Application of online support vector regression for soft sensors

Soft sensors have been widely used in chemical plants to estimate process variables that are difficult to measure online. One of the crucial difficulties of soft sensors is that predictive accuracy drops due to changes in state of chemical plants. Characteristics of adaptive soft sensor models such as moving window models, just‐in‐time models and time difference models were previously discussed. The predictive accuracy of any traditional models decreases when sudden changes in processes occur. Therefore, a new soft sensor method based on online support vector regression (SVR) and the time variable was developed for constructing soft sensor models adaptive to rapid changes of relationships among process variables. A nonlinear SVR model with the time variable is updated with the most recent data. The proposed method was applied to simulation data and real industrial data, and achieved higher predictive accuracy than traditional ones even when time‐varying changes in process characteristics happen. © 2013 American Institute of Chemical Engineers AIChE J 60: 600–612, 2014     

Development of a vision‐based online soft sensor for oil sands flotation using support vector regression and its application in the dynamic monitoring of bitumen extraction

Extraction from oil sands is a crucial step in the industrial recovery of bitumen. It is challenging to obtain online measurements of process outputs such as bitumen grade and recovery. Online measurements are a prerequisite for innovating better process control solutions for process efficiency and cost reduction. We have developed a soft sensor to provide online measurements of bitumen grade and recovery in a flotation‐based oil sand extraction process. Continuous froth images were captured using a VisioFroth camera system on a batch flotation unit. A support vector regression (SVR) model with a Gaussian kernel was constructed to develop a soft sensor for bitumen grade and recovery using froth image features as the inputs. The model was trained and validated for batch flotation of different grades of oil sands ore at industry‐relevant process conditions. A Dean‐Stark analyzer was used to obtain offline grade and recovery measurements that were used to calibrate the soft sensor. Mean squared errors (MSE) of 62 and 74 were achieved for grade (%) and recovery (%), respectively, and this was obtained using 5‐fold cross validation. The developed soft sensor model has been applied successfully in the real‐time dynamic monitoring of flotation grade and recovery for different grades of ore and operating conditions.

     

Dynamic nonlinear soft sensor modelling method using linear slow feature analysis and least squares support vector regression for batch processes

Data-driven soft sensor models have been extensively utilized in industrial processes. Batch processes are usually employed to manufacture low-volume and high value-added products in chemical, materials, and pharmaceutical industries. The most distinctive features of batch process lie in nonlinear, repetition, and slow time varying characteristics. In this paper, a data-driven soft sensor modelling method based on linear slow feature analysis (LSFA) and least squares support vector regression (LSSVR) is proposed. In this method, LSFA was used to effectively capture the driving force behind the data features that change as slowly as possible. Then, a LSSVR model was constructed between the extracted slow feature variables and quality variables. Finally, a numerical example, industrial penicillin fermentation processes, and cobalt oxalate synthesis process were utilized to confirm the prediction accuracy and model reliability of the proposed approach.     

局部最小二乘支持向量机回归在线建模方法及其在间歇过程的应用

当间歇生产切换于不同的工艺条件时,由于新工况下的样本一般很少,且批次间存在着不确定性（由于原材料波动或过程动态特性波动等）,基于全局学习的建模方法（如最小二乘支持向量机回归,LSSVR）建立的模型泛化性能不强。将局部学习融入LSSVR中,提出一种局部LSSVR（local LSSVR, LLSSVR）的间歇过程在线建模方法。结合前一批次离线优化后的LSSVR参数,针对待预测新样本在线选择与之相关的近邻样本集并基于此进行建模。以建立青霉素发酵过程的菌体浓度为例,验证了LLSSVR算法能够从过程的第2个生产批次开始在线建立较准确的预报模型,较LSSVR有着更好的推广能力、适应性和鲁棒性。     

