模糊免疫算法及其在溶剂脱水塔软测量建模中的应用

本文针对基本免疫算法收敛速度慢、计算精度低等缺点,提出了模糊免疫算法.该算法引入模糊技术,对关键参数(交叉概率和变异概率)实现了模糊自适应调整.通过标准测试函数实验结果的对比,其可行性和有效性得到证明,不仅减轻了原始算法中参数确定存在的困难,而且提高了算法的计算速度和精度.其次,本文将模糊免疫算法用于径向基神经网络的训练,并将该神经网络应用于溶剂脱水塔软测量模型.仿真实验证明,模糊免疫算法优化的径向基函数神经网络具有良好的泛化性能.     

基于模糊神经网络和遗传算法的智能控制器

PID控制算法简单、鲁棒性强,但其参数整定过程繁琐,整定时需要控制对象的精确数学模型,而且整定往往是针对某一种具体工况进行的,缺乏自学习和自适应能力.模糊神经网络则兼备了模糊逻辑和神经网络的优点,具有函数逼近功能,具有较强的自适应、自学习能力、容错能力和泛化能力.借助于遗传算法对全局性参数进行优化设计,借助于BP算法对局部性参数进行优化,将模糊神经网络和遗传算法引入PID控制参数的整定过程,构造出一种基于模糊神经网络和遗传算法的智能PID控制器.     

基于改进FCM模糊神经网络的水处理过程建模

本文结合改进的FCM聚类分析算法,提出了一种自适应T-S模糊神经网络用于建立水处理过程的模型.该方法通过减法聚类初始化FCM聚类算法,加快了FCM聚类收敛速度,利用改进后的FCM算法对数据集聚类,从而产生输入空间的模糊划分和模糊规则;并用混合BP和递推最小二乘学习算法对前件和后件参数进行优化.最后,将本文的方法用于建立水处理过程的模型,仿真实验的结果表明该方法具有收敛快、精度较高、泛化能力好的优点.     

结合模糊(C+P)均值聚类和SP-V-支持向量机的TSK分类器

为获得具有模糊规则自适应约简性能和较好的泛化性能的TSK分类器,本文提出了一种结合模糊(C+P)均值聚类(FCPM)算法和SP-V-支持向量机(SVM)分类算法来构建TSK(Takagi-Sugeno-Kang)分类器的方法。该方法首先用FCPM聚类算法对训练数据进行聚类;然后根据聚类结果确定TSK分类器的模糊规则前件中的高斯隶属度函数的中心和宽度参数;最后采用成组稀疏约束SP-V-SVM算法对模糊规则后件参数进行学习,该算法不仅改善了系统的泛化性能,还使系统具有模糊规则自适应约简功能,使得系统更为紧凑。与相关算法在UCI和IDA标准数据集分类实验中的模糊规则数和分类性能对比表明:用提出的分类算法所构造的TSK分类器不仅具有较好的分类性能,而且模糊规则数少,有利于构建更为紧凑的模糊分类系统。     

基于数据场聚类的模糊神经网络在发酵过程中的应用

针对传统模糊神经网络方法中模糊规则难以提取、网络结构优化时间过长以及易于早熟的问题,提出了一种基于数据场聚类的免疫模糊神经网络方法.该方法将物理学中场的理念引入到抽象的数域空间,通过模拟对象在虚拟数据场中的相互作用实现数据对象的自组织聚类,提取模糊规则建立初始模糊神经网络模型,并运用免疫遗传算法优化构成隶属函数的网络结构.以实验室赖氨酸发酵过程关键生物量参数软测量为例,进行了仿真验证.结果表明,与常规方法相比,该方法具有较好的建模精度和实用性.     

RBF模糊自适应控制在电液速度伺服系统中的应用

由于电液速度伺服系统的非线性和参数的不确定性,难以建立精确的数学模型,文中引入RBF（径向基函数）模糊自适应控制,利用RBF神经网络进行自学习,修改和完善模糊规则,改善其动态性能.仿真结果表明该方法具有较强的自适应和自学习能力,即使对复杂的非线性系统也能取得良好的控制效果.     

自组织递归区间二型模糊神经网络在动态时变系统辨识中的应用

针对动态时变系统辨识过程中存在噪声干扰的问题,本文将区间二型模糊集结合到递归神经网络中,提出了自组织递归区间二型模糊神经网络以增强动态时变系统的抗噪能力.该自组织递归区间二型模糊神经网络由前件和后件两部分构成:前件为区间二型模糊集模型,用于将每个规则的激活强度反馈到自身构成内反馈回路,其参数学习采用梯度下降算法;后件为...     

