UNet Based onMulti-Object Segmentation and Convolution Neural Network for Object Recognition

The recent advancements in vision technology have had a significant impact on our ability to identify multiple objects and understand complex scenes.Various technologies,such as augmented reality-driven scene integration,robotic navigation,autonomous driving,and guided tour systems,heavily rely on this type of scene comprehension.This paper presents a novel segmentation approach based on the UNet network model,aimed at recognizing multiple objects within an image.The methodology begins with the acquisition and preprocessing of the image,followed by segmentation using the fine-tuned UNet architecture.Afterward,we use an annotation tool to accurately label the segmented regions.Upon labeling,significant features are extracted from these segmented objects,encompassing KAZE(Accelerated Segmentation and Extraction)features,energy-based edge detection,frequency-based,and blob characteristics.For the classification stage,a convolution neural network(CNN)is employed.This comprehensive methodology demonstrates a robust framework for achieving accurate and efficient recognition of multiple objects in images.The experimental results,which include complex object datasets like MSRC-v2 and PASCAL-VOC12,have been documented.After analyzing the experimental results,it was found that the PASCAL-VOC12 dataset achieved an accuracy rate of 95%,while the MSRC-v2 dataset achieved an accuracy of 89%.The evaluation performed on these diverse datasets highlights a notably impressive level of performance.

Cardiac arrhythmias detection in an ECG beat signal using fast fourier transform and artificial neural network

Cardiac Arrhythmias shows a condition of abnor-mal electrical activity in the heart which is a threat to humans. This paper presents a method to analyze electrocardiogram (ECG) signal, extract the fea-tures, for the classification of heart beats according to different arrhythmias. Data were obtained from 40 records of the MIT-BIH arrhythmia database (only one lead). Cardiac arrhythmias which are found are Tachycardia, Bradycardia, Supraventricular Tachycardia, Incomplete Bundle Branch Block, Bundle Branch Block, Ventricular Tachycardia. A learning dataset for the neural network was obtained from a twenty records set which were manually classified using MIT-BIH Arrhythmia Database Directory and docu- mentation, taking advantage of the professional experience of a cardiologist. Fast Fourier transforms are used to identify the peaks in the ECG signal and then Neural Networks are applied to identify the diseases. Levenberg Marquardt Back-Propagation algorithm is used to train the network. The results obtained have better efficiency then the previously proposed methods.

基于Fourier-TOuNN的鲁棒性拓扑优化设计

为推广拓扑优化设计方法的工程应用,需要在设计过程中考虑结构鲁棒性以应对实际工程荷载的随机性。本文基于神经网络提出了鲁棒性结构拓扑优化设计的高效方法。该方法通过优化Fourier-TOuNN神经网络的权值更新描述结构拓扑的密度变量,并引入随机荷载下结构柔顺度平均值和标准差的加权总和作为目标函数,从而定义了随机荷载下的结构鲁棒性优化问题。利用神经网络的自动反向微分功能,实现了优化过程中灵敏度的直接求解。借助Fourier-TOuNN细部尺寸可控特性,可在结构中生成细小支撑以抵抗随机荷载。数值算例表明,采用本文提出的方法可以高效地获得鲁棒性稳健的优化设计结果。

基于改进FNN-CCC的双伺服压力机同步控制策略研究

目的改善双伺服压力机同步控制策略的动态响应性能和鲁棒性,提升双伺服压力机的单轴跟踪精度和双轴同步精度,实现成形过程的高精度位置控制。方法建立双伺服压力机驱动系统数学模型,分析系统同步误差来源,结合模糊神经网络单轴控制算法,引入迭代学习律,设计一种改进模糊神经网络-交叉耦合(FNN-CCC)同步控制器。基于系统控制模型进行单轴阶跃响应特性与双轴正弦跟随特性仿真,搭建嵌入式双伺服压力机驱动系统试验平台,在偏载干扰条件下进行双轴同步控制试验,验证所提出理论的有效性。结果仿真结果表明,与模糊控制算法和BP神经网络控制算法相比,该控制器单轴控制算法的超调量分别减少了11.5%和25.5%,调节时间分别减少了48.8%和34.4%,具有更好的动态响应性能。与原控制器相比,改进后的交叉耦合同步控制器最大双轴同步误差降低了65.7%,同步控制精度有所提高。试验结果表明,与传统PID-交叉耦合控制器相比,改进的FNN-CCC控制器有更好的控制性能,在热冲压合模成形阶段,单轴跟踪误差分别减小了81.8%和75.0%,双轴同步误差减小了69.2%。结论所提出的同步控制策略在偏载干扰条件下具有较好的动态响应性能和鲁棒性,能够使同步误差快速收敛,提高了双伺服压力机驱动系统的单轴跟踪精度和双轴同步控制精度,实现了对双伺服压力机的高精度控制。

