基于混沌神经网络理论的城市地面沉降量预测模型

通过分析城市地面沉降量时间序列的非线性动力学系统,认为该时间序列具有混沌特性。在此基础上,通过相空间重构的方法建立了用于城市地面沉降量预测的混沌神经网络模型;并利用此模型对高桥地面沉降量进行了预测,并和实际监测沉降量进行了比较,最大绝对预测误差为1. 7,预测的平均误差为0. 0833,研究结果表明,应用混沌神经网络模型进行城市沉降预测是可行、精确的。     

滑坡时间预报的非平稳时间序列方法研究

本文基于边坡变形破坏的特点,对位移观测数据进行统计分析,利用非平稳时间序列理论就位移观测值建立模型进而作出预报。通过实例验证,该理论方法在边坡变形破坏时间预报中是有效可行的。     

混沌时间序列建模及预测

许多看似随机的时间序列可能是非线性确定系统的混沌行为所导致的结果。本文讨论了混沌时间序列的建模及预测方法,给出了各重要参数的选择取算法,并应用于实例,与传统的时间序列预测方法相比较,取得了精度更高的预测结果,从而为一类非线性时间序列提供了从数据采集识别到建模预测的完整技术。     

一类非线性时间序列的建模

在非线性科学的应用方面，时间序列分析和处理是一个典型问题。本文以太阳黑子为例，阐述了对混沌背景下的一类非线性时间序列。传统的AR模型的预测值可信度不高，而采用基于混沌吸引子的时间序列预测方法，可获得较好的预测效果。     

面向工程全局优化的混沌优化算法研究进展Research advances of chaos optimization algorithms for engineering global optimization

近年来,基于混沌的初值敏感性、伪随机性、遍历性以及自相似分形等非线性动力学特性所发展的混沌优化方法,是一种有潜力的工程全局优化新工具,已广泛应用于科学与工程技术的各学科领域。根据混沌优化方法的发展历程,以算法基本思想和工程应用研究状况为重点,评述了混沌神经网络优化方法、第一类混合混沌优化算法（基于混沌搜索）、第二类混合混沌优化算法（混沌序列代替随机序列）以及混沌分形优化四种主要混沌优化算法。混沌映射最早被引入神经网络,发展了混沌神经网络优化方法,可解决复杂的组合优化等全局优化问题。遗传算法及粒子群等启发式随机算法虽具全局搜索能力,但易出现早熟并陷入局部最优。然后,出现了混沌搜索的概念,研究者将其嵌入启发式算法建立了第一类混合混沌优化算法,可有效克服原启发式算法早熟收敛的缺点。随后,利用混沌映射产生的混沌序列代替启发式算法中的随机参数形成了第二类混合混沌优化算法。混合混沌优化算法有益于实现快速全局收敛和提高计算精度。最后,利用混沌分形特性,从分形理论出发提出一类新颖的混沌分形优化算法,可搜索到优化问题的所有全局最优解。此外,对混沌优化算法研究的几个发展方向进行了展望,诸如加强混沌优化算法的参数设计、处理大规模优化、多目标优化问题以及使用代理模型等。     

加权函数组合预测边坡变形模型的研究

边坡变形监测是边坡监测的主要内容之一,其变形预测问题是边坡工程中主要技术难题之一。考虑边坡位移变形预测模型的局限性,如神经网络预测方法需要大量的实测数据作为学习样本,灰色系统模型要求原始数据序列必须满足指数规律,且数据序列变化速度不能太快等。建立了边坡变形反向传播神经网络预测模型,同时给出了灰色GM(1,1)边坡预测模型。提出边坡的神经网络与灰色系统加权函数组合预测模型,采用动态规划解法,将原模型转化为多阶段决策问题,使组合预测误差的平方和最小,得到组合权重,这样得到的变形预测结果的精度将大大提高,弥补了单一方法的局限性,满足工程预测的需要。通过边坡实例加以验证,加权函数组合预测模型的预测结果精度有一定提高,能够与实际监测数据相吻合,达到准确预测的目的。     

