Fully Distributed Learning for Deep Random Vector Functional-Link Networks

In the contemporary era, the proliferation of information technology has led to an unprecedented surge in data generation, with this data being dispersed across a multitude of mobile devices. Facing these situations and the training of deep learning model that needs great computing power support, the distributed algorithm that can carry out multi-party joint modeling has attracted everyone’s attention. The distributed training mode relieves the huge pressure of centralized model on computer computing power and communication. However, most distributed algorithms currently work in a master-slave mode, often including a central server for coordination, which to some extent will cause communication pressure, data leakage, privacy violations and other issues. To solve these problems, a decentralized fully distributed algorithm based on deep random weight neural network is proposed. The algorithm decomposes the original objective function into several sub-problems under consistency constraints, combines the decentralized average consensus (DAC) and alternating direction method of multipliers (ADMM), and achieves the goal of joint modeling and training through local calculation and communication of each node. Finally, we compare the proposed decentralized algorithm with several centralized deep neural networks with random weights, and experimental results demonstrate the effectiveness of the proposed algorithm.

Optimized functional linked neural network for predicting diaphragm wall deflection induced by braced excavations in clays

Deep excavation during the construction of underground systems can cause movement on the ground,especially in soft clay layers.At high levels,excessive ground movements can lead to severe damage to adjacent structures.In this study,finite element analyses(FEM)and the hardening small strain(HSS)model were performed to investigate the deflection of the diaphragm wall in the soft clay layer induced by braced excavations.Different geometric and mechanical properties of the wall were investigated to study the deflection behavior of the wall in soft clays.Accordingly,1090 hypothetical cases were surveyed and simulated based on the HSS model and FEM to evaluate the wall deflection behavior.The results were then used to develop an intelligent model for predicting wall deflection using the functional linked neural network(FLNN)with different functional expansions and activation functions.Although the FLNN is a novel approach to predict wall deflection;however,in order to improve the accuracy of the FLNN model in predicting wall deflection,three swarm-based optimization algorithms,such as artificial bee colony(ABC),Harris’s hawk’s optimization(HHO),and hunger games search(HGS),were hybridized to the FLNN model to generate three novel intelligent models,namely ABC-FLNN,HHO-FLNN,HGS-FLNN.The results of the hybrid models were then compared with the basic FLNN and MLP models.They revealed that FLNN is a good solution for predicting wall deflection,and the application of different functional expansions and activation functions has a significant effect on the outcome predictions of the wall deflection.It is remarkably interesting that the performance of the FLNN model was better than the MLP model with a mean absolute error(MAE)of 19.971,root-mean-squared error(RMSE)of 24.574,and determination coefficient(R^(2))of 0.878.Meanwhile,the performance of the MLP model only obtained an MAE of 20.321,RMSE of 27.091,and R^(2)of 0.851.Furthermore,the results also indicated that the proposed hybrid models,i.e.,ABC-FLNN,HHO-FLNN,HGS-FLNN,yielded more superior performances than those of the FLNN and MLP models in terms of the prediction of deflection behavior of diaphragm walls with an MAE in the range of 11.877 to 12.239,RMSE in the range of 15.821 to 16.045,and R^(2)in the range of 0.949 to 0.951.They can be used as an alternative tool to simulate diaphragm wall deflections under different conditions with a high degree of accuracy.

Functional Link Neural Network for Predicting Crystallization Temperature of Ammonium Chloride in Air Cooler System

The air cooler is an important equipment in the petroleum refining industry.Ammonium chloride(NH4 Cl)deposition-induced corrosion is one of its main failure forms.In this study,the ammonium salt crystallization temperature is chosen as the key decision variable of NH4 Cl deposition-induced corrosion through in-depth mechanism research and experimental analysis.The functional link neural network(FLNN)is adopted as the basic algorithm for modeling because of its advantages in dealing with non-linear problems and its fast-computational ability.A hybrid FLNN attached to a small norm is built to improve the generalization performance of the model.Then,the trained model is used to predict the NH4 Cl salt crystallization temperature in the air cooler of a sour water stripper plant.Experimental results show the proposed improved FLNN algorithm can achieve better generalization performance than the PLS,the back propagation neural network,and the conventional FLNN models.