A Precise Algorithm for Non-Integer Harmonics Analysis Based on FFT and Neural Network

By means of an arificial neural network (ANN) model, higher measurement accuracy of integer harmonics can be obtained. Combining the windowed fast Fourier transform (FFT) algorithm with the improved ANN model, we present a new precise algorithm for non-integer harmonics analysis. According to the result obtained from the Hanning-windowed FFT algorithm, we choose the initial values of orders of harmonics for the neural network. Through such processing, the time of iterations is shortened and the convergence rate of neural network is raised thereby. The simulation results show that close non-integer harmonics can be separated from a signal with higher accuracy and better real-time by using the algorithm presented in the paper. Key words fast Fourier transform (FFT) - artificial neural network (ANN) - Hanning-window - harmonics analysis CLC number TM 935 Foundation item: Supported by the Teaching and Research Award Program for Outstanding Young Teachers in Higher Education Institutions of China (2001-182) and the Science Foundation of Naval University of Engineering(HGDJJ03001).Biography: WANG Gong-bao (1962-), male, Professor, research direction: artificial neural network, wavelet analysis and their applications to signal processing in electric power systems.

Parameter Optimization of Interval Type-2 Fuzzy Neural Networks Based on PSO and BBBC Methods

Interval type-2 fuzzy neural networks(IT2FNNs)can be seen as the hybridization of interval type-2 fuzzy systems(IT2FSs) and neural networks(NNs). Thus, they naturally inherit the merits of both IT2 FSs and NNs. Although IT2 FNNs have more advantages in processing uncertain, incomplete, or imprecise information compared to their type-1 counterparts, a large number of parameters need to be tuned in the IT2 FNNs,which increases the difficulties of their design. In this paper,big bang-big crunch(BBBC) optimization and particle swarm optimization(PSO) are applied in the parameter optimization for Takagi-Sugeno-Kang(TSK) type IT2 FNNs. The employment of the BBBC and PSO strategies can eliminate the need of backpropagation computation. The computing problem is converted to a simple feed-forward IT2 FNNs learning. The adoption of the BBBC or the PSO will not only simplify the design of the IT2 FNNs, but will also increase identification accuracy when compared with present methods. The proposed optimization based strategies are tested with three types of interval type-2 fuzzy membership functions(IT2FMFs) and deployed on three typical identification models. Simulation results certify the effectiveness of the proposed parameter optimization methods for the IT2 FNNs.

Research on Prediction of Red Tide Based on Fuzzy Neural Network

In this paper, a four-layer fuzzy neural network using the Back Propagation （BP） Algorithm and the fuzzy logic was built to study the nonlinear relationships between different physical -chemical factors and the denseness of red tide algae, and to anticipate the denseness of the red tide algae. For the first time, the fuzzy neural network technology was applied to research the prediction of red tide. Compared with BP network and RBF network, the outcome of this method is better.

Artificial neural network approach to assess selective flocculation on hematite and kaolinite

Because of the current depletion of high grade reserves, beneficiation of low grade ore, tailings produced and tailings stored in tailing ponds is needed to fulfill the market demand. Selective flocculation is one alternative process that could be used for the beneficiation of ultra-fine material. This process has not been extensively used commercially because of its complex dependency on process parameters. In this paper, a selective flocculation process, using synthetic mixtures of hematite and kaolinite in different ratios, was attempted, and the ad-sorption mechanism was investigated by Fourier transform infrared （FTIR） spectroscopy. A three-layer artificial neural network （ANN） model （4？4？3） was used to predict the separation performance of the process in terms of grade, Fe recovery, and separation efficiency. The model values were in good agreement with experimental values.

New shape clustering method based on contour DFT descriptor and modified SOFM neural network

A contour shape descriptor based on discrete Fourier transform （DFT） and a K-means al- gorithm modified self-organizing feature map （SOFM） neural network are established for shape clus- tering. The given shape is first sampled uniformly in the polar coordinate. Then the discrete series is transformed to frequency domain and constructed to a shape characteristics vector. Firstly, sample set is roughly clustered using SOFM neural network to reduce the scale of samples. K-means algo- rithm is then applied to improve the performance of SOFM neural network and process the accurate clustering. K-means algorithm also increases the controllability of the clustering. The K-means algo- rithm modified SOFM neural network is used to cluster the shape characteristics vectors which is previously constructed. With leaf shapes as an example, the simulation results show that this method is effective to cluster the contour shapes.

