Genetically optimized fuzzy polynomial neural networks with fuzzy set-based polynomial neurons

In this paper, we propose and investigate a new category of neurofuzzy networks—fuzzy polynomial neural networks (FPNN) endowed with fuzzy set-based polynomial neurons (FSPNs) We develop a comprehensive design methodology involving mechanisms of genetic optimization, and genetic algorithms (GAs) in particular. The conventional FPNNs developed so far are based on the mechanisms of self-organization, fuzzy neurocomputing, and evolutionary optimization. The design of the network exploits the FSPNs as well as the extended group method of data handling (GMDH). Let us stress that in the previous development strategies some essential parameters of the networks (such as the number of input variables, the order of the polynomial, the number of membership functions, and a collection of the specific subset of input variables) being available within the network are provided by the designer in advance and kept fixed throughout the overall development process. This restriction may hamper a possibility of developing an optimal architecture of the model. The design proposed in this study addresses this issue. The augmented and genetically developed FPNN (gFPNN) results in a structurally optimized structure and comes with a higher level of flexibility in comparison to the one we encounter in the conventional FPNNs. The GA-based design procedure being applied at each layer of the FPNN leads to the selection of the most suitable nodes (or FSPNs) available within the FPNN. In the sequel, two general optimization mechanisms are explored. First, the structural optimization is realized via GAs whereas the ensuing detailed parametric optimization is carried out in the setting of a standard least square method-based learning. The performance of the gFPNN is quantified through experimentation in which we use a number of modeling benchmarks—synthetic and experimental data being commonly used in fuzzy or neurofuzzy modeling. The obtained results demonstrate the superiority of the proposed networks over the models existing in the references.     

基于DBSCAN-FPN的输电线路螺栓缺销检测方法

螺栓作为输电线路上数量最大的紧固件,其缺陷检测是输电线路巡检工作中的一项重要内容。针对螺栓缺销为小目标,其定位困难、特征难提取的问题,提出一种基于DBSCAN算法与FPN模型相结合的螺栓缺销检测方法。首先,利用FPN模型定位螺栓缺销目标区域,同时基于DBSCAN聚类算法对具有相同形态结构的区域进行聚类;然后,改进FPN模型：基于螺栓先验知识,利用卷积网络实现自底向上的特征提取,采用双线性插值方法将特征的高层语义信息自顶向下地传递到各个层级,通过卷积滤波方法横向加强高层语义特征与高分辨率特征的融合信息,获得更优化的螺栓缺销特征表达;利用改进FPN模型实现螺栓缺销的初步检测;最后,采用DBSCAN聚类算法对初步检测结果进行误检甄别,实现了螺栓缺销的精确检测。实验结果表明,DBSCAN-FPN在自建数据集上的检测精度达到76.23%,检测效果优于FPN、R-FCN和Faster R-CNN。所提方法可以有效提高螺栓缺销检测精度,对输电线路运维有实际意义。     

基于结果反馈的模糊Petri网学习算法

针对模糊Petri网模型的复杂结构,在不增加虚库所和虚变迁的情况下改进了模糊Petri网分层算法,从而简化模糊Petri网学习和训练方法。为提高收敛速率,本文从一个全新的角度考虑模糊Petri网的学习和训练,提出了基于结果反馈的模糊Petri网学习的新算法（FBFPN）。该算法通过对纯网进行层次式分层及建立变迁点燃的近似连续函数后,调整权值、变迁的阈值、变迁的可信度的同时又调整输入矢量的多重作用来最小化误差函数。仿真结果分析表明,该算法具有良好的学习效率和泛化能力。     

基于FPN—RSVM的电梯故障诊断方法

模糊Petri网(FPN)作为产生式规则系统的建模、知识表示和诊断推理的工具,在故障诊断领域得到了广泛的应用。但调整学习机制的缺乏使其不能处理专家系统的变化。使用回归型支持向量机(RSVM)对电梯在不同运行状态下的专家诊断数据进行训练,可以得到与电梯状态相对应的FPN诊断网络权值。故障发生时,根据电梯运行状态选择相应的权值,可以使得诊断结果更加准确,更加符合实际情况。     