基于稀疏最小二乘支持向量机的软测量建模

针对传统最小二乘支持向量机非稀疏化解问题,提出了基于遗传算法的最小二乘支持向量机稀疏化及参数优化方法,稀疏化的基本思想是给训练样本赋予一个概率值,将概率值小于0.5的样本作为测试样本,从而将总的训练样本集分成测试样本集和保留的训练样本集。定义了包括稀疏率、训练误差及测试误差在内的适应度函数。种群个体的前N维表示每个样本对应的概率,后m维表示要优化的参数。通过选择、交叉和变异操作对所有参数进行整体优化,取适应度最小的个体对应的保留的训练样本及优化参数建立最小二乘支持向量机模型。并用该方法用于PX氧化过程4-CBA含量的软测量中,工业数据仿真结果表明,用本文提出的方法稀疏化率达到87%,核参数选取自动完成,与稀疏前建立的模型相比推广能力更高。     

基于深度集成支持向量机的工业过程软测量方法

基于支持向量机（SVM）的软测量建模方法已经在工业过程控制领域得到广泛应用,然而传统支持向量机直接针对原始测量变量建立模型,未能充分挖掘数据的内在特征信息以提高预测精度。针对该问题,本文提出一种基于深度集成支持向量机（DESVM）的软测量建模方法。该方法首先利用深度置信网络（DBN）来对数据进行深层次的信息挖掘,提取出数据的内在特征,然后引入基于Bagging算法的集成学习策略,构建基于深度数据特征的集成支持向量机模型,以提升软测量预测模型的泛化能力。最后通过数值系统和真实工业数据对方法进行应用分析,结果表明本文提出的方法能够有效提升支持向量机软测量模型的预测精度,能够更好地预测过程质量指标的变化。     

用于发酵过程在线建模的自适应局部最小二乘支持向量机回归方法

提出一种基于自适应局部学习的最小二乘支持向量机回归（LSSVR）在线建模方法。考虑样本间的距离和角度信息以获得更全面合理的相似样本集,推导了采用快速留一法在线优化模型参数的准则,并给出了发酵过程在线自适应模型选择的策略。以链激酶流加发酵过程为例,验证了所提出算法能够从过程的第2批次开始,同时对活性菌体浓度和链激酶浓度进行较准确的在线预报,较普通的局部LSSVR等建模方法具有更高的预报精度和自适应性。     

Applicability domain of soft sensor models based on one‐class support vector machine

Soft sensors are widely used to estimate process variables that are difficult to measure online. By using soft sensors, analyzer faults can be detected by estimation errors. However, it is difficult to detect abnormal data and determine the reasons because estimation errors increase not only due to analyzer faults but also due to variations caused by changes in the state of chemical plants. To separate those factors, we previously proposed to construct the relationships between distances to soft sensor models (DMs) and the accuracy of prediction of the models quantitatively and estimate the prediction accuracy of new data online. In this article, we used a one‐class support vector machine (OCSVM) to estimate data density and the output of an OCSVM as a DM. The proposed method was applied to real industrial data and the superiority of the proposed DM to the traditional ones was demonstrated by comparing their results. © 2013 American Institute of Chemical Engineers AIChE J, 59: 2046–2050, 2013     

基于最小二乘支持向量回归机的光管污垢特性预测

搭建了污垢实验系统以测得管壁温度和出、入口温度等参数,并将这些参数作为模型的输入变量,以污垢热阻值作为模型的输出变量,利用最小二乘支持向量回归机搭建了污垢预测模型,对光管的污垢特性进行了预测。一方面,通过与测量结果相比较,验证所搭建的模型是合理可行的;另一方面,通过对多次预测结果分析比较得出,该模型不但适用于流速、水浴温度、材质等参数为定值的情况,而且当这些参数发生改变时,该模型也是适用的。     

Adaptive local kernel-based learning for soft sensor modeling of nonlinear processes

Soft sensor techniques have been widely used to estimate product quality or other key indices which cannot be measured online by hardware sensors. Unfortunately, their estimation performance would deteriorate under certain circumstances, e.g., the change of the process characteristics, especially for global learning approaches. Meanwhile, local learning methods always only utilize input information to select relevant instances, which may lead to a waste of output information and inaccurate sample selection. To overcome these disadvantages, a new local modeling algorithm, adaptive local kernel-based learning scheme (ALKL) is proposed. First, a new similarity measurement using both input and output information is proposed and utilized in a supervised locality preserving projection technique to select relevant samples. Second, an adaptive weighted least squares support vector regression (AW-LSSVR) is employed to establish a local model and predict output indices for each query data. In AW-LSSVR, instead of using traditional cross-validation methods, the trade-off parameters are adjusted iteratively and the local model is updated recursively, which reduces the computational complexity a lot. The proposed ALKL is applied to an online crude oil endpoint prediction in an industrial fluidized catalytic cracking unit (FCCU) process. The experimental results demonstrate the high precision of our ALKL approach.     