Using Neural Networks to Predict Secondary Structure for Protein Folding

Protein Secondary Structure Prediction (PSSP) is considered as one of the major challenging tasks in bioinformatics, so many solutions have been proposed to solve that problem via trying to achieve more accurate prediction results. The goal of this paper is to develop and implement an intelligent based system to predict secondary structure of a protein from its primary amino acid sequence by using five models of Neural Network (NN). These models are Feed Forward Neural Network (FNN), Learning Vector Quantization (LVQ), Probabilistic Neural Network (PNN), Convolutional Neural Network (CNN), and CNN Fine Tuning for PSSP. To evaluate our approaches two datasets have been used. The first one contains 114 protein samples, and the second one contains 1845 protein samples.

Adaptive Iterative Learning Control for Nonlinear Time-delay Systems with Periodic Disturbances Using FSE-neural Network

An adaptive iterative learning control scheme is presented for a class of strict-feedback nonlinear time-delay systems, with unknown nonlinearly parameterised and time-varying disturbed functions of known periods. Radial basis function neural network and Fourier series expansion （FSE） are combined into a new function approximator to model each suitable disturbed function in systems. The requirement of the traditional iterative learning control algorithm on the nonlinear functions （such as global Lipschitz condition） is relaxed. Furthermore, by using appropriate Lyapunov-Krasovskii functionals, all signs in the closed loop system are guaranteed to be semiglobally uniformly ultimately bounded, and the output of the system is proved to converge to the desired trajectory. A simulation example is provided to illustrate the effectiveness of the control scheme.

Feedforward Neural Network for joint inversion of geophysical data to identify geothermal sweet spots in Gandhar,Gujarat,India

Artificial Neural Networks(ANNs)are used in numerous engineering and scientific disciplines as an automated approach to resolve a number of problems.However,to build an artificial neural network that is prudent enough to rely on,vast quantities of relevant data have to be fed.In this study,we analysed the scope of artificial neural networks in geothermal reservoir architecture.In particular,we attempted to solve joint inversion problem through Feedforward Neural Network(FNN)technique.In order to identify geothermal sweet spots in the subsurface,an extensive geophysical studies were conducted in Gandhar area of Gujarat,India.The data were acquired along six profile lines for gravity,magnetics and magnetotellurics.Initially low velocity zone was identified using refraction seismic technique in order to set a common datum level for other potential data.The depth of low velocity zone in Gandhar was identified at 11 m.The FNN backpropagation method was applied to gain the global minima of the data space and model space as desired.The input dataset fed to the inversion algorithm in the form of gravity,magnetic susceptibility and resistivity helped to predict the suitable model after network training in multiple steps.The joint inversion of data is conducive to understanding the subsurface geological and lithological features along with probable geothermal sweet spots.The results of this study show the geothermal sweet spots at depth ranging from 200 m to 300 m.The results from our study can be used for targeted zones for geothermal water exploitation.

The Neural Network Model of Sinusoid Activation Transfer Function

A new type of neural network is described, which is basing on Fourier series, and the activation transfer function in its neuron model is sinusoid, ft can approximate to any function, which is continuum in every segment, with any precision with by layers only. We also provide the computer approach emulation results of different kinds of static function.

Application of fuzzy neural network to the nuclear power plant in process fault diagnosis

The fuzzy logic and neural networks are combined in this paper, setting upthe fuzzy neural network (FNN ) ; meanwhile, the distinct differences and connections between thefuzzy logic and neural network are compared. Furthermore, the algorithm and structure of the FNN areintroduced. In order to diagnose the faults of nuclear power plant, the FNN is applied to thenuclear power planl, and the intelligence fault diagnostic system of the nuclear power plant isbuilt based on the FNN . The fault symptoms and the possibility of the inverted U-tube breakaccident of steam generator are discussed. In order to test the system' s validity, the invertedU-tube break accident of steam generator is used as an example and many simulation experiments areperformed. The test result shows that the FNN can identify the fault.

