查询词典 neural network

At first, this paper surveys nonlinear control and the applications of neural networks in control respectively. Then it studies definitions, mathematical descriptions and conditions of invertibility for nonlinear systems, which are used to analyze the feasibility of inverse system control method. Conclusions about the feasibility of inverse control and the realizability of αth-order inverse systems in engineering are achieved. Furthermore, this paper discusses the feasibility, structure and identification of neural network αth-order inverse system for general single-input single-output continuous or discrete nonlinear systems and various composite control methods based on the αth-order inverse. At last, as a part work of an actual project, this paper focuses on the design of neural network αth-order inverse control method for thyristor controlled series compensation in power systems and its effect in improving the stability of power systems.

本文首先分别概述非线性系统的控制与人工神经网络在控制中的应用；然后研究非线性系统各种可逆性的定义、数学描述及可逆性条件，并用于分析逆系统控制方法的可行性，得出有关逆系统控制方法可行性的结论，阐明一般情况下α阶逆系统才工程可实现；接着本文针对一般单输入单输出连续、离散非线性系统，研究神经网络α阶逆系统的可行性、结构与辨识以及基于神经网络α阶逆系统的各种复合控制方法，并给出应用例子；最后本文结合实际项目，针对采用可控串补的电力系统，设计神经网络α阶逆系统控制方法，研究其对提高电力系统稳定性的作用。

Firstly, reconstruct attractors in phase spaces using chaotic theory,Secondly fit the attractor s evolvement using BP neural networks, because selecting neural network s input training data using Euclid distance and correlation, improve neural network s associative memory and ratiocinative ability, can better fit the attractor s evolvement.

提出一种将混沌理论、关联度和神经网络相结合的短期负荷预测模型，首先利用混沌理论重构负荷时间序列的相空间吸引子，然后用BP神经网络来拟合空间吸引子的演化，由于使用空间欧氏距离和关联度联合来选取神经网络的训练样本，这样就提高了神经网络对负荷序列混沌特性的联想和泛化推理能力，能够更好的拟合吸引子的演化。

Computer-simulation results related to the TNN model substantiate that the timing neural network could handle well the inverse-interpolation problems of one-to-one and multiple-to-one mappings. Keywords: neural networks, weights direct determination, inverse problems, timing neural network

基于前面提出的网络模型，本文进一步提出一种增加时序控制条件的神经网络，即时序神经网络模型，并给出理论推导和进行仿真验证，结果表明该时序神经网络能够成功解决一一映射及非一一映射反插值问题。

According to the basic principles of neural networks and in combination with 35 examples of actual landslide projects, this paper establishes RBF and BP neural network models respectively. The comparative results of the two models show that RBF neural network model has higher accuracy and fewer errors and is more suitable for the stability evaluation of landslide. Therefore, this model is used to calculate the stability coefficients of Yellow Caoba ancient landslide and predict the stability of the landslide.

根据神经网络的基本原理，结合35个实际滑坡工程实例，分别采用RBF神经网络和BP神经网络建立滑坡稳定性评价模型，利用MATLAB编程求解，并对比了两种模型的计算结果，表明采用RBF网络模型计算出的结果精度高、误差小，更适合于对滑坡的稳定性评价；同时，运用RBF网络模型计算了黄草坝古滑坡的稳定性系数，对其稳定性进行了预测，以验证该方法的有效性。

The first part is the introduction, which generally introduces the research purpose and significance of this thesis, the summary of the financial pre-warning researches both home and abroad, and the research main contexts and research methods; The second part is the comparison on the financial pre-warning methods existent, which compares the different methods by categories, makes a further analysis on the limits of existent methods, and proposes the improve thoughts of this research based on that; The third part is the basic theory, which makes a summary of the referent theories exerted on the research, and exposit the comparing mechanism of the grey prediction theory and the artificial neural network method concise; The fourth part is the frame work of the model, which illuminates the research designs step by step, such as the definition of the financial crisis, the method of sample select, the select and definition of the pre-warning variables, and makes an exposition of the establishment procedures of the whole models at last; The fifth part is the financial pre-warning based on the theory of grey prediction and artificial neural network, which mainly demonstrate the model establishment in the concrete, dividing into four part to carry out the part of pragmatic research integrally, building of indexes system, establishment of BP neural networks, dynamic financial pre-warning and test, analysis on the pre-warning result; The sixth part is conclusion, which sums up the main results of the whole research, clarifies the innovative points, and makes a discussion on the limits and future study directions of the research in this thesis.

