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

This paper introduces a novel soft sensor modeling method based on BDA-IPSO-LSSVM designed to address the issue of model failure caused by varying fermentation data distributions resulting from different operating conditions during the fermentation of different batches of Pichia pastoris. First, the problem of significant differences in data distribution among different batches of the fermentation process is addressed by adopting the balanced distribution adaptation (BDA) method from transfer learning. This method reduces the data distribution differences among batches of the fermentation process, while the fuzzy set concept is employed to improve the BDA method by transforming the classification problem into a regression prediction problem for the fermentation process. Second, the soft sensor model for the fermentation process is developed using the least squares support vector machine (LSSVM). The model parameters are optimized by an improved particle swarm optimization (IPSO) algorithm based on individual differences. Finally, the data obtained from the Pichia pastoris fermentation experiment are used for simulation, and the developed soft sensor model is applied to predict the cell concentration and product concentration during the fermentation process of Pichia pastoris. Simulation results demonstrate that the IPSO algorithm has good convergence performance and optimization performance compared with other algorithms. The improved BDA algorithm can make the soft sensor model adapt to different operating conditions, and the proposed soft sensor method outperforms existing methods, exhibiting higher prediction accuracy and the ability to accurately predict the fermentation process of Pichia pastoris under different operating conditions.

The traditional template matching strategy of optical and synthetic aperture radar (SAR) is sensitive to the nonlinear transformation between two images. In some cases, the optical and SAR image pairs do not conform to the affine transformation condition. To address this issue, this study presents a novel template matching strategy which uses the One-Class Support Vector Machine (SVM) to remove outliers. First, we propose a method to construct the similarity map dataset using the SEN1-2 dataset for training the One-Class SVM. Second, a four-step strategy for optical and SAR image registration is presented in this paper. In the first step, the optical image is divided into some grids. In the second step, the strongest Harris response point is selected as the feature point in each grid. In the third step, we use Gaussian pyramid features of oriented gradients (GPOG) descriptor to calculate the similarity map in the search region. The trained One-Class SVM is used to remove outliers through similarity maps in the fourth step. Furthermore, the number of improve matches (NIM) and the rate of improve matches (RIM) are designed to measure the effect of image registration. Finally, this paper designs two experiments to prove that the proposed strategy can correctly select the matching points through similarity maps. The experimental results of the One-Class SVM in dataset show that the One-Class SVM can select the correct points in different datasets. The image registration results obtained on the second experiment show that the proposed strategy is robust to the nonlinear transformation between optical and SAR images.

Huberized one-class support vector machine with truncated loss function in the primal space

Image segmentation is of great importance in the field of image processing. A wide variety of approaches have been proposed for image segmentation. However, SAR image segmentation poses a difficult challenge owing to the high levels of speckle noise. In this paper, we proposed a SAR image segmentation method based on one-class support vector machines (SVM) to solve this problem. One-class SVM and two-class SVM for segmentation is discussed. One-class way is a kind of unsupervised learning, and one-class SVM based segmentation method reduces greatly human interactions, while yielding good segmentation results compared to two-class SVM based segmentation method. The segmentation results based on SVM are also compared to threshold method and adaptive threshold method. Experimental results demonstrate that the proposed method works well for image segmentation while reducing the speckle noise.

The main objective of structural health monitoring is to provide reliable information about the health state of the critical structures by implementing a damage characterization strategy to detect the presence of damage, location, severity as possibly failure prediction as soon as the damage occurs. This paper presents a robust approach to detect and characterize a gradually evolving damage based on time responses data captured from a steel reinforced concrete structure. The presented method is in the context of unsupervised and non-model-based approaches, hence, there is no need for any representative numerical/finite element model of the structure to be built. In this work, we propose one-class support vector machine as an anomaly detection method. One-class support vector machine fits well for damage diagnosis in structural health monitoring since there may exist many damaged patterns and one-class support vector machine can detect all of them as anomalies. To demonstrate the feasibility of the method in the detection and assessment of a gradually evolving deterioration, a test bed was established to replicate a concrete jack arch which is a main structural component on the Sydney Harbour Bridge – one of Australia’s iconic structures, see Figure 1. The structure is a concrete cantilever beam with an arch section which is located on the eastern side of the bridge underneath the bus lane. It is assumed that the structure is subjected to Gaussian white noise excitation. The response of the structure was collected by 10 uniaxial accelerometers placed in frontal face of the jack arch. The measurement was conducted for 2 seconds at a sampling rate of 8 kHz, resulting in 16000 samples for each event. Figure 1 Illustration of the test bed. https://s3-euw1-ap-pe-df-pch-content-public-p.s3.eu-west-1.amazonaws.com/9781315207681/cd556cd4-4dcf-4efe-8e29-56fc67b8bfbd/content/fig321_1.jpg/> A crack is introduced in the structure using a cutting saw and its length is progressively increased in four stages while the depth was constant; these four damage cases correspond to less than 0.5% reduction in the first three modes of the structure. Figure 2 presents the average of all decision values obtained from all sensors for healthy state and different damage states using one-class support vector machine. As seen, the method is able to successfully separate the healthy state from the damage states and as the severity of damage increases, higher negative decision values are obtained. Figure 2 Damage detection using decision values of all test events. https://s3-euw1-ap-pe-df-pch-content-public-p.s3.eu-west-1.amazonaws.com/9781315207681/cd556cd4-4dcf-4efe-8e29-56fc67b8bfbd/content/fig321_2.tif/>

