Novel Neural Network Inspired by Neuro-Endocrine-Immune System with Its Application to Beam Pumping Unit

Inspired by the modulation mechanism of neuroendocrine-immune system(NEIs),a novel structure of artificial neural network(ANN) named NEI-NN and its learning method are presented.The NEI-NN includes two parts,i.e.,positive subnetwork(PSN) and negative sub-network(NSN).The neuron functions of PSN and NSN are designed according to the increased and decreased secretion functions of hormone,respectively.In order to make the novel neural network learn quickly,the novel neuron based on some characteristics of NEIs is also redesigned.Besides the normal input signals,two control signals are considered in the proposed solution.One is the enable/disable signal,and the other is the slope control signal.The former can modify the structure of NEI-NN,and the later can regulate the evolutionary speed of NEINN.The NEI-NN can obtain the optimized network structure by using error back-propagation(BP) learning algorithm.Since the modeling of the beam pumping unit is very difficult by using the conventional method,the modeling of bean bump unit is chosen to examine the performance of the NEI-NN.The experiment results show that the optimized structure and learning speed of NEI-NN are better than those of the conventional neural network.

Immune Feedforward Neural Network for Fault Detection

An immune algorithm-based approach was developed to optimize a feedforward neural network. The network architecture, activation functions, and training method were encoded as individuals with an ap- propriate method for individual selection. The immune feedforward neural network is then applied to fault detection of water quality monitoring equipment. This gives better performance than a feedforward neural network.

Real-time immune-inspired optimum state-of-charge trajectory estimation using upcoming route information preview and neural networks for plug-in hybrid electric vehicles fuel economy

The main proposition of the current investigation is to develop a computational intelligence-based framework which can be used for the real-time estimation of optimum battery state-of-charge （SOC） trajectory in plug-in hybrid electric vehicles （PHEVs）. The estimated SOC trajectory can be then employed for an intelligent power management to significantly improve the fuel economy of the vehicle. The devised intelligent SOC trajectory builder takes advantage of the upcoming route information preview to achieve the lowest possible total cost of electricity and fossil fuel. To reduce the complexity of real-time optimization, the authors propose an immune system-based clustering approach which allows categoriz- ing the route information into a predefined number of segments. The intelligent real-time optimizer is also inspired on the basis of interactions in biological immune systems, and is called artificial immune algorithm （AIA）. The objective function of the optimizer is derived from a computationally efficient artificial neural network （ANN） which is trained by a database obtained from a high-fidelity model of the vehicle built in the Autonomic software. The simulation results demonstrate that the integration of immune inspired clustering tool, AIA and ANN, will result in a powerful framework which can generate a near global optimum SOC trajectory for the baseline vehicle, that is, the Toyota Prius PHEV. The outcomes of the current investigation prove that by taking advantage of intelligent approaches, it is possible to design a computationally efficient and powerful SOC trajectory builder for the intelligent power management of PHEVs.

Novel design concepts for network intrusion systems based on dendritic cells processes

An abstraction and an investigation to the worth of dendritic cells (DCs) ability to collect, process and present antigens are presented. Computationally, this ability is shown to provide a feature reduction mechanism that could be used to reduce the complexity of a search space, a mechanism for development of highly specialized detector sets as well as a selective mechanism used in directing subsets of detectors to be activated when certain danger signals are present. It is shown that DCs, primed by different danger signals, provide a basis for different anomaly detection pathways. Different antigen-peptides are developed based on different danger signals present, and these peptides are presented to different adaptive layer detectors that correspond to the given danger signal. Experiments are then undertaken that compare current approaches, where a full antigen structure and the whole repertoire of detectors are used, with the proposed approach. Experiment results indicate that such an approach is feasible and can help reduce the complexity of the problem by significant levels. It also improves the efficiency of the system, given that only a subset of detectors are involved during the detection process. Having several different sets of detectors increases the robustness of the resulting system. Detectors developed based on peptides are also highly discriminative, which reduces the false positives rates, making the approach feasible for a real time environment.

Multi-Agent Network Intrusion Active Defense Model Based on Immune Theory

Inspired by the immune theory and multi-agent systems, an immune multi-agent active defense model for network intrusion is established. The concept of immune agent is introduced, and its running mechanism is established. The method, which uses antibody concentration to quantitatively describe the degree of intrusion danger, is presented. This model implements the multi-layer and distributed active defense mechanism for network intrusion. The experiment results show that this model is a good solution to the network security defense.