Adaptive soft sensor model using online support vector regression with time variable and discussion of appropriate hyperparameter settings and window size

Soft sensors have been widely used in chemical plants to estimate process variables that are difficult to measure online. One crucial difficulty of soft sensors is that predictive accuracy drops due to changes in state of chemical plants. The predictive accuracy of traditional soft sensor models decreases when sudden process changes occur. However, an online support vector regression (OSVR) model with the time variable can adapt to rapid changes among process variables. One crucial problem is finding appropriate hyperparameters and window size, which means the numbers of data for the model construction, and thus, we discussed three methods to select hyperparameters based on predictive accuracy and computation time. The window size of the proposed method was discussed through simulation data and real industrial data analyses and the proposed method achieved high predictive accuracy when time-varying changes in process characteristics occurred.     

Illumination correction model with sine-cosine algorithm to optimize gray wolf population for least-squares support vector regression

The color of an object appears different from its true color when illuminated with light sources of different hues. To solve this problem, this article proposes a combination algorithm (SCA-GWO-LSSVR) based on the sine-cosine algorithm (SCA) and the gray wolf optimization (GWO) algorithm to optimize the regression prediction model of the least-squares support vector regression (LSSVR) algorithm. The performance of the traditional LSSVR is significantly affected by the penalty parameter (gamma) and the sig2 kernel function parameter. The proposed method uses the improved GWO algorithm to search the population to find the best LSSVR parameter solution. The proposed algorithm uses the SCA to create multiple random candidate solutions in population initialization to avoid blind initialization of the GWO algorithm. In the process of iterative optimization, the SCA is infiltrated, and its sine-cosine wave mathematical model is used to quickly identify the best outward or inward position of the gray wolf. Finally, the LSSVR combines the optimal sig2 kernel function parameters and penalty parameters (gamma) to obtain a highly versatile illumination correction model. The experimental results show that the fitting accuracy of the proposed method reaches 86.8%, which is 5% higher than that of the LSSVR algorithm alone.     

A novel unified correlation model using ensemble support vector regression for prediction of flooding velocity in randomly packed towers

Traditional empirical correlations and models have found insufficient to predict the flooding velocity accurately mainly because there are many kinds of random packings which exhibit different characteristics. In this work, a novel data-driven modeling method, i.e. ensemble least squares support vector regression (ELSSVR), is proposed to construct a unified correlation for prediction of the flooding velocity for packed towers with random packings. The flooding data are first clustered into several classes by the fuzzy c-means clustering algorithm. Then, several single LSSVR models can be trained using each sub-class of samples to capture the special characteristics. Moreover, a weighted least squares approach is adopted to integrate these single LSSVR models. Consequently, the ELSSVR model can extract the feature information of flooding data effectively and improve the prediction performance. The proposed ELSSVR method is applied to construct a unified correlation for prediction of the flooding velocity in randomly packed towers. The obtained results for several kinds of random packings demonstrate that the ELSSVR-based correlation can obtain better prediction performance, compared with the traditional semi-empirical correlations and artificial neural networks-based models. Finally, a database containing the modeling information of flooding velocity in randomly packed towers of China is provided for academic research.     

Weighted target feature regression neural networks based soft sensing for industrial process

There are often nonlinear and time-varying characteristics in industrial processes. These characteristics cause difficulty in measuring product quality online. To address these issues, a weighted target feature regression neural network (WTFAER) was proposed for soft sensor modelling in this paper. The Pearson correlation coefficient was calculated to assign corresponding weights to process variables and design a weighted objective function. A target feature regression network (TFAER) was constructed using target correlation autoencoder with fully connected layer. After that, the weighted reconstructed information was applied to the TFAER model to extract deep quality-related features and realize feature reuse. A deep network was formed by layered stacking to fully exploit the deep features for quality prediction. To make the proposed method domain adaptive, a maximum mean squared deviation (MMD) based regularization term was introduced in the loss function. Through the simulation experiments of debutanizer column and industrial polyethylene process, and compared with stacked autoencoder (SAE), variable-wise weighted stacked autoencoder (VW-SAE) and stacked target-related autoencoder (STAE) methods, the effectiveness and generalization performance of the proposed modelling method were verified.     

