基于PLS快速剪枝法的RBF神经网络软测量模型建模方法和应用

提出一种从RBF神经网络隐含层的输出信息出发，通过PLS快速剪枝法，一次性剪去多余节点，生成最优规模的数学解析模型的方法。并用该方法建立了某化工企业精对苯二甲酸(PTA)晶体平均粒径的软测量模型，针对实际对象进行仿真研究，结果表明，该方法计算速度快，建立的模型精度高，适合实际工程应用的需求。     

嵌入岭回归的BP算法及其在软测量中的应用

针对三层神经网络(ANN)最佳隐节点个数难以确定和随着隐节点个数增加ANN模型易出现过拟合等缺点,提出了嵌入岭回归(RR)的误差反传算法(BP).BP-RR根据样本规模自适应确定隐节点个数,并通过BP算法充分提取样本数据信息.然后,针对隐含层输出可能存在的复共线性,采用RR以预测性能为指标,通过进化算法确定最佳岭参数,进而重新确定隐含层与输出层之间最佳的权值和阈值,克服ANN过拟合,建立具有良好预测性能的模型.将BP-RR应用于建立石脑油干点软测量,结果显示,BP-RR模型具有良好的预测性能.与ANN相比,BP-RR模型鲁棒性强,预测精度高.     

管壳式换热器ANN静态预测最优结构分析

采用人工神经网络技术(ANN)对连续螺旋折流板换热器的沿程压降进行了辨识及预测,开发了4-3-1型人工神经网络结构及计算程序,人工神经网络的预测结果与实验数据良好.通过固定网络结构,讨论了训练与预测样本比例对网络性能的影响;在网络结构及训练与预测样本比例不变的情况下,讨论隐含层节点对网络性能的影响,对人工神经网络进行了优化计算和分析.计算结果表明:随着学习样本数据的增多,预测精度变高,但是同时由于预测样本变少,神经网络模型的泛化能力变弱;ANN选取的隐含层节点数不能过多,否则会导致训练过度,节点数也不能太少,否则精度会降低.     

基于改进聚类算法的RBF网络及其应用

RBF网络可以逼近任意连续非线性函数,且训练速度快,性能好,被广泛应用于过程建模和预测。RBF网络的一个重要因素是隐层节点的选择,隐层节点过多或过少都会影响最终网络的性能。提出一种改进的k-means聚类算法,可以自动确定最优的聚类区数,并且可使最终的聚类中心合理地分布在数据空间中。在应用RBF网络进行建模和预测时,采用该方法确定隐层节点的中心,跟用通常的聚类方法相比,可以大大减小网络规模。仿真和实际应用结果都证明该方法的有效性。     

BP神经网络在优化配煤预测模型中的研究

由于掺配煤种煤质的波动 ,配煤是一个在不确定的条件下的优化问题 .传统的线性规划模型已不能解决这种非线性问题 ,而 BP神经网络这一非线性优化工具已成功地应用于混煤煤质特性的预测模型 .本文详细地分析了不同 BP神经网络模型的预测效果以及制约其预测效果的主要因素 (网络结构、学习样本数量、隐层节点数、学习精度 ) ,并发现学习样本数是影响 BP神经网络性能的关键     

建筑卫生陶瓷缺陷分析神经网络系统

尝试利用BP神经网络模型,确定其输出输入值的模式,确定隐层节点数,探讨除用Sigmoid函数外,还有什么函数更能提高自学习效率及发挥其功能.探讨切合建筑卫生陶瓷缺陷的神经网络相关参数,以保持陶瓷的网络模型的稳定性.通过收集样本来训练、测试和考核结论.     

基于EGK'M-RBF网络的顺丁橡胶门尼黏度预测

提出一种基于增强的全局K'-means算法(EGK'M)-RBF网络的建模方法,该方法采用作者提出的EGK'M来确定RBF网络隐含层的结构,包括隐含层中心个数、中心位置以及隐含层扩展常数,采用KPCA提取非线性特征信息,实现辅助变量的二次选择.并与基于PCA和EGK'M-RBF网络模型、基于KPCA和K-means算法...     