Modified Self-adaptive Immune Genetic Algorithm for Optimization of Combustion Side Reaction of p-Xylene Oxidation

In recent years, immune genetic algorithm (IGA) is gaining popularity for finding the optimal solution for non-linear optimization problems in many engineering applications. However, IGA with deterministic mutation factor suffers from the problem of premature convergence. In this study, a modified self-adaptive immune genetic algorithm (MSIGA) with two memory bases, in which immune concepts are applied to determine the mutation parameters, is proposed to improve the searching ability of the algorithm and maintain population diversity. Performance comparisons with other well-known population-based iterative algorithms show that the proposed method converges quickly to the global optimum and overcomes premature problem. This algorithm is applied to optimize a feed forward neural network to measure the content of products in the combustion side reaction of p-xylene oxidation, and satisfactory results are obtained.

Optimizing control of coal flotation by neuro-immune algorithm

Coal flotation is widely used to separate commercially valuable coal from the fine ore slurry, and is an industrial process with nonlinear, multivariable, time-varying and long time-delay characteristics. The online detection of ash content of products as the operation performance evaluation in the flotation system is extraordinarily difficult because of the low solid content and numerous micro-bubbles in the slurry. Moreover, it is time-consuming by manual analysis. Consequently, the optimal separation is not usually maintained. A novel technique, called the neuro-immune algorithm (NIA) inspired by the biological nervous and immune systems, is presented in this paper for predicting the ash content of clean coal and performing the optimizing control to the coal flotation system. The proposed algorithm integrates the deeply-studied artificial neural network (ANN) and the developing artificial immune system (AIS). A two-layer back-propagation network was constructed offline based on the historical process data under the best system situation, using five parameters: the flow and the density of raw slurry, the input flows of water, the kerosene and the GF oil, as the inputs and the ash content of clean coal as the output. The immune cell of AIS is made up of six parameters above as the antigen. The cytokine based clone selection algorithm is used to produce the relative antibody. The detailed computation procedures about the hybrid neuro-immune algorithm are minutely discussed. The ash content of clean coal was predicted by NIA using the practical process data s: (308.6 174.7 146.1 43.6 4.0 9.4), and the absolute difference between the actual and computed ash content values was 0.0967%. The optimizing control on NIA was simulated considering two different situations where the ash content of clean coal was controlled downward from 10.00% or upward from 9.20% predicted by ANN to the target value 9.50%. The results indicate that the target ash content and the value of controlling parameters are obtained after several control cycles.

人工神经网络筛选类风湿关节炎的特征基因及免疫细胞浸润分析

目的:运用人工神经网络确定类风湿关节炎(RA)的特征基因,并分析免疫细胞在RA相关微环境中的作用。方法:GSE1919和GSE77298芯片均来自GEO数据库。运用R语言将两芯片进行合并与批次矫正,得到一个新数据集,并进行差异分析。对差异表达基因(DEGs)进行Metascape富集分析和GO、KEGG富集分析。运用R软件“randomForest”包在随机森林(RF)算法下筛选RA特征基因,并依据基因评分构建人工神经网络模型。提取排名前4的基因(HubGene)进行后续分析。应用单样本基因集富集分析(ssGSEA)计算样品中免疫细胞丰度并进行相关性分析。结果:人工神经网络模型筛选出排名前15的基因作为RA特征基因:STAT1、RUNX3、AR、CDH11、LMO4、TIMP1、PLXNC1、CAP2、PRKAA2、VDR、SPP1、HCK、EPHB2、KCNAB1、ITGB7,其中STAT1、RUNX3、CDH11在RA滑膜组织中均上调,AR在RA滑膜组织中下调。免疫细胞浸润结果显示,RA与活化CD4 T细胞相关性最显著。与RUNX3存在显著相关性的免疫细胞最多。RUNX3与活化B细胞、活化CD4 T细胞、活化CD8 T细胞、中央记忆CD4+T细胞、效应性记忆CD8 T细胞、调节性T细胞、γδT细胞和巨噬细胞呈显著正相关,但与NK细胞呈显著负相关。结论:通过人工神经网络确定了与RA相关的15个特征基因,其中排名前4的基因为STAT1、RUNX3、AR、CDH11。强调了活化CD4 T细胞、调节性T细胞、γδT细胞、巨噬细胞、NK细胞、活化B细胞等免疫细胞在RA发病机制中的重要性,为RA诊断和免疫细胞分子机制研究提供了新见解。

基于人工智能技术的轻量级网络入侵检测系统设计

以提升网络入侵检测技术水平为目的,设计基于人工智能技术的轻量级网络入侵检测系统。该系统数据采集层利用若干个用户探针连接IDS检测服务器后,使用网络数据包捕获模块捕获用户网络运行数据,再通过传输层内防火墙、核心交换机和MQTT/CoAP通信协议将用户网络运行数据发送到逻辑运算层内,该层利用数据预处理模块对用户网络运行数据进行去噪预处理后,将其输入到基于人工智能的网络入侵检测模块内,通过该模块输出轻量级网络入侵检测结果,然后将检测结果发送到展示层,通过入侵告警信息、数据可视化展示等模块实现人机交互。实验表明:该系统运行较为稳定,可有效检测不同类型网络入侵的同时,其检测及时性和入侵告警能力较好,应用效果良好。