基于决策逻辑的模糊粗糙神经网络建模

为建立相关量的预测模型,提出了一种新的基于决策逻辑的模糊粗糙神经网络建模方法.首先对原始数据进行预处理,并基于粗糙集理论进行属性约简,得到最简决策表.然后基于决策逻辑建立模糊粗糙神经网络.最后提出了一种结合混沌搜索算法和最小二乘法的Chaos-LS算法,训练模糊粗糙神经网络的参数,从而建立起系统的模糊粗糙神经网络模型.实验证明,这种建模方法建立的模糊粗糙神经网络模型具有较高的精度和泛化能力a     

基于遗传算法的模糊神经网络控制器在烘干炉温度控制系统中的设计与仿真

以烘干炉温度为被控对象,由于烘干炉温度控制具有非线性、大滞后和无法建立精确数学模型等特点,传统的控制器很难达到理想的控制效果,为此设计了一种基于遗传算法的模糊神经网络控制器.基于遗传算法的模糊神经网络控制器是将遗传算法的全局寻优和BP算法的在线学习结合起来,先用遗传算法对神经网络的参数进行离线训练,然后再用BP算法对模糊神经网络控制器进一步在线学习.仿真结果表明,基于遗传算法的模糊神经网络控制器与模糊控制、传统PID控制相比较,改善了系统的动态性能和静态性能,能使非线性、大滞后等特殊的系统达到良好的控制效果.     

基于Takagi-Sugeno模糊神经网络的欠驱动无人艇直线航迹跟踪控制

研究一类欠驱动无人艇的直线航迹跟踪控制问题,提出了一种自适应T-S(Takagi-Sugeno)模糊神经网络控制方法。首先在神经网络体系结构中设计前件网络匹配T-S模糊控制器的模糊规则前件,设计后件网络进行T-S模糊运算推理从而生成模糊规则后件;其次基于梯度下降法原理,设计了T-S模糊规则参数的优化学习算法;然后结合BP神经网络的误差反向传播原理和梯度下降法,设计了模糊神经网络体系误差的反向传播迭代算法,用于高斯隶属度函数参数的学习优化;最后设计了基于T-S模型的模糊神经网络控制器,并通过仿真实验验证了所提出方法和所设计控制器的有效性。     

Extraction Fuzzy Linguistic Rules from Neural Networks for Maximizing Tool Life in High-speed Milling Process

In metal cutting industry it is a common practice to search for optimal combination of cutting parameters in order to maximize the tool life for a fixed minimum value of material removal rate(MRR). After the advent of high-speed milling(HSM) pro cess, lots of experimental and theoretical researches have been done for this purpose which mainly emphasized on the optimization of the cutting parameters. It is highly beneficial to convert raw data into a comprehensive knowledge-based expert system using fuzzy logic as the reasoning mechanism. In this paper an attempt has been presented for the extraction of the rules from fuzzy neural network(FNN) so as to have the most effective knowledge-base for given set of data. Experiments were conducted to determine the best values of cutting speeds that can maximize tool life for different combinations of input parameters. A fuzzy neural network was constructed based on the fuzzification of input parameters and the cutting speed. After training process, raw rule sets were extracted and a rule pruning approach was proposed to obtain concise linguistic rules. The estimation process with fuzzy inference showed that the optimized combination of fuzzy rules provided the estimation error of only 6.34 m/min as compared to 314 m/min of that of randomized combination of rules.     

基于知识的神经网络推理机制的研究

在传统专家系统模糊知识规则库的基础上,给出模糊规则库规范化的处理方法,并将模糊规则通过正向推理处理以消解规则的冗余,最终以利于网络推理时的正确输入.给出了网络推理的原理、具体实现方法及网络推理的自组织学习算法.通过网络推理迭代次数的变化来改变学习率因子的方法,从而大大提高网络推理的效率.     

Nonlinear Fuzzy Linguistics for the Determination of Nitrogen Monoxide,Nitrogen Dioxide,and Sulfur Dioxide by Molecular Absorption Spectroscopy

The nonlinearity and high dimensionality of spectra data affect the precision and the complexity of molecular absorption spectroscopy models. This article proposes a nonlinear fuzzy linguistic method for spectral quantitative analysis. A nonlinear fuzzy linguistic rule is presented. In the rule antecedent, a set operation was used to express the input variables by the fuzzy linguistic terms. A flexible polynomial equation of the input variables was the rule consequent. The fuzzy linguistic terms, the membership functions, and the nonlinear linguistic rules were initialized automatically by Gaussian kernel fuzzy clustering analysis, and the related parameters of nonlinear fuzzy linguistic rules were tuned by the iterative optimization for minimizing the root-mean square error. The principal components of the absorption measurements were extracted as input variables to reduce the complexity of the model. Experimental measurements employed a spectral dataset of flue gas for quantitative determination of the components that included sulfur dioxide, nitric oxide, and nitrogen dioxide. The experimental results verify the effectiveness of the theoretical approach.     

