欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
对乙烯裂解炉建立实时监控模型具有重要的现实意义，而传统的多元统计过程监控方法都是假设过程处于单一工况下，而随着过程参数（进料负荷、产品组分等）的改变，工况也随之改变，传统方法便不再适用。本文针对工业过程中的多工况问题，提出了一种基于自适应模糊聚类的多模型过程监控方法，该方法可以减少监控方法对过程知识的依赖性，并且能够适应实际工业过程的非高斯性和非线性特征。首先对影响工况的过程变量利用自适应模糊聚类进行工况划分，然后对每种工况的建模数据分别利用最大方差展开（MVU）提取低维信息，再用支持向量数据描述（SVDD）建立多模型过程监控模型，最后再利用相应的统计指标进行过程监控。将上述方法应用在乙烯裂解炉上，并与基于高斯混合模型的多PCA方法（GMM-MPCA）进行了比较。仿真实验中，监控对裂解炉运行影响最大的33个变量，根据聚类有效性指标，将数据划分为5类时可以得到最佳的聚类效果。通过实验，将33维建模数据降到20维时误报率最小。仿真结果表明该方法在对非线性和非高斯性过程的监控上，能达到很好的效果，误报率和检测率均优于GMM-MPCA方法。
英文摘要：
Modem manufacturing equipment has the characteristics of large scale, high complexity and multi-variable. It usually runs under closed-loop control. Conducting fault detection and diagnosis of this equipment at early time can reduce downtime, increase process safety and cut manufacturing costs. With the development of computers and intemet, a large amount of process data are collected and stored into the database which can be used to carry out fault detection and diagnosis. Traditional methods for process monitoring usually works under the assumption that the process only has one steady state mode, but when applied to the process with multi-state modes these methods don＇t have a good monitoring result. In order to solve this problem, this paper proposes a multiple models process monitoring method based on self-adaptive fuzzy c means cluster. It can have a good monitoring result of the process with multiple operation modes and reduce dependence on the process knowledge. First, classify the mode according to the process variables which have influence on the operating mode. Second, use Maximum Variance Unfolding （MVU） to reduce the dimension of each child-model data, and then build the Support Vector Data Description （SVDD） process monitoring model. At last, corresponding monitoring indices are constructed to detect process fault. The proposed method has been applied to the process monitoring of ethylene cracking furnace to show its efficiency. Gaussian Mixture Model-Multiple PCA （GMM-MPCA） method is chosen to be compared with. In the simulation, 33 variables which have main effect on the cracking furnace are chosen to monitor. According to the cluster validity index, the model data are divided into 5 sub-models. When the data＇s dimensions are reduced to 20, minimum false detection rate can be reached. The simulation result indicates that the proposed method has good monitoring result when applied to the nonlinear and nonGaussian process. The false detection rate and detection rate are all better tha
同期刊论文项目
同项目期刊论文
Modified Self-adaptive Immune Genetic Algorithm for Optimization ofCombustion Side Reaction of p-Xyl
Optimization of p-xylene oxidation reaction process based onself-adaptive multi-objective differenti
Adaptive sampling for surrogate modeling with artificial neural network and its application in an in
Dynamic Optimization of Chemical Engineering Problems Using a Control Vector Parameterization Method
A Hybrid Algorithm Based on Differential Evolution and Group Search Optimization and Its Application
Optimization of p-xylene oxidation reaction process based on self-adaptive multi-objective different
Dynamic Optimization of Chemical Processes Using Nonuniform Discretization Control Vector Parameteri
Dynamic Optimization of Chemical Engineering Problems Using Affinity Propagation Based Estimation of
A New Sample Selection and Modeling Method Based on Near-Infrared Spectroscopy and Its Industrial Ap
Novel Control Vector Parameterization Method with Differential Evolution Algorithm and its Applicati
Synchronization error estimation and controller design for delayed Lur’e systems with parameter mism
Modified Self-adaptive Immune Genetic Algorithm for Optimization of Combustion Side Reaction of p-Xy
Development of a free radical kinetic model for industrial oxidation of p-Xylene based on ANN and AI
Multiobjective optimization of the pseudo-dynamic operation of naphtha pyrolysis process using surro
Outlet Temperature Correlation and Prediction of Transfer Line Exchanger in an Industrial Steam Ethy
Application of the COSMO-RS and COSMO-SAC models for the predictions of partition coefficients, vapo
A novel adaptive algorithm with near-infrared spectroscopy and its application in online gasoline bl
Novel hybrid evolutionary algorithm for dynamic optimization problems and its application in an ethy
Comprehensive Simulation and Optimization of an Ethylene Dichloride Cracker Based on the One-Dimensi
Synthesis and Optimization of the Utility System using parameter adaptive differential evolution alg
Gaussian process assisted coevolutionary estimation of distribution algorithm for computationally ex
期刊信息
位置：