基于硬皮病线粒体相关基因的人工神经网络模型的构建

目的建立基于GEO数据库硬皮病线粒体相关基因的机器学习和人工神经网络联合诊断模型并评价其效果。方法通过GEO数据库获取3份硬皮病芯片。其中GSE95065及GSE59785合并作为实验数据集并提取线粒体相关基因表达量,使用随机森林、LASSO回归和SVM算法筛选硬皮病线粒体相关特征基因,并用特征基因构建人工神经网络模型,用10折交叉验证模型准确性。来用验证数据集GSE76807对模型进一步验证,利用ROC曲线下面积值评估模型准确性。用RT-qPCR实验验证关键基因mRNA相对表达量。最后用CIBERSORT算法预估硬皮病与筛选出的潜在生物标志物的生物信息学关联。结果共获取差异基因24个,其中上调基因11个,下调基因13个。通过3种机器学习算法筛选到最相关的7个线粒体相关特征基因(POLB、GSR、KRAS、NT5DC2、NOX4、IGF1、TGM2),并构建人工神经网络诊断模型。使用该模型绘制了实验组和验证组诊断的ROC曲线,AUC值为0.984。验证组AUC为0.740。10折交叉验证AUC平均值大于0.980。RT-qPCR结果显示,与对照组相比,硬皮病中POLB(P=0.004)、GSR(P=0.029)、KRAS(P=0.007)、NOX4(P=0.019)、IGF1(P=0.008)、TGM2(P<0.0001)表达量明显上调,而NT5DC2(P=0.001)表达量在硬皮病组中明显下调。免疫细胞浸润显示,特征基因与滤泡辅助T细胞、幼稚B细胞、静息树突状细胞、记忆激活CD4+T细胞、巨噬细胞M0、单核细胞、记忆静息CD4+T细胞和肥大细胞激活等相关。结论构建了硬皮病特征基因的人工神经网络诊断模型,为探索硬皮病发病机制提供了一个新视角。

基于人工神经网络模型分析硬皮病关键基因和免疫浸润机制及靶向中药预测

目的建立基于基因表达综合数据库(Gene Expression Omnibus,GEO)硬皮病相关基因的机器学习和人工神经网络(Artificial neural network,ANN)联合诊断模型并评价其效果及预测和分析靶向中药。方法通过GEO数据库获取两份硬皮病芯片GSE23741与GSE95065合并作为训练组数据集,使用随机森林和lasso回归算法筛选硬皮病关键基因并构建用于诊断硬皮病的ANN模型。使用验证数据集GSE76807、GSE32413和GSE59785对模型进行验证,通过曲线下面积(Area under the curve,AUC)分析评价ANN模型的临床应用价值。用实时荧光定量PCR(Real-time quantitative PCR,RT-qPCR)验证关键基因mRNA相对表达量。利用CIBERSORT算法预估硬皮病与筛选出的潜在生物标志物的生物信息学关联。最后用关键基因进行靶向中药筛选。结果共获取差异基因167个。进一步通过机器学习筛选得到5个关键基因(SERPINE2、SFRP4、SUGCT、FBLN5、NRXN2),并构建了ANN诊断模型。使用该模型绘制了训练组和验证组诊断的受试者操作特征(Receiver operating characteristic,ROC)曲线,训练组AUC值为1.000。验证组AUC为0.770、0.795及0.872。RT-qPCR验证结果与机器学习算法相一致。免疫细胞浸润分析显示硬皮病组记忆CD4+T细胞激活相对含量明显增加,而正常组γδT细胞的相对含量明显增加。关键基因与巨噬细胞M1、T细胞、记忆激活CD4+T细胞、静息肥大细胞、CD8+T细胞等相关。根据关键基因筛选出12味中药,其四气五味多属温、寒、平,甘、辛、苦,多归肝、脾、肺经,功效多为补虚祛风湿药。结论构建了硬皮病关键基因的人工神经网络诊断模型,可用于临床诊断硬皮病,并预测出了潜在治疗硬皮病的靶向中药,为探索其发病机制和治疗方向提供了一个新视角。