基于改进支持向量机的高炉一氧化碳利用率预测方法

高炉冶炼是一个具有非线性、大时滞、大噪声、分布参数等特征的高度复杂生产过程。针对目前高炉现场以焦比为能耗评价指标却无法提供实时指导的问题, 研究以一氧化碳利用率为能耗评价指标, 提出一种基于改进支持向量机的高炉一氧化碳利用率预测方法。首先分析高炉炼铁过程机理, 结合互信息法得出影响一氧化碳利用率的相关操作因素。然后鉴于生产数据含噪高的特点, 采用小波去噪方法去除数据噪声干扰, 并且利用灰色相对关联度分析方法对操作参数进行时序配准, 消除时滞影响, 建立高炉一氧化碳利用率预测模型。在建模过程中, 将自适应粒子群与支持向量机回归方法相结合, 以克服模型参数选择的随机性, 提高了模型预测精度。现场实际数据的预测结果表明所提出方法的有效性, 能够实时精确地预测高炉一氧化碳利用率, 为后续高炉的优化操作和节能减排提供了及时有效的决策支持。     

生物反应过程的在线支持向量机建模优化

针对生物反应过程具有较强的非线性、时变性,建立准确的机理模型较为困难,并且复杂的机理模型也无法用于在线控制和优化。将在线支持向量机和机理模型结合,提出串并联在线自校正混合建模方法。通过对典型生化过程谷氨酸的生产过程分析,找到影响谷氨酸浓度的关键参数;从现场历史数据中选取样本,建立基于在线向量机的软测量模型。实验结果表明该模型对谷氨酸浓度预测效果较好。     

Mixture robust L1 probabilistic principal component regression and soft sensor application

In this paper, the multivariate Laplace distribution (also called L1 distribution) is adopted to construct a robust probabilistic principal component regression model (MRPPCR-L1) under multiple operating modes. In the practical industrial chemistry process, outliers exist due to incorrect recording, disturbances, and process noises and might result in modelling distortion. To address this problem, Laplace distribution, instead of the Gaussian distribution in traditional methods, is introduced to reduce the negative influence of outliers. Moreover, probabilistic principal component regression is employed for dealing with the mixture modelling problem owing to its probabilistic property to determine the operating modes. The formulation of this approach is derived with the expectation maximum algorithm and the soft sensing model is also developed for prediction. Compared to the conventional method, a numerical example and the Tennessee Eastman process are used to demonstrate the robust modelling performance of the proposed method.     

基于支持向量机煤炭产地的鉴别研究

利用日照局历年来积累的出口煤炭的检测数据和出口企业自身检测的数据,用支持向量机技术分析其化学性质、物理性质、指标范围等情况,对检测的山东省出口煤炭的产地样品按决策树方法进行特性分类,并结合支持向量机方法实现了山东省15个煤炭出口产地或品牌的鉴别。该方法属于一种新的分类鉴别方法,为煤炭产地的鉴别提供了一个可借鉴的实例。     

Mixture semisupervised principal component regression model and soft sensor application

Traditionally, data‐based soft sensors are constructed upon the labeled historical dataset which contains equal numbers of input and output data samples. While it is easy to obtain input variables such as temperature, pressure, and flow rate in the chemical process, the output variables, which correspond to quality/key property variables, are much more difficult to obtain. Therefore, we may only have a small number of output data samples, and have much more input data samples. In this article, a mixture form of the semisupervised probabilistic principal component regression model is proposed for soft sensor application, which can efficiently incorporate the unlabeled data information from different operation modes. Compared to the total supervised method, both modeling efficiency and soft sensing performance are improved with the inclusion of additional unlabeled data samples. Two case studies are provided to evaluate the feasibility and efficiency of the new method. © 2013 American Institute of Chemical Engineers AIChE J 60: 533–545, 2014