A novel hybrid undersampling method for mining unbalanced datasets in banking and insurance

Resting state functional magnetic resonance imaging (fMRI) aims to measure baseline neuronal connectivity independent of functional tasks. Most existing network detection methods rely on a fixed threshold to identify functionally connected voxels under the resting state. Due to fMRI non-stationarity, a fixed threshold cannot adapt to inter-session and inter-subject variation. In this work, a new method is proposed for resting state fMRI data analysis. Specifically, the resting state network mapping is formulated as an outlier detection process that is implemented using one-class support vector machine (SVM). The results are refined by using a spatial-feature domain prototype selection method and two-class SVM reclassification. The final decision on each voxel is made by comparing its probabilities of functionally connected and unconnected. The proposed method was evaluated using synthetic and experimental fMRI data. A comparison study was also performed with independent component analysis (ICA) and correlation analysis. The experimental results show that the proposed method can provide comparable or better network detection performance than ICA and correlation analysis. The method is potentially applicable to various resting state quantitative fMRI studies.

High-precision prediction of material properties is an advanced direction of hot rolled product technology in manufacturing processes. However, complex disturbance usually produces wrong or missing data in the acquisition and entry stages. The long process and the diverse operating conditions bring great challenges to constructing soft sensors for the hot rolling process. To address this problem, a new time-series soft sensor model framework is proposed for mechanical properties prediction. Firstly, in order to process the raw dataset, the isolation forest algorithm is used to remove the outliers, and the random forest algorithm is used to fill in missing values. Secondly, the auxiliary variables are extracted by the combination of the XGBoost algorithm and prior mechanistic knowledge. Then, a DA-CBGRU-Seq2Seq time-series model based on a two-stage attention mechanism is designed for soft sensor modeling. In the encoder stage, external factors are introduced by an entity embedding layer. A combination structure of bidirectional gated recurrent units and convolutional layers is designed for extracting and learning the input features adaptively with the input attention mechanism. In the decoder stage, the long-term time dependence of the time series is learned by the temporal attention mechanism. Finally, the reliability of the proposed sensor model is compared and verified by the hot rolling process. The results indicate that the proposed method has an accuracy of over 95% in predicting the mechanical properties of a real hot-rolling process dataset, and has better performance compared to other comparative soft sensor prediction methods.

Recently, One Class Support Vector Machines (OCSVM) have been the subject of much research. This paper introduces a novel pattern classification approach for Broken Rotor Bar (BRB) detection in Induction Motors (IM), which combines Stationary Wavelet Packet Transform (SWPT) and OCSVM. Among all the kernels available to OCSVM, wavelet kernels are tuned to improve accuracy and detection time of fault. The SWPT is applied for the relevant data under lower sampling rate and a reduced number of samples. The experimental results on different load levels prove that the proposed method has a better performance than other methods.

