检索结果-南通市图书馆

在线全文

学校读者我要写书评

暂无评论

基于FGCN的针铁矿沉铁过程建模

化工学报 2018年第3期69卷 1141-1148页

作者：陈宁周佳琪桂卫华王磊中南大学信息科学与工程学院湖南长沙410083

针铁矿沉铁过程是由多个连续反应器级联,并且包含氧化反应、还原反应以及中和反应等一系列复杂化学反应的复杂过程,具有强非线性、不确定性的特点,难以建立精确的数学模型。提出一种基于模糊灰色认知网络(fuzzy gray cognitive network,... 详细信息

针铁矿沉铁过程是由多个连续反应器级联,并且包含氧化反应、还原反应以及中和反应等一系列复杂化学反应的复杂过程,具有强非线性、不确定性的特点,难以建立精确的数学模型。提出一种基于模糊灰色认知网络(fuzzy gray cognitive network,FGCN)的针铁矿沉铁过程的建模方法。根据专家经验和历史数据,建立针铁矿沉铁系统的模糊灰色认知网络模型,利用带终端约束的非线性Hebbian学习算法(nonlinear Hebbian learning,NHL)对权值进行学习。在不同程度的不确定性环境下对系统进行分析,结果表明模糊灰色认知网络能够在不确定性高的环境下对复杂工业系统进行有效模拟,收敛到一个灰度为零或者灰度很小的灰数平衡点,利用白化函数得到一个准确的控制输出。

关键词：模糊灰色认知网络针铁矿沉铁过程非线性Hebbian学习化学反应氧化水解

Cloud-Based Predictive Control Strategy with Deep Learning Modeling for Efficient Zinc Rotary Kiln Process Control

在线全文

学校读者我要写书评

暂无评论

Cloud-Based Predictive Control Strategy with Deep Learning M...

第43届中国控制会议

作者： Xiaoming Wang Yiqi Liu Chaobo Zhang Xiaojun Liang Chunhua Yang weihua gui Pengcheng Laboratory School of Future Technology South China University of Technology

This study aims to explore a strategy that combines cloud control technology and deep learning predictive control to enhance the efficiency of process control for zinc rotary kilns. Firstly, a deep learning model using state-space long short-term memory(SSLSTM) networks to capture the dynamic characteristics of the zinc rotary kiln system. This approach allows us to model complex nonlinearities and time dependencies that are difficult to capture using traditional modeling ***, a predictive control algorithm is designed that utilizes the SSLSTM model to predict the future state of the system and generate optimal control inputs. Finally, by integrating cloud computing technology, the designed predictive control algorithm is deployed in the cloud for efficient computation of control inputs, which then fed back to the zinc rotary kiln system to optimize production efficiency and minimize energy consumption. The proposed cloud-based predictive control strategy was evaluated on a zinc rotary kiln system. The experiment results showed that our approach achieved more precise control than traditional control methods, resulting in a significant reduction in energy consumption while maintaining the desired production efficiency. Moreover, the cloud-based design enabled high scalability, flexibility, and adaptability, making the approach suitable for large-scale and complex industrial systems.

关键词： Cloud control technology deep learning predictive control zinc rotary kiln long short-term memory(LSTM)

DMA-Trans: A Difference Metric Attention with Position Distancebased Weighting for Transformer in Data Sequence Modeling of Industrial Processes

在线全文

cnki会议

学校读者我要写书评

暂无评论

DMA-Trans: A Difference Metric Attention with Position Dista...

第35届中国过程控制会议

作者： Ziyi Yang Xiaofeng Yuan Kai Wang Yalin Wang Chunhua Yang weihua gui School of Automation Central South University

Accurate feature extraction and quality variable prediction are critical problems for time sequences in industrial processes. However, industrial samples are often time-relevant and have strong correlations with each other that has different positional distance, making it challenging for conventional data-driven models like LSTM and Vanilla Transformer to capture these underlying features. In this paper, A Difference Metric Attention with Position Distance-based Weighting is proposed for Transformer(DMA-Trans) in industrial time series modeling. Firstly, the Difference Metric Attention is established to calculate the difference of query-key vector pair to measure the spatial similarity. In this fashion, the difference can accurately represent the similarity of vectors, compared with the original dot product by directly using any two vectors. Then, Positional Distancebased Weights are designed to capture the distance relevance of samples that have different positional distance. This information may help to extract more potential features because the closer samples tend to have higher relevance while there are weak correlations if samples are far in distance. The effectiveness of the DMA-Trans model is validated in industrial hydrocracking processes for C5 content in the light naphtha and the final boiling point of jet fuel.

