用神经网络进行连续时间非线性系统建模的研究

在用神经网络进行系统建模时,建模误差的存在是难免的。为了减小这种误差,本文对连接时间非线性系统提出了一种新的神经网络辨识模型,它是由带有输入修正的神经网络和稳定滤波器组合而成。文中给出了权值的学习算法,即权值是根据辨识误差的投影算法来改变,证明了在一定条件下辨识误差的收敛性。     

一种基于函数型神经网络的传感器静态模型辨识方法

本文提出一种基于函数型神经网络的传感器静态模型辨识方法,该神经网络连接系数直接反映了传感器静成模型中的被辨识参数,网络结构简单,具有良好的收敛性,文章将这一方法实际应用到铂热电阻静态模型辨识,仿真结果表明,本方法是可行的。     

时变大时滞系统参数辨识的仿真研究

研究了时变大时滞系统的参数辨识问题.大时滞系统大多采用补偿控制方法,但是补偿控制方法需要系统的精确数学模型,因而获得大时滞系统的数学模型成为了补偿控制的关键,时变特性使问题复杂化,从而影响了大时滞系统的控制精度.为解决上述问题,提出了一种神经网络的参数辨识策略,利用一个神经元对系统的时滞参数进行辨识,从而可以将时滞从系统模型中分离出来,可利用一个RBF神经网络模型辨识系统的其它参数,使神经元的输出作为RBF神经网络的一个输入,从而实现了串-并联结构的双神经网络拓扑.拓扑结构可以比串级的神经网络提高训练速度,因而也就更适合于实时控制.针对工业锅炉回水温度控制系统的仿真结果验证了所提辨识算法的正确性.     

基于ANN逆模型补偿的复合控制系统仿真

为对复杂非线性系统进行辨识建模和实施有效控制，分析了基于神经网络的非线性系统逆模型的辨识和控制原理，研究了基于神经网络的非线性系统逆模型补偿的复合控制方法。基于复合控制思想，时常规PID控制器＋前馈神经网络逆模型补偿的复合控制结构方案进行了仿真。仿真结果表明，基于神经网络的非线性系统逆模型补偿的复合控制结构方案是有效的、相对简单的网络结构，可提高逆模型的泛化能力和非线性系统的控制精度。     

非线性系统辨识方法研究

讨论了利用小波神经网络对非线性系统辨识的新方法。在辨识过程中,为了提高小波神经网络对非线性系统的辨识性能,使用一种改进粒子群优化算法对BP小波神经网络参数进行训练,求得最优值,达到对非线性系统辨识目的。在数值仿真中,与采用标准粒子群优化算法相比,结果显示了提出的方法在收敛性和稳定性等方面均得到了明显的改善。     

多级PSO神经网络在手写体字符识别中的应用

提出一种用于手写体字符识别的三级神经网络模型，各子神经网络均用粒子群优化算法(PSO)训练。在该模型中，各个神经网络与不同的图像特征提取方法相结合；识别时，三个神经网络先串联再并联。该模型充分有效地利用了各种特征信息，从实验结果看，也达到了较好的辨识目的。文中主要讨论手写字符图像的特征提取、粒子群优化算法及其在网络训练上的应用，最后分析了识别结果并与采用改进BP训练算法的综合识别效果进行了比较。     

二阶微粒群优化神经网络的混沌系统辨识方法

针对BP神经网络在学习算法中的不足,将BP神经网络的权值和阀值训练问题转换为优化问题,提出一种利用二阶微粒群算法优化的神经网络的算法。其次,运用基于二阶微粒群算法训练的神经网络模型对混沌系统进行辨识,并与传统的BP神经网络、RBF网络对同一混沌系统辨识的结果进行比较。实验表明,利用二阶微粒群优化算法训练神经网络进行混沌系统辨识,辨识的效果优于其它几种神经网络模型,可有效用于混沌系统的辨识。     