This paper proposes a new texture image retrieval scheme based on contourlet transform and support vector machines (SVMs). In the scheme, the energies and the generalized Gaussian distribution (GGD) parameters are used to represent the contourlet subband features. Using the representations, a two-run SVM retrieval algorithm which employs an one-class SVM followed by a two-class SVM is proposed to carry out the perceptual similarity measurement. For the query image, the one-class SVM is used to obtain the effective initial training set with positive and negative samples. Using these initial samples, the two-class SVM is applied to refine on the image classification subject to the user’s relevance feedback. Compared with existing texture image retrieval methods, the proposed retrieval scheme is demonstrated respectively to be effective on the VisTex database of 640 texture images and the Brodatz database of 1760 texture images. Experimental results have shown that the proposed retrieval scheme can attain 99.38% and 98.07% of the average rates respectively for the two databases.KeywordsContourlet transformgeneralized Gaussian distributionsupport vector machinerelevance feedbacktexture image retrieval

Structural health monitoring has been increasingly used due to the advances in sensing technology and data analysis, facilitating the shift from time-based to condition-based maintenance. This work is part of the efforts which have applied structural health monitoring to the Sydney Harbour Bridge – one of Australia’s iconic structures. It combines dimensionality reduction and pattern recognition techniques to accurately and efficiently distinguish faulty components from well-functioning ones. Specifically, random projection is used for dimensionality reduction on the vibration feature data. Then, healthy and damaged patterns of bridge components are learned in the lower dimensional projected space using supervised and unsupervised machine learning methods, namely, support vector machine and one-class support vector machine. The experimental results using data from a laboratory-based building structure and the Sydney Harbour Bridge showed high feasibility of applying machine learning techniques to dimensionality reduction and damage detection in structural health monitoring. Random projection combined with support vector machine significantly reduces the computational time while maintaining the detection accuracy. The proposed method also outperformed popular dimensionality reduction techniques. The computational time of the method using random projection can be more than 200 times faster than that without using dimensionality reduction while still achieving similar detection accuracy.

In order to fully utilize the local geometric information of the given training set consisting of the normal data, locality correlation preserving (LCP) is introduced into the traditional one-class support vector machine (OCSVM). The proposed method, named as locality correlation preserving based one-class support vector machine (LCP-OCSVM), inherits the merits of LCP and OCSVM. It can keep locality correlation of the normal data and margin maximization between the normal data and the origin in the high-dimensional feature space. Experimental results on one synthetic data set and ten benchmark data sets demonstrate that the proposed method is superior to the traditional OCSVM and two related approaches.

In this paper, we propose a method by engaging the One Class Support Vector Machine (OC-SVM) in the identification of Diffractive Optically Variable Images (DOVIs). OC-SVM, as a special SVM, can solve the problems of high-dimensional data sets and small sample size (SSS) with positive and negative unbalance training data. Image feature matrix is built by extracting image features from texture aspects. OC-SVM can be trained with the high-dimensional matrix directly, and does not have to reduce the dimensionality of feature matrix as the usual methods. The experiment results show the effectiveness of the proposed approach against Linear Discriminant Analysis. Considering time cost and correct classification rate, OC-SVM is suitable for the identification of DOVIs.

In recent years, deep learning has been applied to build soft sensor models. Compared to the classical latent variable models, the model based on deep learning has a good performance in tackling nonlinearity in process data. However, static soft sensor based on deep neural network (SSSDNN) fails to take into account the dynamic characteristics, which are unavoidable in some applications. To improve the performance of the soft sensor, a novel dynamic soft sensor model based on impulse response template and deep neural network (DSSDNN) is proposed by means of Wiener structure, and an iterative algorithm will be used to train this novel model. A case study based on real debutanizer column data demonstrates the desirable prediction performance of DSSDNN and shows that the DSSDNN is able to obtain better approximation accuracy than the static soft sensor model in nonlinear processes with dynamic characteristics.

Erythromycin fermentation process is a typical microbial fermentation process. Soft sensors can be used to estimate biomass of Erythromycin fermentation process for their relative low cost, simple development, and ability to predict difficult-to-measure variables. However, traditional soft sensors, e.g. artificial neural network (ANN) soft sensors, support vector machine (SVM) soft sensors, etc., cannot represent the uncertainty (measurement precision) of outputs. That results in difficulties in practice. Gaussian process regression (GPR) provides a novel framework to solve regression problems. The output uncertainty of a GPR model follows Gaussian distribution, expressed in terms of mean and variance. The mean represents the predicted output. The variance can be viewed as the measure of confidence in the predicted output that distinguishes the GPR from NN and SVM soft sensor models. We proposed a systematic approach based on GPR and principal component analysis (PCA) to establish a soft sensor to estimate biomass of Erythromycin fermentation process. Simulations on industrial data from an Erythromycin fermentation process show the proposed GPR soft sensor has high performance of modeling the uncertainty of estimates.