Underwater Image Classification Based on EfficientnetB0 and Two-Hidden-Layer Random Vector Functional Link

The ocean plays an important role in maintaining the equilibrium of Earth’s ecology and providing humans access to a wealth of resources.To obtain a high-precision underwater image classification model,we propose a classification model that combines an EfficientnetB0 neural network and a two-hidden-layer random vector functional link network(EfficientnetB0-TRVFL).The features of underwater images were extracted using the EfficientnetB0 neural network pretrained via ImageNet,and a new fully connected layer was trained on the underwater image dataset using the transfer learning method.Transfer learning ensures the initial performance of the network and helps in the development of a high-precision classification model.Subsequently,a TRVFL was proposed to improve the classification property of the model.Net construction of the two hidden layers exhibited a high accuracy when the same hidden layer nodes were used.The parameters of the second hidden layer were obtained using a novel calculation method,which reduced the outcome error to improve the performance instability caused by the random generation of parameters of RVFL.Finally,the TRVFL classifier was used to classify features and obtain classification results.The proposed EfficientnetB0-TRVFL classification model achieved 87.28%,74.06%,and 99.59%accuracy on the MLC2008,MLC2009,and Fish-gres datasets,respectively.The best convolutional neural networks and existing methods were stacked up through box plots and Kolmogorov-Smirnov tests,respectively.The increases imply improved systematization properties in underwater image classification tasks.The image classification model offers important performance advantages and better stability compared with existing methods.

Numeral eddy current sensor modelling based on genetic neural network

This paper presents a method used to the numeral eddy current sensor modelling based on the genetic neural network to settle its nonlinear problem. The principle and algorithms of genetic neural network are introduced. In this method, the nonlinear model parameters of the numeral eddy current sensor are optimized by genetic neural network （GNN） according to measurement data. So the method remains both the global searching ability of genetic algorithm and the good local searching ability of neural network. The nonlinear model has the advantages of strong robustness, on-line modelling and high precision. The maximum nonlinearity error can be reduced to 0.037% by using GNN. However, the maximum nonlinearity error is 0.075% using the least square method.

A New Modeling Method Based on Genetic Neural Network for Numeral Eddy Current Sensor

In this paper,we present a method used to the numeral eddy current sensor modeling based on genetic neural network to settle its nonlinear problem.The principle and algorithms of genetic neural network are introduced.In this method, the nonlinear model parameters of the numeral eddy current sensor are optimized by genetic neural network (GNN) according to measurement data.So the method remains both the global searching ability of genetic algorithm and the good local searching ability of neural network.The nonlinear model has the advantages of strong robustness,on-line scaling and high precision.The maximum nonlinearity error can be reduced to 0.037% using GNN.However,the maximum nonlinearity error is 0.075% using least square method (LMS).

Multilayer perceptron and Chebyshev polynomials-based functional link artificial neural network for solving differential equations

This paper discusses the issues of computational efforts and the accuracy of solutions of differential equations using multilayer perceptron and Chebyshev polynomials-based functional link artificial neural networks.Some ordinary and partial differential equations have been solved by both these techniques and pros and cons of both these type of feedforward networks have been discussed in detail.Apart from that,various factors that affect the accuracy of the solution have also been analyzed.

Pan evaporation modeling in different agroclimatic zones using functional link artificial neural network

Pan evaporation is an important climatic variable for developing efficient water resource management strategies.In the past,many machine learning models are reported in the literature for pan evaporation modeling using the different combinationof available climatic variables.In order to develop a novel model with improved accuracy and reduced computational complexity,the functional link artificial neural network(FLANN)is chosen as an architecture to estimate daily pan evaporation in three agro-climatic zones(ACZs)of Chhattisgarh state in east-central India.Single neuron and single layer in its structure make it less complex as compared to other multilayer neural networks and neuro-fuzzy based hybrid models.Estimation results obtained with the FLANN model are compared with those obtained by multi-layer artificial neural networks(MLANN)and two empirical methods using the same raw data and corresponding features.Statistical indices like root mean square error(RMSE),mean absolute error(MAE)and efficiency factor(EF)is also computed to evaluate the model performance.It is demonstrated that pan evaporation estimates obtained with the proposed FLANN models provide an improved estimation of pan evaporation(RMSE=0.85 to 1.27 mm d^(-1),MAE=0.63 to 0.95 mm d^(-1) and EF=0.70 to 0.89)as compared to MLANN(RMSE=0.94 to 1.58 mm d^(-1),MAE=0.73 to 1.14 mm d^(-1) and EF=0.62 to 0.88)and empirical(RMSE=1.19 to 2.19 mm d^(-1),MAE=0.91 to 1.62 mm d^(-1) and EF=0.49 to 0.88)models in different ACZs.