水处理过程的RBF和BP神经网络建模

针对水处理过程非线性、时变和大滞后的特点，本文采用RBF和BP神经网络分别建立了水处理过程模型，利用水厂实际运行数据对两个模型分别进行了训练和检验。与BP神经网络模型相比，RBF神经网络模型具有逼近能力强、收敛速度快等优点。该模型可以实现对水处理过程的在线辨识，并可进一步用于该过程的神经网络预测控制。     

NSG水位特性神经网络辨识方法研究

核动力蒸汽发生器(NSG)是压水堆核动力装置中把一回路冷却剂从反应堆堆芯带出的热量传递给二回路水的关键性设备。在瞬态、启动和低功率下的“收缩”与“膨胀”现象引起的逆动力学效应使核动力蒸汽发生器水位呈现瞬时“虚假水位”现象,并使其水位特性难以辨识。为了提高辨识效果,提出了NSG水位神经网络辨识的新方法。采用串—并联型辨识结构,以保证辨识的收敛性和稳定性。网络训练采用带动量因子与自适应学习率的BP学习算法。仿真结果表明,所提出的方法能够正确地辨识核动力蒸汽发生器的水位特性,且具有较高的辨识精度。     

基于ANN的动态系统状态方程辨识建模仿真

对系统辨识原理、基于神经网络（ANN）的动态系统辨识进行了分析，针对动态系统辨识模型描述的复杂性，为简化ANN辨识建模的输入／输出关系的表达，提高算法的简洁性，采用了状态方程辨识模型，并给出了基于ANN的动态系统状态方程辨识模型。为比较分析不同网络结构的辨识建模效果及网络模型泛化能力，针对三种不同网络结构方案进行了辨识建模仿真研究。仿真结果最示，基于ANN的动态系统状态方程模型的辨识建模是有效的，并且简单合理的网络结构方案，可提高网络辨识模型的泛化能力。     

Integrated OBF-NN models with enhanced extrapolation capability for nonlinear systems

This paper proposes a nonlinear system identification using parallel linear-plus-neural network models that provide more accurate predictions on the process behavior even on extrapolated regions. For this purpose, a residuals-based identification algorithm using parallel integration of linear orthonormal basis filters (OBF) and neural networks model is developed and analyzed under range extrapolations. Results on the van de Vusse reactor case study show enhanced extrapolation capability when compared to the conventional neural network (NN) and the series Wiener-NN models.     

PNNARMA model: an alternative to phenomenological models in chemical reactors

This paper is focused on the development of non-linear neural models able to provide appropriate predictions when acting as process simulators. Parallel identification models can be used for this purpose. However, in this work it is shown that since the parameters of parallel identification models are estimated using multilayer feed-forward networks, the approximation of dynamic systems could be not suitable. The solution proposed in this work consists of building up parallel models using a particular recurrent neural network. This network allows to identify the parameter sets of the parallel model in order to generate process simulators. Hence, it is possible to guarantee better dynamic predictions. The dynamic behaviour of the heat transfer fluid temperature in a jacketed chemical reactor has been selected as a case study. The results suggest that parallel models based on the recurrent neural network proposed in this work can be seen as an alternative to phenomenological models for simulating the dynamic behaviour of the heating/cooling circuits.     

Neural network augmented identification of underwater vehicle models

In this article the use of neural networks in the identification of models for underwater vehicles is discussed. Rather than using a neural network in parallel with the known model to account for unmodelled phenomena in a model wide fashion, knowledge regarding the various parts of the model is used to apply neural networks for those parts of the model that are most uncertain. As an example, the damping of an underwater vehicle is identified using neural networks. The performance of the neural network based model is demonstrated in simulations using the neural networks in a feed forward controller. The advantages of online learning are shown in case of noise impaired measurements and changing dynamics due to a change in toolskid.     

