三维谱特征下的汽车尾气评估方法

针对汽车尾气排放的非线性、时变性问题,提出一种三维谱特征下的汽车尾气评估方法。该方法利用频谱分析的原理对汽车尾气进行时频转换,得到尾气的三维谱特征。这些三维谱特征作为输入被提交给径向基神经网络,在K均值聚类算法的驱动下,径向基神经网络完成训练与测试,实现对三维谱特征的分类,从而评估相应的汽车尾气排放水平。数值实验结果表明,提出的汽车尾气评估方法具有较高的准确性。     

Fuzzy adaptive tracking control for MIMO nonlinear time-varying delay systems

A fuzzy adaptive tracking control scheme is proposed for MIMO nonlinear time-varying delay systems. A fuzzy T-S model-based adaptive time-varying delay fuzzy logic systems is developed to approximate the unknown nonlinear time-varying delay functions.Thus, the modeling to nonlinear systems is implemented. The update laws for parameters of the fuzzy logic systems are derived by the tracking error. A Hcompensator is designed to eliminate fuzzy approximation errors and external disturbances. Based on Lyapunov stability theorem, the proposed control scheme can guarantee the stability of the closed loop systems and obtain anticipant Htracking performance as well. Simulation results of the manipulator demonstrate the effectiveness of the control scheme.     

Improved Ellipsoidal unit Neural Networks and Its Applications in CSTR

To overcome the limitations of the standard ellipsoidal unit neural networks, some new approaches used in ellipsoidal unit neural networks have been proposed. These new approaches address three main issues: firstly, to understand better and represent the nature of fault classification boundaries; secondly, to determine the network structure without the usual trial and error schemes; lastly, to avoid erroneous generalizations. The application in CSTR shows that the ellipsoidal unit networks can possess arbitrary nonlinear classifying ability, nonlinear interfacial describing ability, and obtain accurate and efficient diagnosis results.     

Vibration characteristics of face-gear transmission system with parametric excitation

In order to study the influences of different factors on nonlinear dynamics of face-gear transmission system, a nonlinear dynamic model is presented. This model includes backlash, general transmission error, time-varying meshing stiffness, meshing damping, bearings and external load, etc. Based on numerical analysis theory of nonlinear dynamics, the equations are solved and the bifurcation characteristics with different parameters are obtained. And the calculation results show that increasing the backlash and time-varying meshing stiffness will augment the dynamic load of system, but increasing the mesh damping will reduce the dynamic load effectively.     

The Research of Using BP Neural Network in Prediction of Woods Fire Insurance Grade

The relation between circumfluence index and woods fire.index is described. BP neural network is trained by the known circumfluence index and woods fire index. The restricted feedback and stimulant feed forwardcalculation are to realize self-organizational learningof BP Network.Thereby the nonlinear mapping between circumfluence index and woods fire.index isperformed. With this network, characteristic choice can be carriedout and the accurate prediction of woods fire insurance grade can be given. Its accurate prediction rate is 89%in spring and 82%in summer. The correctnessand feasibilityare illustratedwith the simulate results.     

Multiple Turbo Codes for Satellite Communications

Satellite communication channel has the characteristic of power limited, band abundance, and powerful error correcting ability FEC scheme is needed. For its near Shannon limit error correcting capability, Turbo codes may be suitable for satellite communication systems. The multiple Turbo codes can utilize channel capacity sufficiency and overcome the disadvantages of power limited for satellite communication systems. Force on the needed for satellite communication systems, multiple Turbo codes is adopted and server decoding schemes were proposed with simulation. Simulation results show that, the schemes proposed can get near channel capacity error correcting performance with short code block.     

Frequency Estimate Basing on Synergetic Neural Network

This text solves the basic and important frequency estimate problem in communication by synergetic neural network. Basing frequency estimate in satellite communication, concretely studies the frequency estimate about BPSK signal. When the large Doppler frequency is partitioned several smaller area, each order parameter stands for each area. Once get the winner of the order parameters, the smaller frequency is confirmed. The approximation of the adjoint vectors simplify the complexity of arithmetic, but the system error is caused. This method has rapid capture speed, and smaller cost by analyzing and simulation.     