遗传算法结合FLNN实现加速度传感器动态特性补偿

对加速度传感器动态性能进行分析,利用遗传算法与函数链神经网络相结合实现其动态性能补偿的方法,介绍补偿原理以及算法,给出了用遗传算法和函数链神经网络相结合建立的加速度传感器动态补偿网络的数学模型。结果表明,这种补偿模型具有精度高、有良好的鲁棒性以及动态补偿器实现简单等优点,在测试领域中有很好的应用前景。

基于FLNN的多粘菌素发酵过程建模

为解决由于缺乏传感器使众多状态参数难以在线测量的问题,建立了多粘菌素的发酵过程模型,对许多重要的状态参数进行了预估.通过在FLNN内部增加一个带有局部激活反馈和一个局部输出反馈的自回归滑动平均滤波器使其成为动态的FLNN网络,并把它运用于多粘菌素发酵过程的建模中,结合遗传算法实现对其发酵过程的菌体浓度、总糖浓度和相对效价进行预估,为实际生产和优化控制提供了有利条件.仿真结果表明,基于改进的FLNN建立的多粘菌素发酵过程模型预估效果良好.

基于FLNN的加速度传感器动态特性补偿方法

对加速度传感器动态性能进行分析,提出其动态性能补偿的神经网络方法,介绍了补偿原理以及神经网络算法,给出用函数连接型神经网络建立的加速度传感器动态补偿网络的数学模型。结果表明:这种补偿模型精度高、能实现在线修正,有良好的鲁棒性及动态补偿器实现简单等优点。

一种新型的动态FLNN及其在系统辨识中的应用

本文介绍了一种新型的人工神经网络的改进 ,并把它运用于非线性系统的辨识中。这种新型的网络就是带有内部动态元的FLNN(Functional-LinkNeuralNetwork)。其中内部动态元分别由带有局部激活反馈和局部输出反馈的自回归滑动平均滤波器构成。其具体的动态网络参数寻优由遗传算法来决定。仿真结果表明 ,把这种改善了的FLNN与原有的外部带动态元的FLNN分别应用于系统辨识中 。

基于FLNN的塔机起重量软测量方法研究

通过起重量限制器的受力分析,指出在塔式起重机(塔机)起重量软测量中钢丝绳张力与拉力传感器拉力之间呈非线性关系,提出应用函数型连接神经网络(FLNN)建立软测量模型,为塔机起重量的间接在线测量提供了新方法。结合塔机QTZ63给出了网络的实现过程,实测研究表明:该测量方法具有误差小、精度高、容易实现等优点。

无人机纵向飞行控制器的FLNN轨迹线性化设计

提出一种将函数连接神经网络(FLNN)和轨迹线性化方法(TLC)相结合的新的研究控制方法,并采用此方法设计了无人机纵向飞行控制器。介绍了采用此方法设计无人机纵向飞行控制器的过程,并利用新控制方案对控制器进行了仿真验证,且与TLC控制器进行了比较。结果表明所提出的控制方案可有效地降低建模误差和外部干扰对控制器的影响,提高系统的鲁棒性。

基于MFLNN的变参数非线性系统辨识

函数型连接神经网络的网络结构简单,计算复杂度低。该文提出了一种外积扩展型连接神经网络(MFLNN),用于辨识变参数非线性系统,仿真结果表明,MFLNN实现了变参数非线性系统的辨识,效果显著。

基于外积FLNN的非线性系统辨识

函数型连接神经网络通过对输入模式预先进行非线性扩展,增强了输入信号的模式表达,从而大大简化网络结构,降低计算复杂度。本文提出一种外积扩展型连接神经网络,用于辨识幂函数非线性系统,并与MLP和CFLNN网络对比,仿真结果表明,外积型辨识幂函数非线性系统结构简单、计算量低、性能最优。

Brain networks modeling for studying the mechanism underlying the development of Alzheimer’s disease