Application of variable-filtrating technique on fuzzy-reasoning neural network system predicting BOF end-point carbon content

Artificial intelligence techniques have been used to predict basic oxygen furnace(BOF) end-points. However,the main challenge is to effectively reduce the input nodes as too many input nodes in neural network increase complexity,decrease accuracy and slow down the training speed of the network.Simply picking-up variables as input usually influence validity of model.It is quite necessary to develop an effective method to reduce the number of input nodes whereby to simplify the network and improve model performance.In this study,a variable-filtrating technique combining both metallurgical mechanism model and partial least-squares(PLS ) regression method has been proposed by taking the advantages of both of them,i.e.qualitive and quantative relationships between variables respectively.Accordingly,a fuzzy-reasoning neural network(FNN) prediction model for basic oxygen furnace(BOF) end-point carbon content based on this technique has been developed.The prediction results showed that this model can effectively improve the hit rate of end-point carbon content and increase network training speed.The successful hit rate of the model can reach up to 94.12%with about 0.02% error range.

Design and Implementation of Computer-Aid Garment Coordination Tool Using Fuzzy Neural Network

By modeling the decision-making process of garment coordination of fashion designers, a kind of computer-aid garment coordination using fuzzy neural network was propesed. The Takagi Sugeno Fuzzy Neural Network （TSFNN） is used to learn the knowledge and rules of fashion designers on garment coordination and calculate the garment coordination satisfaction index （GCSI）. The implementation of the computer-aid garment coordination tool is divided into two stages. The first stage is to acquire the knowledge of garment coordination. The second stage is to train and use the fuzzy neural network to conduct garment coordination. Three layers structure were also discussed for developing the system. By applying the computer-aid garment coordination tool into a real fushionretailing store, the experimental results show the system pexforms well with choosing a suitable value for screening out the satisfaction coordination pairs.

A Fuzzy Neural Network Model of Linguistic Dynamic Systems Based on Computing with Words

Linguistic dynamic systems(LDS)are dynamic processes involving computing with words(CW)for modeling and analysis of complex systems.In this paper,a fuzzy neural network(FNN)structure of LDS was proposed.In addition,an improved nonlinear particle swarm optimization was employed for training FNN.The experiment results on logistics formulation demonstrates the feasibility and the efficiency of this FNN model.

基于短时傅里叶变换和深度网络的模块化多电平换流器子模块IGBT开路故障诊断

针对现有模块化多电平换流器(MMC)子模块故障诊断过程中所需传感器较多、测量干扰较大等问题,提出一种基于深度学习的MMC子模块IGBT开路故障诊断方法。在对MMC子模块开路故障特征进行分析的基础上,利用短时傅里叶变换(STFT)提取桥臂电压信号的谐波分量信息作为故障诊断所需的特征参数。将所得到的特征参数进行处理后构建故障诊断样本,在通过深度置信网络实现故障类型快速检测的基础上,依据不同故障类型,构建多个基于卷积神经网络的故障定位网络,进而实现开路故障的检测与定位。通过129电平的MMC系统仿真模型和降功率的MMC实验系统搭建,对该文所提方法进行了验证。仿真和实验结果表明,所提故障诊断方法可以在减少传感器数量的基础上实现子模块开路故障的诊断,提高系统的可靠性。

融合CNN和ViT的声信号轴承故障诊断方法

针对轴承故障诊断任务数据量少、故障信号非平稳等特点,提出一种短时傅里叶变换、卷积神经网络和视觉转换器相结合的轴承故障诊断方法。首先,利用短时傅里叶变换将原始声信号转换为包含时序信息和频率信息的时频图像。其次,将时频图像作为卷积神经网络的输入,用于隐式提取图像的深层特征,其输出作为视觉转换器的输入。视觉转换器用于提取信号的时间序列信息。并在输出层利用Softmax函数实现故障模式的识别。试验结果表明,该方法对于轴承故障诊断准确率较高。为了更好解释和优化提出的轴承故障诊断方法,利用t-分布领域嵌入算法对分类特征进行了可视化展示。

基于时频图与视觉Transformer的滚动轴承智能故障诊断方法

基于循环神经网络的故障诊断方法在计算过程中难以保存间隔时间过长的信息且无法并行计算,在大型数据建模方面存在不足,为提高轴承故障诊断工作的效率及准确性,提出了一种基于短时傅里叶变换时频图与视觉Transformer(ViT)的轴承故障诊断方法:通过短时傅里叶变换将原始振动信号转换为二维时频图像,再将时频图作为特征图输入ViT网络中进行训练,详细分析网络参数对故障诊断性能和计算复杂度的影响,构建最优模型结构,最终实现轴承的故障诊断。采用凯斯西储大学和江南大学轴承数据对模型进行验证,结果表明该模型可以有效结合短时傅里叶变换在处理时变信号方面的优势和ViT网络强大的图像分类能力,具有更高的诊断精度和更好的泛化性、通用性。