关键词： Soft sensor Attention mechanism Dot product Quality prediction Transformer

A Long-tune Natural Logarithm Method for CO2 Detection in Process Industrial Sites and Its Precision Enhancement at Low Concentration

在线全文

cnki会议

学校读者我要写书评

暂无评论

A Long-tune Natural Logarithm Method for CO2 Detection in Pr...

第35届中国过程控制会议

作者： Changsheng Zhang Lijuan Lan Rongzong Huang Qiwu Luo Chunhua Yang weihua gui School of Energy Science and Engineering Central South University

This paper discusses the feasibility of applying the natural logarithm method combined with long-tune spectroscopy（long-tune ln-WMS） for CO detection in process industrial sites. The study investigates the background signals present in the harmonics obtained using this method. The theory demonstrate that the background signal of the harmonic includes only the relative light intensity items and is related to the modulation depth but not the gas concentration with a nonlinear curve instead of linear in the long-tune ln-WMS. Light intensity fluctuations do not affect the harmonic signal. A measurement system based on the ln-WMS method is built and the experimental validations are conducted. The experimental results verify the theoretical predictions. The long-tune ln-WMS has high anti-interference ability and is suitable for gas monitoring in harsh environments, especially in areas with heavy dust or significant mechanical vibrations. The gas concentration detection results indicate that once the background signal is eliminated, accuracy can be achieved with a relative error below 0.5% compared to the conventional "2f/0f" method, whereas the error exceeds 5% without background subtraction. The background elimination method based on the long-tune ln-WMS, proposed in this paper, significantly enhances detection precision and is effective for trace gas detection（weak absorption） or overmodulation conditions. This method has potential applications in the field inspection of industrial production.

关键词： process industrial sites long-tune spectrum ln-WMS background signal RAM concentration measurement

Linear Convolutional Wavelength Modulation Spectroscopy

在线全文

cnki会议

学校读者我要写书评

暂无评论

Linear Convolutional Wavelength Modulation Spectroscopy

第35届中国过程控制会议

作者： Yu Xie Lijuan Lan Rongzong Huang Qiwu Luo Chunhua Yang weihua gui School of Energy Science and Engineering Central South University

This study introduces a novel Linear Convolution Wavelength Modulation Spectroscopy(LC-WMS) technique for demodulating absorption spectra. This method can directly obtain absolute harmonic signals related to gas concentration without the need of traditional lock-in amplifiers and low-pass filters, thereby simplifying the detection process. Simulation results demonstrate that appropriately increasing the cycles of the reference signal helps to reduce the noise of the harmonic *** findings indicate that the accuracy of LC-WMS in measuring CO2and H2O concentrations is comparable to that of conventional LIA techniques, with concentration errors of less than 5 ppm and 0.05‰, respectively. Moreover, long-term observations have shown that the method can be used for the detection of gases with large concentration fluctuations. Thus, it has potential applications in industrial field measurements.

关键词： Linear convolution harmonic demodulation gas detection LC-WMS

在线全文

cnki会议

学校读者我要写书评

暂无评论

人工智能助力制造业优化升级

中国科学基金 2018年第3期32卷 257-261页

作者：钱锋桂卫华华东理工大学信息科学与工程学院上海200237 中南大学信息科学与工程学院长沙410083

制造业是我国实体经济的基础,而人工智能是经济发展的新引擎。本文基于以人工智能基础理论及应用为主题的"双清论坛"讨论内容,阐述了我国流程工业智能制造的发展现状、存在问题、发展目标以及解决途径,从智能感知、智能建模、智能控制... 详细信息