其中，第一部分为导论，总括性概述本文的研究目的与意义，国内外研究综述，以及本文的研究内容和方法；第二部分为现有财务预警研究方法剖析，通过对现有预警方法进行分类比较，进一步分析现有预警方法的局限性，并在此基础上提出本文研究的改进思路；第三部分基本理论，对本文研究中运用的相关理论进行扼要概述，并简要阐述灰色预测理论与人工神经方法的融合机理；第四部分为模型研究框架，逐步阐释模型对财务困境的界定，样本数据的选取方法，预警变量的选取与定义等各个部分的研究设计，并对整体模型的构建步骤进行简要阐述；第五部分为基于灰色神经网络模型的财务预警，主要论述动态预警模型的具体实现过程，分模型指标体系的建立，BP神经网络的建立，动态财务预警与检验，以及预警结果分析四个部分全面阐述本文实证分析过程；第六部分为结论，总结本文研究得到的主要研究结论，以及本文研究中的创新之处，并对论文的不足与后续发展研究进行探讨。

In order to apply neural network and intelligent control in supermaneuverable flight, the mathematic model of flighter is obtained, and the nonlinear aerodynamic coefficients are identified using computational neural networks, then, a neural network control system for supermaneuver is designed according to the timescale, and a intelligent compensator is givento compensate for the system errors which are caused by the networks training error, no-samples input, disturbance and time-delay, etc, which including the design of knowledge base, blackboard and reasoning engneer, the simulation results of Herbst maneuver demonstrate the designed control system is satisfying.

为了在超机动中应用神经网络和智能控制，本文接着建立了超机动时飞机的数学模型，并利用神经网络对大迎角下非线性非定常气动力进行了辨识建模，然后按时标分离法设计了一个针对超机动飞行的神经网络控制系统。

Fuzzy neural network is used in the three phase adaptive reclosure so that the reclose of permanent faults can be avoided. A multi-input fuzzy neural network is constructed and the algorithm of network is designed. Taking the virtual value of three phase voltage as the input with its scale changed, the network learning can be completed with convergence reached after the process of fuzzification, normalization and perspicuity, so it can disinguish transient faults from permanent faults.

为了避免重合闸重合于永久性故障，将模糊神经网络应用于三相自适应重合闸中，构造了一个多输入的模糊神经网络，设计了网络的算法，将三相端压有效值作为输入，经尺度变换、模糊化、归一化和清晰化，并利用梯度下降法修正误差，使得网络完成学习并最终收敛，在输入故障数据时能根据网络的输出结果准确地判别瞬时性故障与永久性故障。

The simulation results show that the search-optimization capacity of wavelet chaotic neural network has been improved and the reformative wavelet chaotic neural network is superior to the primary wavelet chaotic neural networks.

仿真结果表明，改善了小波混沌神经网络的寻优能力，改进的小波混沌神经网络优于原来的小波混沌神经网络。

Based on neural network methods, this article establishes neural network models which are about using valuable experimental data to assist the design of new experiments. It carries out simulation calculation, taking glucose plasma experiments as studying object, using the quantity of liquefied enzyme, liquefied time, the quantity of saccharifying enzyme and saccharified time as the network input, using DE value as the network output.

基于神经网络理论建立了利用已有的实验数据对新实验进行辅助设计的神经网络模型，并以制作葡萄糖浆实验为研究对象，定义液化酶加量、液化时间、糖化酶加量、糖化时间为网络的输入，DE值为网络的输出，进行了仿真计算。

The dissertation recommends some kinds of methods and measures: such as adjusting the network configuration, connecting value and threshold value by the total value of all stylebooks, adding part of the adjusting value of the last time to the current adjusting value, transforming the stylebooks to standard value, optimizing activation function, appending threshold value to the putout of the nerve cell and so on. Simultaneity, the author brings forward a new method to optimize the model of artificial neural network . It is using the automatically adaptive genetic algorithm to make the network configuration, connecting value and threshold value of artificial neural network better. The method can make the model better and improve the simulating effect and forecasting precision.Genetic algorithm is an arithmetic based on evolution and genetics used to search the optimization .

本文针对人工神经网络应用中存在的上述问题，介绍了各种改进方法与措施：如用所有样本的总效果对网络权值矩阵和阈值向量进行调整、调整量中加入动量项、标准化训练样本数据、优化激励函数以及给神经元的输出值添加偏置量等；同时还提出了一种新的优化人工神经网络模型的方法，即采用自适应遗传算法对人工神经网络模型的网络结构和权值阈值进行全局优化搜索，以提高大坝安全监测人工神经网络模型的拟合成果和预测精度。

The split between the two groups can hardly be papered over.

这两个团体间的分歧难以掩饰。

This approach not only encourages a greater number of responses, but minimizes the likelihood of stale groupthink.

这种做法不仅鼓励了更多的反应，而且减少跟风的可能性。