文法来统计各发音对之间的关系,描述错误发音对上下文关系的依赖.最后使用MLP对发音对之间的关系进行重新建模,以学习到在相似的上下文条件下发生的相似的错误.实验证明使用MLP对高阶模型进行概率重估能有效的平滑概率空间,提高了发音错误检测的性能.

MLP training in a self-organizing state space model using unscented Kalman particle filter

Many Bayesian learning approaches to the multi-layer perceptron （MLP） parameter optimization have been proposed such as the extended Kalman filter （EKF）. This paper uses the unscented Kalman particle filter （UPF） to train the MLP in a self- organizing state space （SOSS） model. This involves forming augmented state vectors consisting of all parameters （the weights of the MLP） and outputs. The UPF is used to sequentially update the true system states and high dimensional parameters that are inherent to the SOSS moder for the MLP simultaneously. Simulation results show that the new method performs better than traditional optimization methods.

融合全局推理和MLP架构的甲状腺结节分割模型

针对甲状腺结节分割中存在的超声图像噪声干扰较大、结节尺寸多变和现有模型计算复杂度较高的问题,文中构建融合全局推理和多层感知机(Multi-layer Perception,MLP)架构的甲状腺结节分割模型.模型以轴向移位MLP模块为基础架构,以更小的计算复杂度实现不同空间位置特征之间的交互.在编码部分,融合端到端的全局推理单元,基于图卷积对图像全局信息进行交互,缓解图像噪声干扰较大的影响.在解码部分,引入金字塔特征层,完成多尺度特征交互,应对结节尺寸多变的问题.在DDIT数据集上的实验表明,文中模型性能较优,此外,文中模型还适用于乳腺结节分割、视网膜血管分割等其它医学图像分割任务.

Constitutive modelling of idealised granular materials using machine learning method

Predicting the constitutive response of granular soils is a fundamental goal in geomechanics.This paper presents a machine learning(ML)framework for the prediction of the stress-strain behaviour and shearinduced contact fabric evolution of an idealised granular material subject to triaxial shearing.The MLbased framework is comprised of a set of mini-triaxial tests which provide a benchmark for the setup and validation of the discrete element method(DEM)model of the granular materials,a parametric DEM simulation programme of virtual triaxial tests which provides datasets of micro-and macro-mechanical information,as well as a multi-layer perceptron(MLP)neural network which is trained and tested using the DEM-based datasets.The ML model only requires the initial void ratio of the granular sample as the input for predicting its constitutive response.The excellent agreement between the ML model prediction and experimental test and DEM simulation results indicates that the MLebased modelling approach is capable of capturing accurately the effects of initial void ratio on the constitutive behaviour of idealised granular materials,bypassing the need to incorporate the complex micromechanics underlying the macroscopic mechanical behaviour of granular materials.Lastly,a detailed comparison between the used MLP model and long short-term memory(LSTM)model was made from the perspective of technical algorithm,prediction accuracy,and computational efficiency.

基于机器学习的低渗透砂岩聚合物驱采收率预测

在恶劣的油藏条件下,化学驱提高采收率方法的可行性主要在实验室进行,以探究化学驱方案在现场实施的可能效果,但此类实验通常昂贵且费时。为了提高筛选效率和研究变量关系,进行了3个聚合物驱油实验项目,其次通过构建14种机器学习基础模型来预测低渗透砂岩聚合物驱油实验的效率。结果表明:多层感知机(multi-layer perception,MLP)、随机树(random forest,RF)和极限梯度上升(extreme gradient boosting,XGB)模型表现最佳,它们在测试集的确定系数均为0.99,均方根误差分别为0.855、0.836和0.859。模型表明特征重要性由强至弱依次为含水率、累积注入孔隙体积、渗透率、非均质系数、孔隙度、聚合物注入量、聚合物浓度、注入压力。研究成果为室内物理低渗透砂岩聚合物驱提供了可靠的数据,给出了14种机器学习模型预测性能直接对比,建立了高拟合高泛化高稳定低误差的低渗透砂岩聚合物驱预测模型,有助于化学驱方案快速在低渗透储层应用,以及降低失败风险。

Multi-objective optimization of the cathode catalyst layer micro-composition of polymer electrolyte membrane fuel cells using a multi-scale,two-phase fuel cell model and data-driven surrogates