复杂交通环境下多层交叉融合多目标检测

针对目前复杂交通环境下还存在多目标检测精度和速度不高等问题,以特征金字塔网络(Feature Pyramid Network, FPN)为基础,提出了一种多层融合多目标检测与识别算法,以提高目标检测精度和网络泛化能力。首先,采用ResNet101的五层架构将空间分辨率上采样2倍构建自上而下的特征图,按照元素相加的方式将上采样图和自下而上的特征图合并,并构建一个融合高层语义信息与低层几何信息的特征层;然后,根据BBox回归存在训练样本不平衡问题,选择Efficient IOU Loss损失函数并结合Focal Loss提出一种改进Focal EIOU Loss;最后,充分考虑复杂交通环境下的实际情况,进行人工标注混合数据集进行训练。该模型在KITTI测试集上的平均检测精度和速度比FPN分别提升了2.4%和5 frame/s,在Cityscale测试集上平均检测精度和速度比FPN提升了1.9%和4 frame/s。     

基于改进胶囊网络的运维知识库故障分类方法

针对传统运维知识库不具备图像故障现象识别能力,无法处理非结构化数据的问题,基于深度学习的故障分类网络,提出改进胶囊网络特征提取结构的Caps-DRFN算法,实现机电设备运维图像自动分类。首先,针对运维图像存在的多噪声问题,引入深度残差收缩网络(deep residual shrinkage networks, DRSN)提高模型在含噪声数据上的特征提取效果。然后,针对实际拍摄的运维图像多尺度问题,结合FPN(feature pyramid networks)算法,实现图像多尺度特征融合提高模型分类准确率。最后,利用胶囊结构构建向量神经元,通过动态路由的特征传递方式,得到分类结构数字胶囊,实现机电设备故障分类。实验结果表明,相较于传统胶囊网络算法,提出的基于特征融合的Caps-DRFN算法准确率提高了15%且有着更强的鲁棒性。     

一种实时多尺度目标检测识别算法

无人机目标检测与识别任务中,目标随着飞行高度的改变尺寸发生显著变化。常规目标检测模型中,获取的小目标细节信息有限,检测精度较低;而适用于小目标的实时检测模型往往容易丢失大目标的背景信息,降低大目标的检测精度。针对以上多尺度目标检测识别任务难点,提出一种基于改进特征金字塔网络(Feature Pyramid Network, FPN)结构的实时多尺度目标检测识别模型。该模型通过增加特征金字塔层级覆盖更广的目标尺度,获取更为丰富的目标信息;同时,利用跨连接增加不同尺度特征融合的多样性,降低特征传导距离,保留更加完整的尺度特征来提高模型检测识别多尺度目标的性能。通过实验发现,相比于原始网络结构和相同特征层级的四层特征金字塔结构,加入改进特征金字塔结构的多尺度目标检测模型识别性能得到了提升。     

基于改进的SSD监理目标检测研究

针对装饰装修工程中由人工验收带来的诸多问题,文中提出了一种改进的SSD算法并将其应用于监理工作来代替人工验收,推动智能监理的实现。由于SSD算法存在对同一目标复检以及小目标检测效果欠佳等问题,故文中利用DPN网络替换基础特征提取网络VGG16。DPN结合了Resnet和Densenet的优点,具有更好的特征提取能力。通过加权FPN融合特征图,突出不同层特征图的贡献,丰富用于预测的特征图语义。利用深度可分离卷积降低模型的参数量,提高算法的推理速度。实验对比发现,改进后模型的平均精度提升了3.47%,对小数目检测平均精度的提升可达15%,证明新模型在监理目标检测任务中效果良好。     

Designing genetic regulatory networks using fuzzy Petri nets approach

In this paper, we have successfully presented a fuzzy Petri net (FPN) model to design the genetic regulatory network. Based on the FPN model, an efficient algorithm is proposed to automatically reason about imprecise and fuzzy information. By using the reasoning algorithm for the FPN, we present an alternative approach that is more promising than the fuzzy logic. The proposed FPN approach offers more flexible reasoning capability because it is able to obtain results with fuzzy intervals rather than point values. In this paper, a novel model with a new concept of hidden fuzzy transition (HFT) to design the genetic regulatory network is developed. We have built the FPN model and classified the input data in terms of time point and obtained the output data, so the system can be viewed as the two-input and one output system. This method eliminates possible false predictions from the classical fuzzy model thereby allowing a wider search space for inferring regulatory relationship. The experimental results show the proposed approach is feasible and acceptable to design the genetic regulatory network and investigate the dynamical behaviors of gene network.     

CMOS图像传感器关键技术及其新进展

阐述了CMOS图像传感器芯片的基本现状,给出了目前应用在民用相机上存在的固定图形噪声、暗电流噪声以及像素间串扰等问题的解决方法与对策。并通过对CMOS图像传感器的新技术进行探讨,预测了CMOS图像传感器的应用前景与发展前途。