Survey on Stability, Bifurcation and Chaos in Delayed Neural Networks

From the viewpoint of nonlinear dynamics, this review outlines the recent advances as well as some open problems in the study of neural networks with time delays, an important class of delayed systems in various neural network models. The survey includes three aspects as fellows: the dynamic features, available approaches and advances in research on most attractive problems. The evolution of a delayed neural network depends not only on the current state of the systems but also on previous ones. Hence, a delayed neural network should be modeled by a functional differential equation, the solution space of which is of infinite dimensions. Therefore, the dynamical behavior of delayed neural networks is very complex.     

Drift error compensation algorithm for fiber optic gyro based on T-S fuzzy modeling

A drift error nonlinear compensation algorithm for Fiber Optic Gyro (FOG) is presented based on T-S fuzzy model with the antecedent parameters identified by G-K clustering algorithm and the error model of T-S fuzzy model with the consequent parameters identified by least square algorithm. The computed results show that this model can compensate the original data effectively, while the error principles of FOG do not need to be understood well. Comparing with the original data, compensation with linear fitting and compensation with neural network, the absolute error of the proposed model reduces by 99%, 96% and 10%, respectively. The error variance reduces by 99%, 98% and 20%, respectively. The results indicate that this proposed algorithm can be simply operated with high precision and easy to realize in engineering.     

Research on Load Prediction Model of Air Condition System Based on Elman Neural Network

The load of air condition system is influenced by many factors, and they are variable and nonlinear, The relation between them is dynamic,It is impossible to forecaste the load of air condition syestem accurately by traditional method. But Recurrent Neural Network is able to reflect the dynamic lively and directly. Elman is one of the typical RNN. Based on the analysis as above, prediction model of air-condition system based on Elman neural network is established, and some prediction is done. The prediction accuracy of Elman neural network and BP neural network is compared, and the experiments show that the Elman neural network is efficiency and accuracy , so Elman neural network is a new and reliable method for predicting the load of air-condition system.     

Pulse Signal Analysis of Druggers with the Wavelet and Neural Network

The most significant difference between the human pulse signals collected from heroin druggers and healthy persons is at their amplitude waveforms as time functions. That is, the amplitude values and change rates of two types of signals, within a particular time range, appear different features. However, the partial components of the scaling and wavelet coefficients of the pulse signals obtained by using wavelet transform can reveal such key features. The pulse signals of 15 heroin druggers and 15 healthy persons are analyzed through using the muhiresolution analysis of wavelet transform. By using db2 orthogonal wavelet, every pulse signal is decomposed into three levels and the absolute values of the sixth component of scaling coefficients and the second component of the wavelet coefficients in the third level are combined to form a feature vector. A probabilistic neural network with good detection performance is successfully designed for automatically detecting 30 feature vectors. During the network design, 20 feature vectors are used as training samples. The remained 10 feature vectors are used as testing samples. Based on these steps, 15 heroin druggers and 15 healthy persons are all correctly identified. In other words, the detection rate arrives at 100%. druggers.     

Using BP Neural Network to Determine the Membership Function of the Input of Fuzzy Control System

While design the fuzzy controller, it is very important to determine the membership function of fuzzy variables.The data can be broadly classified as fuzzy sets by using the classification property of the BP neural network. The author selects a BP neural network with one hide layer and uses S function to the input and hide layer, and linear function to the output layer.Advanced BP algorithm isused to train the BP neural network in the environment of MATLAB . The nearer to the target values is the better the last output is.With the trained BP network , the membership values of the inputs can be got ten. This method has high rate and low error.     

Construction of Wavelet Neural Networks and its Robustness Analysis of Learning Algorithm

In this paper,a method of constructing the wavelet neural networks for nonlin-ear fun ctional approximation is discussed.The expon ential convergence of the training process andits robust stability to the noise perturbances and the network design errors are also proved.     

Application of Intelligent Neural Network on Prediction of Time Series Signal

The research of neural network has been maturated both in theory and practical application since 1980's, and also been employed into the prediction and analysis of nonlinear time series signal in the field of signal process system. Concerning with the problem of time series signal prediction based on traditional neural network, such as black box, poor accuracy, and facing the shortage of post knowledge, this paper presents a different neural network prediction model from the traditional ones, based on intelligent neural cell model and employing the iterative prediction method. Through the example on stock price prediction, the prediction accuracy and practical value are proved.     