BP网络结构和算法对非线性系统进行辨识

利用MATLAB神经网络工具箱进行了基于BP神经网络的非线性系统辨识仿真研究,针对实际问题选取BP网络结构和BP算法,并对网络输入、输出、隐层节点的个数选取作了探讨,选取一个非线性系统,通过改变隐层节点数和训练函数,找出适当的BP网络结构和算法对非线性系统进行辨识。     

基于模糊RBF神经网络的乙烯装置生产能力预测

针对传统的径向基函数(RBF)神经网络隐藏层节点的不确定和初始中心敏感性、收敛速度过慢等问题,提出一种基于模糊C均值的RBF神经网络(FCM-RBF)模型,通过模糊C均值聚类(FCM)得到各聚类中心,基于误差反传的梯度下降法训练隐藏层到输出层之间的权值,克服传统RBF模型对数据中心的敏感性,优化确定RBF神经网络隐藏层的节点数,提高网络训练速度和精度。最后将其用于乙烯装置生产能力预测中,分析预测不同技术、不同规模乙烯装置生产情况,指导乙烯生产,提高生产效率,结果验证了所提出算法的有效性和实用性。     

神经网络系统预测采面来压研究

通过人工神经网络方法,将影响工作面来压的主要因素作为输入层,构建BP神经网络模型,应用黄金分割算法确定最优隐含层节点数,得到最优模型,并对工作面支架平均来压阻力、平均非来压阻力、平均来压步距进行预测,分析可知预测误差在±10%,且符合正态分布,可通过多种方法提高预测精度,控制在±5%,取得了较好的预测结果精度,对于指导工作面的安全生产具有重要意义。     

自增长混合神经网络及其在燃料电池建模中的应用

为了提高非线性辨识的精度, 提出了一种基于混合算子的自增长混合神经网络。该神经网络通过自增长的混合隐含层结构, 包括加算子和乘算子, 形成神经元个数少、结果精确、增长快速的网络。论文在级联神经网络的结构基础上, 提出GQPSOI算法来引导神经网络的结构自增长以及权值更新。通过对燃料电池的建模与比较分析, 证明了方法的有效性和良好的应用前景。     

动态系统神经网络结构的改进及其应用

在对动态神经网络结构分析的基础上,提出了新的网络结构模式,把具有同类特性的输人参数集中在一个节点上以代表该类参数的性质、产生这类参数对网络的综合作用。这样构造的更合乎逻辑的神经网络的权值总数要比传统的BP网络的权值总数大大减少,从而加快网络学习速度,有利于网络的在线学习和提高网络的可靠性与稳定性。此外,本文对神经网络逆动态控制器进行了分析,提出在输入层增加系统偏差作为一个输入变量,从而增强了控制器的控制质量和控制反应能力。最后,应用上述技术对CSTR典型化工实例进行了验证,取得了较好的结果。     

Control of nonisothermal CSTR with time varying parameters via dynamic neural network control (DNNC)

Dynamic neural network control (DNNC) is a model predictive control strategy potentially applicable to nonlinear systems. It uses a neural network to model the process and its mathematical inverse to control the process. The advantages of single hidden layer DNNC are threefold: First, the neural network structure is very simple, having limited nodes in the hidden layer and output layer for the SISO case. Second, DNNC offers potential for better initialization of weights along with fewer weights and bias terms. Third, the controller design and implementation are easier than control strategies such as conventional and hybrid neural networks without loss in performance. The objective of this paper is to present the basic concept of single hidden layer DNNC and illustrate its potential. In addition, this paper provides a detailed case study in which DNNC is applied to the nonisothermal CSTR with time varying parameters including activation energy (i.e., deactivation of catalyst) and heat transfer coefficient (i.e., fouling). DNNC is compared with PID control. Although it is clear that DNNC will perform better than PID, it is useful to compare PID with DNNC to illustrate the extreme range of the nonlinearity of the process. This paper represents a preliminary effort to design a simplified neural network-based control approach for a class of nonlinear processes. Therefore, additional work is required for investigation of the effectiveness of this approach for other chemical processes such as batch reactors. The results show excellent DNNC performance in the region where conventional PID control fails.     