Polymer electrolyte membrane fuel cells(PEMFCs)are considered a promising alternative to internal combustion engines in the automotive sector.Their commercialization is mainly hindered due to the cost and effectiveness of using platinum(Pt)in them.The cathode catalyst layer(CL)is considered a core component in PEMFCs,and its composition often considerably affects the cell performance(V_(cell))also PEMFC fabrication and production(C_(stack))costs.In this study,a data-driven multi-objective optimization analysis is conducted to effectively evaluate the effects of various cathode CL compositions on Vcelland Cstack.Four essential cathode CL parameters,i.e.,platinum loading(L_(Pt)),weight ratio of ionomer to carbon(wt_(I/C)),weight ratio of Pt to carbon(wt_(Pt/c)),and porosity of cathode CL(ε_(cCL)),are considered as the design variables.The simulation results of a three-dimensional,multi-scale,two-phase comprehensive PEMFC model are used to train and test two famous surrogates:multi-layer perceptron(MLP)and response surface analysis(RSA).Their accuracies are verified using root mean square error and adjusted R^(2).MLP which outperforms RSA in terms of prediction capability is then linked to a multi-objective non-dominated sorting genetic algorithmⅡ.Compared to a typical PEMFC stack,the results of the optimal study show that the single-cell voltage,Vcellis improved by 28 m V for the same stack price and the stack cost evaluated through the U.S department of energy cost model is reduced by$5.86/k W for the same stack performance.

Camera-Radar Fusion Sensing System Based on Multi-Layer Perceptron

Environmental perception is a key technology for autonomous driving.Owing to the limitations of a single sensor,multiple sensors are often used in practical applications.However,multi-sensor fusion faces some problems,such as the choice of sensors and fusion methods.To solve these issues,we proposed a machine learning-based fusion sensing system that uses a camera and radar,and that can be used in intelligent vehicles.First,the object detection algorithm is used to detect the image obtained by the camera;in sequence,the radar data is preprocessed,coordinate transformation is performed,and a multi-layer perceptron model for correlating the camera detection results with the radar data is proposed.The proposed fusion sensing system was verified by comparative experiments in a real-world environment.The experimental results show that the system can effectively integrate camera and radar data results,and obtain accurate and comprehensive object information in front of intelligent vehicles.

A hybrid constriction coefficientbased particle swarm optimization and gravitational search algorithm for training multi-layer perceptron

Purpose-In this paper,a newly proposed hybridization algorithm namely constriction coefficient-based particle swarm optimization and gravitational search algorithm(CPSOGSA)has been employed for training MLP to overcome sensitivity to initialization,premature convergence,and stagnation in local optima problems of MLP.Design/methodology/approach-In this study,the exploration of the search space is carried out by gravitational search algorithm(GSA)and optimization of candidate solutions,i.e.exploitation is performed by particle swarm optimization(PSO).For training the multi-layer perceptron(MLP),CPSOGSA uses sigmoid fitness function for finding the proper combination of connection weights and neural biases to minimize the error.Secondly,a matrix encoding strategy is utilized for providing one to one correspondence between weights and biases of MLP and agents of CPSOGSA.Findings-The experimental findings convey that CPSOGSA is a better MLP trainer as compared to other stochastic algorithms because it provides superior results in terms of resolving stagnation in local optima and convergence speed problems.Besides,it gives the best results for breast cancer,heart,sine function and sigmoid function datasets as compared to other participating algorithms.Moreover,CPSOGSA also provides very competitive results for other datasets.Originality/value-The CPSOGSA performed effectively in overcoming stagnation in local optima problem and increasing the overall convergence speed of MLP.Basically,CPSOGSA is a hybrid optimization algorithm which has powerful characteristics of global exploration capability and high local exploitation power.In the research literature,a little work is available where CPSO and GSA have been utilized for training MLP.The only related research paper was given by Mirjalili et al.,in 2012.They have used standard PSO and GSA for training simple FNNs.However,the work employed only three datasets and used the MSE performance metric for evaluating the efficiency of the algorithms.In this paper,eight different standard datasets and five performance metrics have been utilized for investigating the efficiency of CPSOGSA in training MLPs.In addition,a non-parametric pair-wise statistical test namely the Wilcoxon rank-sum test has been carried out at a 5%significance level to statistically validate the simulation results.Besides,eight state-of-the-art metaheuristic algorithms were employed for comparative analysis of the experimental results to further raise the authenticity of the experimental setup.

Prediction of diabetes and hypertension using multi-layer perceptron neural networks

Background:Diabetes and hypertension are two of the commonest diseases in the world.As they unfavorably affect people of different age groups,they have become a cause of concern and must be predicted and diagnosed well in advance.Objective:This research aims to determine the effectiveness of artificial neural networks(ANNs)in predicting diabetes and blood pressure diseases and to point out the factors which have a high impact on these diseases.Sample:This work used two online datasets which consist of data collected from 768 individuals.We applied neural network algorithms to predict if the individuals have those two diseases based on some factors.Diabetes prediction is based on five factors:age,weight,fat-ratio,glucose,and insulin,while blood pressure prediction is based on six factors:age,weight,fat-ratio,blood pressure,alcohol,and smoking.Method:A model based on the Multi-Layer Perceptron Neural Network(MLP)was implemented.The inputs of the network were the factors for each disease,while the output was the prediction of the disease’s occurrence.The model performance was compared with other classifiers such as Support Vector Machine(SVM)and K-Nearest Neighbors(KNN).We used performance metrics measures to assess the accuracy and performance of MLP.Also,a tool was implemented to help diagnose the diseases and to understand the results.Result:The model predicted the two diseases with correct classification rate(CCR)of 77.6%for diabetes and 68.7%for hypertension.The results indicate that MLP correctly predicts the probability of being diseased or not,and the performance can be significantly increased compared with both SVM and KNN.This shows MLPs effectiveness in early disease prediction.