Robust adaptive control for nonlinear chaotic systems

Tracking control of a nonlinear uncertain Chua's chaotic system is studied. With coordinate transform, Chua's chaotic system is transformed to a general form of a strict feedback control system. Combining the backstepping method with robust control technology, an adaptive parameter control law for a robust output feedback control scheme is developed for output tracking of nonlinear unknown systems. It is shown that the derived robust adaptive controller based on Lyapunov stability theory can guarantee global uniformly bounded ultimate property for all states of the closed loop system, and lead to tracking error decreasing at exponential speed. The simulation results show the effectiveness of the proposed approach.     

河南省粮食产量影响因素和预测方法

粮食生产是国民经济重要的组成部分,粮食产量对于保证我国的粮食安全具有重要意义。旨在提高粮食产量预测的科学性和准确性,在分析现有预测方法的基础上,文中将灰色理论和神经网络有机地结合在一起,通过灰色理论关联度分析,在众多影响粮食产量的因素中确定出主要的、客观的因素指标,通过这些指标利用人工神经网络具有的非线性建模和极高的拟合精度特点,应用到粮食产量预测中去。结果表明：人工神经网络预测的最大误差1.21%,平均误差0.63%。预测精度较高。为粮食产量预测提供了一种科学的、有效的预测方法。     

The application of rough sets-neural network theory to mine ventilation system evaluation

To solve the instability problem of established sample in the neural network evaluation method for mine ventilation system, a comprehensive evaluation of the ventilation system is carried out based on rough sets and BP neural networks. Taking the ventilation system of a mine as an example, the classification quality of raw data samples are tested by using rough set data analysis system. Then, based on artificial neural network theory, a rough sets-neural network evaluation model of a mine ventilation system is established and a new rough sets-neural network evaluation method of mine ventilation system is formed. The results show that, after the model validation of data and application, its theoretical evaluation results are in line with the actual situation, and the network total error is less than 0.004. It shows that the comprehensive evaluation method based on rough sets-neural networks has a good effect in evaluating mine ventilation system in practical application.     

基于主成分分析的BP神经网络估算藏北高寒草地覆盖度变化

藏北高寒牧区草地是中国高寒草地分布面积最大的地区。为了及时准确地获得该区域草地覆盖度的变化趋势,本研究利用多年气象数据、社会统计数据、GIMMS、MODIS两种归一化植被指数（NDVI）数据作为参数,构建 BP神经网络模型,估算2010—2014年藏北高寒草地年际变化趋势,并用主成分分析方法优化参数来改进模型。结果表明,① BP神经网络模型及其改进模型对藏北高寒草地覆盖度年际变化趋势与遥感值的相关系数为0.16、0.47,表明通过主成分分析优化参数后的BP神经网络模型具有较好的模拟效果。 ②两种BP神经网络估算的植被指数值与NDVI值平均误差率分别为2.36%、2.20%。均有较高的模拟精度。③从神经网络训练步数上看,BP神经网络结果训练收敛步长为5000,基于主成分分析的BP神经网络模型训练收敛步长为454,表明后者提高了计算效率,体现出良好的收敛性。因此,无论从年际变化趋势拟合程度、植被指数估算值精度、还是从计算效率来看,改进的BP神经网络模型对于估算藏北高寒草地覆盖度变化更加行之有效。     

基于人工神经网络理论的土壤水分预测研究

土壤水分含量是影响作物生长的重要因素,精确的预测技术对水资源的合理利用与管理具有重要的指导意义。利用人工神经网络理论,建立了以降水量、蒸发量、相对湿度和地下水埋深为输入因子,土壤水分含量为输出因子的预测模型,并对其预测精度进行了评价。结果表明,BP神经网络模型预测土壤含水率的最大误差为8.66%,平均误差为4.27%,预测精度达到0.989。模型具有较高的预测精度,其结果可为制定合理的水资源调配方案和调度计划提供科学依据。     

神经网络自变量输入模式的视电阻率求解算法

根据瞬变电磁场理论公式中的响应和自变量之间的关系特点,提出用非线性方程模式的BP神经网络求解电阻率。通过构造单输入单输出网络结构,建立以不同时间点上的电流归一化的感应电压值为输入、视电阻率值为输出的神经网络,来拟合瞬变电磁场的二次涡流曲线。利用数值方法计算出的数据验证该训练网络的精确性,比较了不同算法对训练精度和收敛速度产生的影响。以重庆大学某处的防空洞探测实验为例验证了该算法的有效性,该算法避开具体的复杂电磁场计算或数值反问题计算,从而实现电阻率快速计算,为快速成像准备必要条件。