Prediction Performance Of Soft Sensor Research Articles

Probabilistic stationary subspace regression model for soft sensing of nonstationary industrial processes

AbstractBenefitting from the development of industrial intelligence, data‐driven soft sensors have been widely applied in industrial processes in recent years. Driven by fluctuations in raw material quality, varying loads, and uncertain disturbances, industrial processes are often characterized by nonstationary properties. However, the nonstationary characteristics of the process are not considered in traditional data‐driven methods, leading to the poor prediction performance of soft sensors. Meanwhile, numerous nonstationary modelling methods have been proposed for process monitoring, with probabilistic stationary subspace analysis (PSSA) showing significant application potential. Unfortunately, PSSA has not yet been used to develop effective soft sensors for nonstationary processes. Therefore, the PSSA model is extended to the regression form (PSSR), and a corresponding soft sensing model is developed in this paper. Unlike previous approaches, the proposed PSSR can offer a suitable description of nonstationary processes and fully capture the mathematical correlation between input and output variables. Finally, the prediction performance of PSSR is verified through case studies on a numerical example and penicillin fermentation process. The experimental results indicate that PSSR can provide a superior solution for soft sensing modelling of nonstationary industrial processes.

Research on adaptive soft sensing modeling method of photovoltaic power generation process based on online semi-supervised selective ensemble learning

Soft sensors often estimate product quality variables that are difficult to measure in real time. However, strong nonlinearity and dynamic and time-varying characteristics lead to poor prediction performance of soft sensors for the photovoltaic power generation process. This paper proposes an adaptive soft sensor model construction strategy based on extreme learning machine online semi-supervised selective ensemble learning (SEMI-SEL-ELM). Firstly, a local learning strategy based on the Spatio-temporal criterion is proposed to limit the impact of nonlinear dynamic characteristics. The unlabeled samples are then examined to provide a semi-supervised sample set reconstruction. After that, using the extreme learning machine method, a local soft sensor model is created. Furthermore, the statistical hypothesis test is used to create the statistical information of the test samples, and the t-test is utilized to acquire the quantitative local model satisfaction. Additionally, the adaptive computation of mixed weights of distinct sub-models is performed using the improved information entropy. The black hole algorithm determines the parameters of this approach automatically. Finally, the method is applied to the historical data set of a photovoltaic power station in Australia. The results show that the method effectively deals with nonlinear, dynamic, and time-varying regression problems in PV power prediction.

Streaming parallel variational Bayesian supervised factor analysis for adaptive soft sensor modeling with big process data

Time-varying and state shifting are two of the main process factors that cause poor prediction performance of soft sensors. Adaptive soft sensor is commonly an alternative practice to ensure high predictive accuracy. However, the large scale of process data often leads to inefficiency of model updating. In this paper, a streaming variational Bayesian supervised factor analysis (S-VBSFA) model is first proposed to capture the process time-varying and state shifting features through online updating of the posterior of model parameters. During the updating process, the symmetric Kullback–Leibler (SKL) divergence is utilized to determine priors of the next variation Bayesian inference. To improve the modeling efficiency for large-scale process data, the parallel computing strategy is further applied to the streaming model. As a result, the proposed streaming parallel VBSFA (SP-VBSFA) algorithm not only relieves the computing pressure of modeling big process data, but also improves the prediction accuracy and further reduces the tracking time delay for process variations. Two case studies demonstrate the superiority of the proposed method, compared to conventional methods.

Probabilistic slow feature analysis‐based representation learning from massive process data for soft sensor modeling

Latent variable (LV) models provide explicit representations of underlying driving forces of process variations and retain the dominant information of process data. In this study, slow features (SFs) as temporally correlated LVs are derived using probabilistic SF analysis. SFs evolving in a state‐space form effectively represent nominal variations of processes, some of which are potentially correlated to quality variables and hence help improving the prediction performance of soft sensors. An efficient expectation maximum algorithm is proposed to estimate parameters of the probabilistic model, which turns out to be suitable for analyzing massive process data. Two criteria are also proposed to select quality‐relevant SFs. The validity and advantages of the proposed method are demonstrated via two case studies. © 2015 American Institute of Chemical Engineers AIChE J, 61: 4126–4139, 2015

Soft Sensor Model Maintenance: A Case Study in Industrial Processes

One of the challenges of utilizing soft sensors is that their prediction accuracy deteriorates with time due to multiple factors, including changes in operating conditions. Once soft sensors are designed, a mechanism to maintain or update these models is highly desirable in industry. This paper proposes an index that can monitor the prediction performance of soft sensor models and provide guidance about when to update these models. In the proposed approach, a Kalman filter based model mismatch index is developed to monitor the prediction performance of soft sensors with the support of traditional process monitoring indexes, T2 and SPE. Then, the soft sensor model can be updated through partial least squares (PLS) regression by using samples from the off-line training set and new process conditions. The proposed online update method is applied to an industrial process case study and the effectiveness of the proposed approach is demonstrated by comparing with traditional recursive partial least squares (RPLS).

Investigation of a novel soil analysis method in agricultural areas of Çarşamba plain for fertilizer recommendation

In this study, a novel soil analysis method for fertilization recommendation was developed and validated with 161 soil samples taken from Turkey - Carsamba plain for determination of potassium as a plant nutrient. In conventional soil analysis methods, available potassium (K) nutrient was determined by ammonium acetate extraction with flame photometer. In this study an alternative to existing method was proposed by developing extraction solutions suitable for interference dynamics of ion selective electrodes in a flow injection setup. Flow injection analysis system was optimized and K ion concentration of 161 soil samples taken from Turkey – Carsamba plain was determined with potentiometrically. For the same soil samples, K + ion concentration was determined with ammonium acetate extraction using flame photometer in parallel. Fertilization recommendations for potassium was calibrated on ammonium acetate extraction based measurements. In order to evaluate available potassium nutrient analysis results from new generation soil analysis method in fertilization recommendation process, a correlation model is required for relating new generation method results to conventional method results. An artificial neural network based soft sensor system was developed for this task. Potentiometric K + ion measurement of soil sample in flow injection analysis system was presented as input to soft sensor system. Soft sensor predicted available K in soil sample based on artificial neural network model which can be used in fertilizer recommendation. Prediction performance of soft sensor was validated with experimental data and fitted with high correlation coefficient (R2= 0.902). Experimental studies have shown that K determined by potentiometric measurements can be used in fertilization recommendations in Carsamba plain by using soft sensor approach.

A Multi-model Approach for Soft Sensor Development Based on Feature Extraction Using Weighted Kernel Fisher Criterion

Multi-model approach can significantly improve the prediction performance of soft sensors in the process with multiple operational conditions. However, traditional clustering algorithms may result in overlapping phenomenon in subclasses, so that edge classes and outliers cannot be effectively dealt with and the modeling result is not satisfactory. In order to solve these problems, a new feature extraction method based on weighted kernel Fisher criterion is presented to improve the clustering accuracy, in which feature mapping is adopted to bring the edge classes and outliers closer to other normal subclasses. Furthermore, the classified data are used to develop a multiple model based on support vector machine. The proposed method is applied to a bisphenol A production process for prediction of the quality index. The simulation results demonstrate its ability in improving the data classification and the prediction performance of the soft sensor.