制造业是我国实体经济的基础,而人工智能是经济发展的新引擎。本文基于以人工智能基础理论及应用为主题的"双清论坛"讨论内容,阐述了我国流程工业智能制造的发展现状、存在问题、发展目标以及解决途径,从智能感知、智能建模、智能控制、智能优化决策和智能安环五个方面介绍了人工智能技术助力流程制造业优化升级,打造新一代智能制造体系的核心内容,并给出了流程工业智能制造的发展愿景。

关键词：人工智能流程工业智能制造优化升级

基于模糊聚类和模糊支持向量机的湿法炼锌净化除钴过程建模

在线全文

学校读者我要写书评

暂无评论

高技术通讯 2010年第10期20卷 1068-1071页

作者：朱红求阳春华桂卫华中南大学信息科学与工程学院长沙410083

提出了一种基于模糊C-均值(FCM)聚类和模糊支持向量机(SVM)方法相结合的湿法炼锌净化除钴过程建模方法。该方法针对样本空间影响支持向量机泛化性能和样本数量影响计算复杂度的问题,首先采用模糊聚类将学习样本分类,然后在各个类的样本... 详细信息

提出了一种基于模糊C-均值(FCM)聚类和模糊支持向量机(SVM)方法相结合的湿法炼锌净化除钴过程建模方法。该方法针对样本空间影响支持向量机泛化性能和样本数量影响计算复杂度的问题,首先采用模糊聚类将学习样本分类,然后在各个类的样本空间内采用模糊支持向量机进行训练,并对各支持向量机模型的输出加权作为过程模型的输出。以净化除钴过程生产数据进行实验验证的结果表明,该方法明显减少了模型的训练时间,模型具有精度高、泛化性能好等特点,可以用于净化过程的优化控制。

关键词：净化除钴过程模糊聚类支持向量机(SVM)

基于模糊LS-SVM的净化过程钴离子浓度软测量

在线全文

学校读者我要写书评

暂无评论

仪器仪表学报 2009年第6期30卷 1224-1227页

作者：朱红求阳春华桂卫华中南大学信息科学与工程学院长沙410083

针对锌湿法冶炼净化过程的复杂性,提出一种基于模糊最小二乘支持向量机方法的净化过程钴离子浓度软测量模型。考虑到不同时期以及不同区域样本数据所起的作用不一样,提出一种综合模糊加权函数,有效提高了最小二乘支持向量机的性能。以... 详细信息

针对锌湿法冶炼净化过程的复杂性,提出一种基于模糊最小二乘支持向量机方法的净化过程钴离子浓度软测量模型。考虑到不同时期以及不同区域样本数据所起的作用不一样,提出一种综合模糊加权函数,有效提高了最小二乘支持向量机的性能。以净化过程生产数据进行实验验证分析,计算结果表明:模糊最小二乘方法性能优于传统最小二乘方法,模型具有精度高、泛化性能好等特点,钴离子浓度软测量结果满足现场工艺参数要求,可以作为过程信息用于净化过程的优化控制。

关键词：净化过程离子浓度软测量模糊LS-SVM

不确定性关联时滞大系统的分散鲁棒控制——LMI方法

在线全文

学校读者我要写书评

暂无评论

控制理论与应用 2001年第2期18卷 263-265,269页

作者：谢永芳桂卫华吴敏陈宁中南工业大学信息科学与工程学院长沙410083

对一类满足匹配条件的不确定性关联时滞大系统 ,根据Lyapunov稳定性原理 ,给出了其可分散状态反馈镇定的充分条件 ,即一组线性矩阵不等式 (LMI)有解 .在此基础上 ,通过求解一凸优化问题 ,提出了具有较小反馈增益的分散稳定化状态反馈控... 详细信息

对一类满足匹配条件的不确定性关联时滞大系统 ,根据Lyapunov稳定性原理 ,给出了其可分散状态反馈镇定的充分条件 ,即一组线性矩阵不等式 (LMI)有解 .在此基础上 ,通过求解一凸优化问题 ,提出了具有较小反馈增益的分散稳定化状态反馈控制律的设计方法 .

关键词：不确定性关联时滞大系统分散鲁棒控制 LMI 鲁棒镇定