面向工程全局优化的混沌优化算法研究进展

近年来,基于混沌的初值敏感性、伪随机性、遍历性以及自相似分形等非线性动力学特性所发展的混沌优化方法,是一种有潜力的工程全局优化新工具,已广泛应用于科学与工程技术的各学科领域。根据混沌优化方法的发展历程,以算法基本思想和工程应用研究状况为重点,评述了混沌神经网络优化方法、第一类混合混沌优化算法(基于混沌搜索)、第二类混合混沌优化算法(混沌序列代替随机序列)以及混沌分形优化四种主要混沌优化算法。混沌映射最早被引入神经网络,发展了混沌神经网络优化方法,可解决复杂的组合优化等全局优化问题。遗传算法及粒子群等启发式随机算法虽具全局搜索能力,但易出现早熟并陷入局部最优。然后,出现了混沌搜索的概念,研究者将其嵌入启发式算法建立了第一类混合混沌优化算法,可有效克服原启发式算法早熟收敛的缺点。随后,利用混沌映射产生的混沌序列代替启发式算法中的随机参数形成了第二类混合混沌优化算法。混合混沌优化算法有益于实现快速全局收敛和提高计算精度。最后,利用混沌分形特性,从分形理论出发提出一类新颖的混沌分形优化算法,可搜索到优化问题的所有全局最优解。此外,对混沌优化算法研究的几个发展方向进行了展望,诸如加强混沌优化算法的参数设计、处理大规模优化、多目标优化问题以及使用代理模型等。     

亚音速气流作用下薄板结构混沌运动的时滞反馈控制

研究了亚音速气流下非线性二维薄板结构在横向周期载荷作用下的混沌运动及控制问题。基于von Karman大变形板理论和分离变量法,建立了亚音速下薄板结构的运动控制方程。对于未控系统,采用Melnikov方法判断其混沌运动阈值,并用Runge-Kutta法进行数值验证。对处于混沌运动状态的系统,采用时滞反馈控制方法对混沌运动进行控制。结果表明,Melnikov方法可以有效地预测系统的混沌运动行为,时滞反馈控制方法可以有效地将系统的混沌运动转化为周期运动。     

异常天气条件下潮位过程神经网络补遗预测方法的研究

以异常天气条件下的实测潮位过程为研究对象,根据其为非平稳时间序列的特点,以人工神经网络BP算法作为预测工具,对潮位时间序列中缺失的数据进行补遗建立:差分方法人工神经网络模型;同一海域邻近潮汐测站潮位过程对应预测的去平均值法人工神经网络模型;对增减水现象潮汐过程预测的气象资料数据库人工神经网络模型。以实测资料验证上述方法的可行性,并取得了很好的预测结果。     

非线性函数的混沌优化方法比较研究

已有的混沌优化方法几乎都是利用Logistic映射作为混沌序列发生器，而Logistic映射产生的混沌序列的概率密度函数服从两头多、中间少的切比雪夫型分布，不利于搜索的效率和能力。为此，首先根据Logistie映射混沌轨道点密度函数的特点，建立改进的混沌-BFGS混合优化算法。之后，考虑到Kent映射混沌轨道点密度为均匀分布，建立了基于Kent映射的混沌-BFGS混合优化算法。然后对五种混合优化方法——不加改进的和改进的基于Logistic映射的混沌-BFGS法，基于Kent映射的混沌-BFGS法，Monte Carlo试验-BFGS法，网格-BFGS法进行了研究，分别对3个低维和2个高维非线性复杂测试函数进行优化计算，对它们的全局优化计算效率和寻优能力做了比较，并探讨了混合优化方法全局优化性能差异的原因。结果表明，混沌优化方法是与Monte Carlo方法类似的一种随机性试验优化方法。而且，这类优化方法的计算性能至少与以下因素有关：混沌／随机序列的统计性质，优化问题全局最优点位置。     

STUDY ON PREDICTION METHODS FOR DYNAMIC SYSTEMS OF NONLINEAR CHAOTIC TIME SERIES＊

The prediction methods for nonlinear dynamic systems which are decided by chaotic time series are mainly studied as well as structures of nonlinear self-related chaotic models and their dimensions. By combining neural networks and wavelet theories, the structures of wavelet transform neural networks were studied and also a wavelet neural networks learning method was given. Based on wavelet networks, a new method for parameter identification was suggested, which can be used selectively to extract different scales of frequency and time in time series in order to realize prediction of tendencies or details of original time series. Through pre-treatment and comparison of results before and after the treatment, several useful conclusions are reached:High accurate identification can be guaranteed by applying wavelet networks to identify parameters of self-related chaotic models and more valid prediction of the chaotic time series including noise can be achieved accordingly.     