基于IGA-BP神经网络的PEMFC供氢系统模型预测控制算法

对化石能源的过度使用导致了严重的环境问题和能源担忧。氢能作为一种清洁的能源,被认为是实现能源转型和可持续发展的重要资源。在此背景下,氢燃料电池作为一种将氢能高效转化为电能的技术,展现出了巨大潜力。本文以质子交换膜燃料电池阳极供氢系统为研究对象,以氢气计量比和阴、阳极压强差为控制目标,设计了基于神经网络的模型预测控制算法。首先基于MATLAB/Simulink搭建了面向控制的燃料电池集总参数机理模型,通过实验验证了模型的可靠性;然后通过免疫遗传算法优化神经网络的学习过程,实现了对燃料电池系统状态的精确拟合与预测;最后,将离线训练的神经网络应用于模型预测控制器,并验证了控制算法的有效性。

基于人工免疫网络的短期负荷预测模型

为了克服传统神经网络预测方法在网络结构设计、学习算法和收敛效果等方面存在的缺陷,通过借鉴免疫网络调节与免疫规划,该文提出了一种基于人工免疫网络的短期负荷预测模型。在人工免疫网络的设计中,创造性地融入了免疫调节原理,利用免疫规划来进化网络结构,采用了新的个体编码方式、神经元适应度函数和自适应混沌变异算子,通过免疫规划进行网络结构的设计,并结合免疫网络调节的进化算法进行网络的学习。电力系统短期负荷预测的计算实例表明,基于人工免疫网络的负荷预测方法与传统神经网络预测方法相比,具有较强的自适应能力和较好的效果。

人工免疫系统研究的新进展

综述了人工免疫系统的最新研究成果。首先简述了生物免疫系统的信息处理机理 ,其次介绍了独特型人工免疫网络、多值免疫网络、免疫联想记忆等人工免疫模型 ,以及反向选择、免疫遗传、克隆选择等五类免疫学习算法 ,最后介绍了人工免疫系统的应用 。

新型RBF神经网络及在热工过程建模中的应用

文中提出了一种基于免疫原理的新型径向基函数(RBF—Radial Basis Function)神经网络模型。该模型利用人工免疫系统的记忆、学习和自组织调节原理,进行RBF神经网络隐层中心数量和位置的选择,并采用递推最小二乘算法确定网络输出层的权值。将这种新型的RBF神经网络应用于建立热工过程的非线性模型。仿真研究表明,这种建模方法不仅计算量较小,而且精度高,并有较好的泛化能力。

人工免疫系统的应用与发展

在现代信息科学和生命科学相互交叉渗透的研究领域,由生物免疫系统启发的人工免疫系统(AIS)是继脑神经系统(神经网络)和遗传系统(进化计算)之后的又一个研究热点。该文首先简要介绍了生物免疫系统的特点,然后系统综述了国内外对人工免疫系统的最新研究和应用成果,最后展望了人工免疫系统进一步的研究方向。

人工免疫系统进展与展望

本文评述人工免疫系统的历史、研究现状和进一步发展的方向 .着重论述人工免疫系统的机理、算法和应用 ,总结了免疫算法的一般步骤 ,比较了其与神经网络、进化计算以及一般确定性优化算法的异同 .在总结人工免疫系统存在问题的基础上 。

基于危险理论的网络风险评估模型

为有效评估网络信息系统的网络风险,提出了一种新的基于危险理论的风险评估模型(DTREM)。在给出网络活动中自体、非自体、免疫细胞的定义,建立由记忆检测器、成熟检测器、未成熟检测器集合构成的入侵检测子模型后,进一步给出了基于危险理论的网络风险定量计算子模型。利用该模型,可以实时定量地计算出主机和网络当前所面临攻击的各类攻击和总体网络风险强度,理论分析和实验结果均表明,该模型为实时网络安全风险评估提供了一种有效的新途径。

基于多值免疫网络的多光谱遥感影像分类

提出了一种基于多值免疫网络的多光谱遥感影像分类方法.该方法用选取的训练样本对多值免疫网络进行网络训练,得到具有记忆功能的免疫网络结构,然后利用多值免疫网络对多光谱遥感影像进行分类.实验结果证明,该算法分类精度上优于传统的分类方法,总精度和Kappa系数分别达到了88.84%和0.8605,因而具有实用价值.

改进人工免疫网络算法在电力短期负荷预测中的应用

针对现有的短期负荷预测方法易陷入局部极值以及预测精度不高等缺陷,文中提出了一种基于改进免疫算法优化BP神经网络的短期智能负荷预测方法。通过利用改进的矢量距优化免疫网络,从而达到优化网络的目的。融入免疫调节原理,引入抗体浓度的概率选择式,采用自适应变化策略重新设计变异算子,利用新的变异尺度设计种群抗体,采用新的神经元适应度函数,并结合免疫网络调节的进化算法进行网络学习。实例分析表明,基于改进免疫网络优化的BP网络短期负荷预测算法比混沌算法优化BP网络算法精度更高,更具实用性。