酪蛋白-胰酶水解历程分子量变化模拟与三维表征

食源性活性多肽的生物活性功能具有很强的分子量依赖性,不同分子量的多肽有不同的功能。本文运用人工神经网络模拟了酪蛋白在胰酶作用下酶解全过程的分子量分布变化,并基于此模型进行了酶解历程三维表征。使用BP神经网络模拟,隐含层为2层,每层含30个节点时拟合效果最好,回归系数R²可达0.9922。将预测数据用于复杂酶解历程的三维表征,可通过该三维表征图迅速判断各集总分子量多肽区的最佳制备水解度。     

Batch-to-batch optimal control of a batch polymerisation process based on stacked neural network models

A neural network based batch-to-batch optimal control strategy is proposed in this paper. In order to overcome the difficulty in developing mechanistic models for batch processes, stacked neural network models are developed from process operational data. Stacked neural networks have enhanced model generalisation capability and can also provide model prediction confidence bounds. However, the optimal control policy calculated based on a neural network model may not be optimal when applied to the true process due to model plant mismatches and the presence of unknown disturbances. Due to the repetitive nature of batch processes, it is possible to improve the operation of the next batch using the information of the current and previous batch runs. A batch-to-batch optimal control strategy based on the linearisation of stacked neural network model is proposed in this paper. Applications to a simulated batch polymerisation reactor demonstrate that the proposed method can improve process performance from batch to batch in the presence of model plant mismatches and unknown disturbances.     

NEURAL NETWORK MODEL FOR FLUIDISED BED DRYERS

The application of an artificial neural network (ANN) to model a continuous fluidised bed dryer is explored. The ANN predicts the moisture and temperature of the output solid. A three-layer network with sigmoid transfer function is used. The ANN learning is made by using a set of data that were obtained by simulating the operation by a classical model of dryer. The number of hidden nodes, learning coefficient, size of learning data set and number of iterations in the learning of the ANN were optimised. The optimal ANN has five input nodes and six hidden nodes. It is able to predict, with an error less than 10%, the moisture and temperature of the output dried solid in a small pilot plant that can treat up to 5 kg/h of wet alpeorujo. This is a wet solid waste that is generated in the two-phase decanters used to obtain olive oil.     

Auto-analysis system for graphite morphology of grey cast iron

The current method to classify graphite morphology types of grey cast iron is based on traditional subjective observation, and it cannot be used for quantitative analysis. Since microstructures have a great effect on the mechanical properties of grey cast iron and different types have totally different characters, six types of grey cast iron are discussed and an image-processing software subsystem that performs the classification and quantitative analysis automatically based on a kind of composed feature vector and artificial neural network (ANN) is described. There are three kinds of texture features: fractal dimension, roughness and two-dimension autoregression, which are used as an extracted feature input vector of ANN classifier. Compared with using only one, the checkout correct precision increased greatly. On the other hand, to achieve the quantitative analysis and show the different types clearly, the region segmentation idea was applied to the system. The percentages of the regions with different type are reported correctly. Furthermore, this paper tentatively introduces a new empirical method to decide the number of ANN hidden nodes, which are usually considered as a difficulty in ANN structure decision. It was found that the optimum hidden node number of the experimental data was the same as that obtained using the new method.     

分离系统与换热网络同步优化方法的改进：神经元网络—遗传算法

采用神经元网络法和遗传算法，在过程系统用能一致性的基础上对分离系统与换热网络同步优化问题提出了改进的优化模型及优化策略。该方法不仅能够自动、迅速地同步得到分离序列与换热网络联合系统的流程结构与操作参数，而且具有获得全局最优解的能力。最后通过实例说明本方法的有效性。