三维激光扫描曲面重构算法研究

围绕三维激光扫描曲面重构做了以下工作:(1)分析了三维激光扫描仪的工作原理;(2)研究了利用多层感知器神经网络用于曲面重构,提出了基于多层感知器网络的曲面重构算法(M LPSR)。为了有效评估算法的性能,通过CYRAX 2500实测数据并利用M ATLAB环境进行仿真实验,结果表明:这种算法具有较高的重构精度和较快的重构速度,尤其在曲面受到破坏或不完全时特别有效,是值得推广的曲面重构算法。

基于前馈多层感知器的网络入侵检测的多数据包分析

提出了一种新型网络入侵检测模型,在该模型中,首先将截获的数据包结合历史数据包数据库进行协议分析,找出可能存在的入侵行为的相关数据包,然后采用前馈多层感知器神经网络对这些相关的数据包进行回归分析,最终获得检测结果。该模型与传统采用单数据包检测方式的网络入侵检测系统(NIDS)模型相比,具有更低的漏检率。

基于AR自相关峰态值的一类轴承故障检测方法

针对轴承故障检测系统中异常样本数据不易收集以及异常样本数据分布不均导致传统分类算法出现过适应现象等现实应用问题,提出了一种基于自回归(AR)模型自相关系数峰态特征的一类故障检测方法。该方法利用正常样本生成AR模型参数,其他样本在该模型的投影形成残差序列,计算残差序列的自相关系数并取其峰态特征作为相似性的度量。实验结果表明该方法能有效地克服以AR模型参数为特征计算复杂度高且检测性能易受样本大小影响的不足。同时,文章给出了单一故障诊断模型并提出基于粒子群优化算法的阈值设定决策方法。实验中将本方法同其他以AR模型为特征的多层感知机(MLP)及自组织映射(SOM)方法进行比较,实验结果验证了本文建议方法的正确性和有效性。

可变神经网络结构下的遥感影像光谱分解方法

多层感知神经网络(MLP)是主流的非线性分解方法,但是目前缺乏有效方法处理MLP分解结果中的丰度负值问题。为此,提出一种可变神经网络结构的方法,逐步去除负值丰度对应的端元,并调整相应的网络结构使之针对剩余的端元进行分解。通过武汉地区模拟TM遥感影像实验可以发现,该方法与传统MLP方法以及线性光谱分解方法的平均误差分别为0.077 7、0.081 9、0.094 3,说明该方法的分解精度高于其他2种分解方法,能克服丰度负值问题。

基于KPCA空间相似度的一类入侵检测方法

为了解决入侵检测系统中异常样本数据不易收集以及异常样本数据分布不均导致传统分类算法出现过适应现象等现实应用问题,提出了一种基于核主成分分析(KPCA)空间相似度的一类入侵检测方法。该方法利用KPCA形成正常样本的非线性特征子空间,其他样本在该空间的投影系数作为相似性的度量。同时,为了有效利用已有的异常训练样本,通过自适应增加免疫因子方法来提高模型的决策性能及增量学习能力。对核函数参数和阈值设定进行了分析,并给出基于粒子群优化算法的决策模型。实验中将该方法同其他多层感知机(MLP),支持向量机(SVM)及自组织映射(SOM)方法进行比较,实验结果验证了该方法的正确性和有效性。

基于MLP和多头自注意力特征融合的双模态情感计算模型

针对情感计算中传统的单模态情感分析通常存在分类准确率不高和不同语言环境间泛化能力较差的问题,提出一种双模态情感计算模型,以同时使用包含中英文两种语言、两种不同模态的情感数据。首先,利用多层感知机(MLP)网络和双向长短时记忆(BiLSTM)网络对数据进行特征提取;其次,基于MLP和自注意力机制分别对提取的特征进行特征融合,得到多模态分析模型;最后,使用该模型在构建的包含中英文两种语言数据的数据集上进行二分类情感计算预测。实验结果表明,所提模型相较于次优的BiLSTM模型,精度提高了1.22%;相较于单模态情感计算模型,精度提高了6.21%~14.00%。

基于CHMT/NN的小波域纹理图象分割新算法

文中提出了一种基于隐Markov模型和多层感知器的小波域图象纹理分割新算法。首先该算法通过图形组合方法有效地提取了图像在小波变换域各子带之间的相关性,然后应用多层感知器进行分类,将HMM的规范性和MLP神经网络的分类能力有效地结合起来。最后给出了文中算法对Brodatz纹理的分类结果。实验证明了文中算法的有效性。