Alzheimer’s disease is a primary age-related neurodegenerative disorder that can result in impaired cognitive and memory functions.Although connections between changes in brain networks of Alzheimer’s disease patients have been established,the mechanisms that drive these alterations remain incompletely understood.This study,which was conducted in 2018 at Northeastern University in China,included data from 97 participants of the Alzheimer’s Disease Neuroimaging Initiative(ADNI)dataset covering genetics,imaging,and clinical data.All participants were divided into two groups:normal control(n=52;20 males and 32 females;mean age 73.90±4.72 years)and Alzheimer’s disease(n=45,23 males and 22 females;mean age 74.85±5.66).To uncover the wiring mechanisms that shaped changes in the topology of human brain networks of Alzheimer’s disease patients,we proposed a local naive Bayes brain network model based on graph theory.Our results showed that the proposed model provided an excellent fit to observe networks in all properties examined,including clustering coefficient,modularity,characteristic path length,network efficiency,betweenness,and degree distribution compared with empirical methods.This proposed model simulated the wiring changes in human brain networks between controls and Alzheimer’s disease patients.Our results demonstrate its utility in understanding relationships between brain tissue structure and cognitive or behavioral functions.The ADNI was performed in accordance with the Good Clinical Practice guidelines,US 21 CFR Part 50-Protection of Human Subjects,and Part 56-Institutional Review Boards(IRBs)/Research Good Clinical Practice guidelines Institutional Review Boards(IRBs)/Research Ethics Boards(REBs).

基于一种NW-FLNN神经网络的短期电价预测

针对传统神经网络收敛速度慢、容易陷入局部极值的问题,文中提出一种改进型小波神经网络以实现网络全局最优化。首先,将小波神经网络与随机矢量函数连接型网络相融合构建一种新型小波链神经网络( NW-FLNN);其次,以小波基函数作为NW-FLNN的隐含层的传递函数,并利用梯度修正法训练该模型各参数;最后,选用澳大利亚新南威尔士州电价数据作为实验数据集,分别对 NW-FLNN神经网络、逆传播 B P神经网络与小波神经网络进行预测性能比较。实验结果表明:该新型网络预测模型较B P神经网络与小波神经网络性能更优,可明显减少网络迭代次数与隐层神经元数目,且平均百分比误差最大降低至0. 0317,满足实时性要求。

特征扩展的随机向量函数链神经网络

基于宽度学习的动态模糊推理系统(broad-learning-based dynamic fuzzy inference system,BL-DFIS)能自动构建出精简的模糊规则并获得良好的分类性能.然而,当遇到大型复杂的数据集时,BL-DFIS因会使用较多模糊规则来试图达到令人满意的识别精度,从而对其可解释性造成了不利影响.对此,提出一种兼顾分类性能和可解释性的模糊神经网络,将其称为特征扩展的随机向量函数链神经网络(FA-RVFLNN).在该网络中,一个以原始数据为输入的RVFLNN被作为主体结构,BL-DFIS则用作性能补充,这意味着FA-RVFLNN包含具有性能增强作用的直接链接.由于主体结构的增强节点使用Sigmoid激活函数,因此,其推理过程可借助一种模糊逻辑算子(I-OR)来解释.而且,具有明确含义的原始输入数据也有助于解释主体结构的推理规则.在直接链接的支撑下,FA-RVFLNN可利用增强节点、特征节点和模糊节点学到更丰富的有用信息.实验表明:FA-RVFLNN既减缓了主体结构RVFLNN中过多增强节点带来的“规则爆炸”问题,也提高了性能补充结构BL-DFIS的可解释性(平均模糊规则数降低了50%左右),在泛化性能和网络规模上仍具有竞争力.

MODEL REFERENCE ADAPTIVE CONTROL BASED ON NONLINEAR COMPENSATION FOR TURBOFAN ENGINE

The design of a turbofan rotor speed control system, using model reference adaptive control（MRAC） method with input and output measurements, is discussed for the purpose of practical application. The nonlinear compensator based on functional link neural network is used to deal with the engine nonlinearity and the hardware-in-loop simulation is also developed. The results show that the nonlinear MRAC controller has the adequate performance of compensating and adapting nonlinearity arising from the change of engine state or working environment. Such feature demonstrates potential practical applications of MRAC for aeroengine control system.