Analysis and applied study of dynamic characteristics of chaotic repeller in complicated system

Introduction Inrecentyears,thestudyondynamicbehaviorofnonlinearsystemhasbecomeanactive subjectinnonlinearscience[1-12].Chaosisakindofcomplicatedandirregularbehaviorcseated bynonlinearsystem,suchirregularphenomenonexistsinnatureandsocietywidely.Itiswell_ known,timeserieswithcomplicatedphenomenonandbehaviorincludingchaosexistinvarious complicatedsystemsandinengineeringtechniques,suchsituationsareusuallytreatedeffectively withchaotictheoriesandmethods.Uptonowseveralmaturedstatisticindexesformeasu…     

STUDY ON THE PREDICTION METHOD OF LOW_DIMENSION TIME SERIES THAT ARISE FROM THE INTRINSIC NONLINEAR DYNAMIC

ＩｎｔｒｏｄｕｃｔｉｏｎＬｏｔｓｏｆｔｉｍｅｓｅｒｉｅｓｆｒｏｍｐｒａｃｔｉｃａｌｐｒｏｂｌｅｍｓｂｅｌｏｎｇｔｏｎｏｎｌｉｎｅａｒｃｈａｏｔｉｃｔｉｍｅｓｅｒｉｅｓ.Ｉｔｈａｓｂｅｅｎｐｒｏｖｅｄｉｎｐｒａｃｔｉｃｅｔｈａｔｔｈｅｌｉｎｅａｒｍｏｄｅｌｓｏｆｅｉｔｈｅｒｌｏｗｏｒｄｅｒｓｏｒｈｉｇｈｏｒｄｅｒｓｃａｎｎｏｔｂｅｕｓｅｄｔｏｄｅｓｃｒｉｂｅｎｏｎｌｉｎｅａｒｃｈａｏｔｉｃｔｉｍｅｓｅｒｉｅｓ.Ｈｅｎｃｅｉｔｉｓｖｅｒｙｉｍｐｏｒｔａｎｔｔｏｉｎｖｅｓｔｉｇａｔｅｃｈａｏｔｉｃｔｉｍ…     

Study on the Prediction Method of Low-Dimension Time Series that Arise from the Intrinsic Nonlinear Dynamics

The prediction methods and its applications of the nonlinear dynamic systems determined from chaotic time series of low-dimension are discussed mainly. Based on the work of the foreign researchers, the chaotic time series in the phase space adopting one kind of nonlinear chaotic model were reconstructed. At first, the model parameters were estimated by using the improved least square method. Then as the precision was satisfied, the optimization method was used to estimate these parameters. At the end by using the obtained chaotic model, the future data of the chaotic time series in the phase space was predicted. Some representative experimental examples were analyzed to testify the models and the algorithms developed in this paper. The results show that if the algorithms developed here are adopted, the parameters of the corresponding chaotic model will be easily calculated well and true. Predictions of chaotic series in phase space make the traditional methods change from outer iteration to interpolations. And if the optimal model rank is chosen, the prediction precision will increase notably. Long term superior predictability of nonlinear chaotic models is proved to be irrational and unreasonable. Paper from Chen Yu-shu, Member of Editorial of Committee, AMM Foundation item: the National Natural Science Foundation of China (19990510); the National Key Basic Research Special Fund(G1998020316) Biography: Ma Jun-hai(1965-), Professor, Doctor     

PREDICTION TECHNIQUES OF CHAOTIC TIME SERIES AND ITS APPLICATIONS AT LOW NOISE LEVEL

The paper not only studies the noise reduction methods of chaotic time series with noise and its reconstruction techniques,but also discusses prediction techniques of chaotic time series and its applications based on chaotic data noise reduction.In the paper,we first decompose the phase space of chaotic time series to range space and null noise space.Secondly we restructure original chaotic time series in range space.Lastly on the basis of the above,we establish order of the nonlinear model and make use of the nonlinear model to predict some research.The result indicates that the nonlinear model has very strong ability of approximation function,and Chaos predict method has certain tutorial significance to the practical problems.     