Parallel Parameter Identification in Industrial Biotechnology

Real-valued black-box optimization of badly behaved and not well understood functions is a wide topic in many scientific areas. Possible applications range from maximizing portfolio profits in financial mathematics over efficient training of neuronal networks in computational linguistics to parameter identification of metabolism models in industrial biotechnology. This paper presents a comparison of several global as well as local optimization strategies applied to the task of efficiently identifying free parameters of a metabolic network model. A focus is being set on the ease of adapting these strategies to modern, highly parallel architectures. Finally an outlook on the possible parallel performance is being presented.     

基于并行自学习的神经网络内模控制及其应用研究

基于在线并行自学习的神经网络内模控制,该方法是借助于神经网络对复杂系统的辩识能力对被控对象进行正模型及逆模型的辩识,用NNM辩识对象的正模型,通过一个并行自学习系统训练的NNC辩识对象的逆模型,然后用做内模控制器去控制对象。将该种控制策略应用于火电厂热工对象中具有大迟延、大惯性和时变等特性的主汽温对象,仿真研究表明,该控制方案适应对象参数的变化并表现出良好的控制特性,具有较强的鲁棒性和自适应能力。在实际应用中具有一定的实用价值。     

Memory neuron networks for identification and control of dynamicalsystems

This paper discusses memory neuron networks as models for identification and adaptive control of nonlinear dynamical systems. These are a class of recurrent networks obtained by adding trainable temporal elements to feedforward networks that makes the output history-sensitive. By virtue of this capability, these networks can identify dynamical systems without having to be explicitly fed with past inputs and outputs. Thus, they can identify systems whose order is unknown or systems with unknown delay. It is argued that for satisfactory modeling of dynamical systems, neural networks should be endowed with such internal memory. The paper presents a preliminary analysis of the learning algorithm, providing theoretical justification for the identification method. Methods for adaptive control of nonlinear systems using these networks are presented. Through extensive simulations, these models are shown to be effective both for identification and model reference adaptive control of nonlinear systems.     

基于并行支持向量机的多变量非线性模型预测控制

提出一种基于并行支持向量机的多变量系统非线性模型预测控制算法．首先，通过考虑输入、输出间的耦合，建立基于并行支持向量机的多步预测模型；然后，将该模型用于非线性预测控制,提出新的适用于并行预测模型的反馈校正策略，得到最优控制律．连续搅拌槽式反应器（CSTR）的控制仿真结果表明,该算法的性能优于基于并行神经网络的非线性模型预测控制和基于集成模型的非线性模型预测控制．     

Identification of non-linear processes using reciprocal multiquadric functions

In this paper radial basis function (RBF) networks are used to model general non-linear discrete-time systems. In particular, reciprocal multiquadric functions are used as activation functions for the RBF networks. A stepwise regression algorithm based on orthogonalization and a series of statistical tests is employed for designing and training of the network. The identification method yields non-linear models, which are stable and linear in the model parameters. The advantages of the proposed method compared to other radial basis function methods and backpropagation neural networks are described. Finally, the effectiveness of the identification method is demonstrated by the identification of two non-linear chemical processes, a simulated continuous stirred tank reactor and an experimental pH neutralization process.     

多核并行脉冲神经网络模拟器的设计

脉冲神经网络属于第三代人工神经网络,它是更具有生物可解释性的神经网络模型。随着人们对脉冲神经网络不断深入地研究,不仅神经元空间结构更为复杂,而且神经网络结构规模也随之增大。以串行计算的方式,难以在个人计算机上实现脉冲神经网络的模拟仿真。为此,设计了一个多核并行的脉冲神经网络模拟器,对神经元进行编码与映射,自定义路由表解决了多核间的网络通信,以时间驱动为策略,实现核与核间的动态同步,在模拟器上进行脉冲神经网络的并行计算。以Izhikevich脉冲神经元为模型,在模拟环境下进行仿真实验,结果表明多核并行计算相比传统的串行计算在效率方面约有两倍的提升,可为类似的脉冲神经网络的模拟并行化设计提供参考。