基于SIRM的倒立摆模糊控制器

探讨了倒立摆控制中的一些方法和应用，指出了通常模糊控制规则过于复杂的问题，设计了基于SIRM(单输入规则模块)的模糊控制器，既解决了模糊控制规则复杂的问题，又实现了倒立摆的快速稳定。最后，通过仿真验证了控制系统的效果。     

塑料注射成型工艺参数优化的模糊规则网络模型

注射成型是塑料产品成型的最主要工艺,工艺参数是影响成型产品外观、尺寸与性能的关键因素之一。工艺参数的设置与优化属于弱理论、强经验的问题,迫切需要发展科学化、系统化的方法。针对产品缺陷修正中人工经验依赖性强的问题,构建知识的统一模糊化规则形式,建立工艺优化知识表示和推理于一体的Takagi-Sugeno-Kang(TSK)模糊规则网络模型。进一步,提出从工艺数据集自动发现工艺参数优化规则的学习方法,采用Dropout策略与Bagging集成学习策略缓解高维工艺数据下工艺知识库增长出现的规则数量爆炸等问题。分析了模糊规则网络参数、结构对知识表示和推理的影响,建立模型的参数学习与结构优化的双重进化方法。提出基于经验回放的工艺数据增量学习方法,建立数据的增量学习策略。在注射成型工艺数据集上的结果表明,模型的规则数量和长度降低了50%,具有高可解释性以及增量学习稳定性。     

基于模糊关联规则的预测系统遗传优化

为提高预测系统中的预测精度,提出了一种基于模糊关联规则的优化的预测系统设计方法。该方法通过两个阶段来实现:首先采用竞争聚集算法得到各数量型属性优化的模糊集个数,从而挖掘出优化的模糊关联规则。在得到用于构建预测系统规则库的模糊关联规则后,采用遗传算法约简冗余规则库,实现精确性和解释性的折衷,以提高预测精度。最后将此方法运用于Abalone样本数据集进行实验分析,证实此方法解决了模糊关联规则的冗余问题,有效提高了预测精度。     

Self-organizing fuzzy control of constant cutting force in turning

Constant force control is gradually becoming an important technique in the modern manufacturing process. Especially, constant cutting force control is a useful approach in increasing the metal removal rate and the tool life for turning systems. However, turning systems generally have nonlinear with uncertainty dynamic characteristics. Designing a model-based controller for constant cutting force control is difficult because an accurate mathematical model in the turning system is hard to establish. Hence, this study employed a model-free fuzzy controller to control the turning system in order to achieve constant cutting force control. Nevertheless, the design of the traditional fuzzy controller (TFC) presents difficulties in finding control rules and selecting an appropriate membership function. Moreover, the database and fuzzy rules of a TFC are fixed after the design step and then cannot appropriately regulate ones real time according to the system output response and the desired control performance. To solve the above problem, this work develops a self-organizing fuzzy controller (SOFC) for constant cutting force control to evaluate control performance of the turning system. The SOFC continually updates the learning strategy in the form of fuzzy rules, during the turning process. The fuzzy rule table of this SOFC can be begun with zero initial fuzzy rules which not only overcome the difficulty in the TFC design, but also establish a suitable fuzzy rules table, and support practically convenient fuzzy controller applications in turning systems control. To confirm the applicability of the proposed intelligent controllers, this work retrofitted an old lathe for a turning system to evaluate the feasibility of constant cutting force control. The SOFC has a better control performance in constant cutting force control than does the TFC, as verified in experimental results.     

Self-organizing fuzzy control of constant cutting force in turning

Constant force control is gradually becoming an important technique in the modern manufacturing process. Especially, constant cutting force control is a useful approach in increasing the metal removal rate and the tool life for turning systems. However, turning systems generally have nonlinear with uncertainty dynamic characteristics. Designing a model-based controller for constant cutting force control is difficult because an accurate mathematical model in the turning system is hard to establish. Hence, this study employed a model-free fuzzy controller to control the turning system in order to achieve constant cutting force control. Nevertheless, the design of the traditional fuzzy controller (TFC) presents difficulties in finding control rules and selecting an appropriate membership function. Moreover, the database and fuzzy rules of a TFC are fixed after the design step and then cannot appropriately regulate ones real time according to the system output response and the desired control performance. To solve the above problem, this work develops a self-organizing fuzzy controller (SOFC) for constant cutting force control to evaluate control performance of the turning system. The SOFC continually updates the learning strategy in the form of fuzzy rules, during the turning process. The fuzzy rule table of this SOFC can be begun with zero initial fuzzy rules which not only overcome the difficulty in the TFC design, but also establish a suitable fuzzy rules table, and support practically convenient fuzzy controller applications in turning systems control. To confirm the applicability of the proposed intelligent controllers, this work retrofitted an old lathe for a turning system to evaluate the feasibility of constant cutting force control. The SOFC has a better control performance in constant cutting force control than does the TFC, as verified in experimental results.     

Part input into a flexible flow system: An evaluation of look-ahead simulation and a fuzzy rule base

This article reports an investigation into part-input methods for an implemented flexible flow system (FFS). Two new dynamic methods—look-ahead simulation and a fuzzy heuristic rule base—are compared to three simple myopic static sequencing rules and one dynamic rule. It is shown using simulation that for the existing system, the best sequence generated from the minimum production set performs very well, but the dynamic methods outperform the sequence when the system is modified and the sequence is unaltered. It is concluded that for a stable FFS, the static determination of the best input sequence is appropriate, but that a rapidly changing FFS—due to machine breakdown, changes in production requirements, etc.—may benefit from a dynamic part-input method. The look-ahead simulation outperforms the fuzzy rule base, and appears to be a more promising dynamic method. Suggestions for future research and evaluation are given.