Can People Predict Chaotic Sequences?

Previous studies suggesting that people predict chaotic sequences better than chance have not discriminated between sensitivity to nonlinear determinism and facilitation using autocorrelation. Since prediction accuracy declines with increases in the look-ahead window in both cases, a decline in prediction accuracy does not imply chaos sensitivity. To overcome this problem, phase-randomized surrogate time series are used as a control. Such series have the same linear properties as the original chaotic sequence but contain no nonlinear determinism, i.e. chaos. In the experimental task, using a chaotic Hénon attractor, participants viewed the previous eight days temperatures and then predicted temperatures for the next four days, over 120 trials. The control group experienced a sample from a corresponding phase-randomized surrogate series. Both time series were linearly transformed to provide a realistic temperature range. A transformation of the correlation between observed and predicted values decreased over days for the chaotic time series, but remained constant and high for the surrogate series. The interaction between the days and series factors was statistically significant, suggesting that people are sensitive to chaos, even when the autocorrelation functions and power spectra of the control and experimental series are identical. Implications for the psychological assessment of individual differences in human prediction are discussed.     

A new type of atomic force microscope based on chaotic motions

Local flow variation (LFV) method of non-linear time series analysis is applied to develop a chaotic motion-based atomic force microscope (AFM). The method is validated by analyzing time series from a simple numerical model of a tapping mode AFM. For both calibration and measurement procedures the simulated motions of the AFM are nominally chaotic. However, the distance between a tip of the AFM and a sample surface is still measured accurately. The LFV approach is independent of any particular model of the system and is expected to be applicable to other micro-electro-mechanical system sensors where chaotic motions are observed or can be introduced.     

Predicting cortical oscillations with bidirectional LSTM network: a simulation study

It has been stated that up-down-state (UDS) cortical oscillation levels between excitatory and inhibitory neurons play a fundamental role in brain network construction. Predicting the time series behaviors of neurons in periodic and chaotic regimes can help in improving diseases, higher-order human activities, and memory consolidation. Predicting the time series is usually done by machine learning methods. In paper, the deep bidirectional long short-term memory (DBLSTM) network is employed to predict the time evolution of regular, large-scale UDS oscillations produced by a previously developed neocortical network model. In noisy time-series prediction tasks, we compared the DBLSTM performance with two other variants of deep LSTM networks: standard LSTM, LSTM projected, and gated recurrent unit (GRU) cells. We also applied the classic seasonal autoregressive integrated moving average (SARIMA) time-series prediction method as an additional baseline. The results are justified through qualitative resemblance between the bifurcation diagrams of the actual and predicted outputs and quantitative error analyses of the network performance. The results of extensive simulations showed that the DBLSTM network provides accurate short and long-term predictions in both periodic and chaotic behavioral regimes and offers robust solutions in the presence of the corruption process.

     

Nonlinear analysis of laser droplet generation by means of 0–1 test for chaos

We apply the recently improved version of the 0–1 test for chaos to real experimental time series of laser droplet generation process. In particular two marginal regimes of dripping are considered: spontaneous and forced dripping. The outcomes of the test reveal that both spontaneous and forced dripping time series can be characterized as chaotic, which coincides with the previous analysis based on nonlinear time series analysis.     

Generating Replications of Chaotic Time Series

This paper addresses the problem of resampling chaotic time series. We propose a method based on resampling distances between nearest neighbours in phase space, so that the resampled time series present the original points differently positioned along the trajectory. This approach allows one to obtain time series with the same length of the original one, so that confidence intervals for Lyapunov exponents, correlation dimension and other invariants would be determined. For its generality this kind of resampling would be applicable to chaotic time series no matter the observations concern natural or life sciences. The method has been tested with common simulated chaotic systems with both clean and noisy data. Short and noisy time series, as the ones we simulated, typically occur in medicine, biology, and social sciences.