Browsing by Author "Pani, Ajaya Kumar"

Now showing 1 - 20 of 29

Adaptive non-linear soft sensor for quality monitoring in refineries using Just-in-Time Learning—Generalized regression neural network approach
(Elsevier, 2022-04) Mohanta, Hare Krishna; Pani, Ajaya Kumar
Real time estimation of target quality variables using soft sensor relevant to time varying process conditions will be a significant step forward in effective implementation of Industry 4.0. Generalized Regression neural network (GRNN) has been used as a steady state quality monitoring soft sensor with reasonable estimation accuracy. However, the accurate prediction capability of GRNN has rarely been explored in a time varying environment. This article reports design of adaptive soft sensor using GRNN as a local model in Just-in-Time learning (JITL-GRNN) framework. The JITL-GRNN adaptive soft sensing technique is further investigated in various dimensions such as, the effect of different similarity index criteria and relevant dataset size on model prediction accuracy and model computation time. Performance of the proposed JITL-GRNN soft sensor is investigated by assessing its prediction accuracy on two benchmark industrial datasets. In addition, dynamic Non-linear autoregressive with exogenous inputs (NARX) neural network model is also developed and the performance of NARX model was compared with the proposed JITL-GRNN model. Results show that the JITL-GRNN adaptive soft sensor has at par or better prediction capability than the NARX model and many other models reported in literature.
Adaptive soft sensor design using a regression neural network and bias update strategy for non-linear industrial processes
(IOP, 2023-05) Rout, Bijay Kumar; Mohanta, Hare Krishna; Pani, Ajaya Kumar
Soft sensing of quality parameters in process industries has been an active area of research for the past two decades. To improve the performance of soft sensors in the scenario of time varying process states, adaptation capability is incorporated into the soft sensor model. In this work, recursive (R), sliding window (SW) and just-in-time learning (JITL) frameworks are used for adaptive soft sensor design. A rarely explored modeling technique in the adaptation framework, the generalized regression neural network (GRNN) is used as a local modeling strategy. A bias update procedure is applied during the model adaptation activity to improve the prediction accuracy. Further, the performances of the developed models are tested against input–output data dimension mismatch along with various concept drift phenomena by considering a different number of labeled samples for inputs and outputs. The proposed adaptation strategy is applied on two benchmark industrial processes. Simulation results show that the GRNN local modeling approach combined with the bias update strategy gives higher prediction accuracy than other adaptive soft sensors proposed in the literature. Moreover, GRNN local modeling strategy using SW adaptation mechanism has the least computation time among the three adaptation methods due to the use of a low number of samples for model development.
Application of Soft Sensors in Process Monitoring and Control: A Review
(SSRN, 2010-01-04) Mohanta, Hare Krishna; Pani, Ajaya Kumar
A major problem in product quality control in process industries is the difficulty of continuous online measurement of certain output variables especially related to composition. Although analytical instruments are available in some cases, significant time delays associated with most of such instruments make timely control difficult and sometimes impossible. Soft sensor is a modeling approach to estimate hard-to-measure process variables (primary variables) from easy-to-measure online process variables (secondary variables). The important steps of soft sensor development are collection of historical plant data for different variables and their processing, development of a model based on the available data and validation of the model. This paper presents the need and advantages of soft sensor implementation in process industries and does a critical review of various techniques available for data handling and modeling.
Asymmetric mixed matrix membranes with zeolite imidazolate frameworks (ZIF-8, ZIF-67, bimetallic ZIF-8/67) and polyethersulfone for high flux and high selective hydrogen separation
(Wiley, 2025-03) Pani, Ajaya Kumar; Kuncharam, Bhanu Vardhan Reddy
For producing high-purity hydrogen (H2) from hydrocarbon reforming, membrane-based separation can be used. In this study, mixed matrix polymer membranes using metal–organic framework (MOF) nanoparticles are explored to overcome the permeability-selectivity trade-off of traditional polymeric membranes. Highly permeable and highly H2 selective MMM using ZIF-8, ZIF-67, and bimetallic ZIF-8/67 MOFs were fabricated via a non-solvent induced phase inversion method by incorporating an intermediate solvent evaporation step. MMMs with 5, 10, and 15 wt.% of nanofillers loadings were prepared and tested for single gas (H2, CO2, CH4, and N2) permeability at 1–2 bar pressures. MMMs permeability and selectivities exceeded the Robeson upper bound (2008) for H2/CO2 separation, demonstrating the potential for obtaining high-purity hydrogen at low pressures. H2/N2 selectivity of 43.4, H2/CO2 selectivity of 27.86 for and H2/CH4 selectivity of 31.36 were obtained. Analytical techniques such as XRD, FTIR, and DSC were used to explain the transport mechanism in the MMMs. The cross-sectional structure and morphology of MMMs were analyzed with field-emission scanning electron microscope (FESEM) to provide insights into the membrane's porous structure.
Back Pressure Monitoring of Power Plant Condenser Using Multiple Adaptive Regression Spline
(Springer, 2020-02) Pani, Ajaya Kumar
This research work involves the application of multivariate adaptive regression spline (MARS) for estimating back pressure (p) created in a condenser of a coal-fired thermal power plant. MARS employs the plant load (L) and temperature of cooling water (T) as input variables. The output of the MARS is condenser back pressure p^. The designed MARS-based model gives equations for determination of p. Further, the MARS-generated objective function is optimized by randomized search cross validation. Simulation study shows that the accuracy of the reported MARS model is quite satisfactory for the prediction of back pressure.
Designing intelligence: harnessing soft sensors and advanced analytics in petroleum refining for industry 4.0
(CRC Press, 2025) Pani, Ajaya Kumar
Industry 4.0 is revolutionizing process industries by transforming vast data sets into actionable insights, improving efficiency, reducing downtime, and ensuring consistent product quality while meeting environmental standards. A key driver of this transformation is the development of intelligent, software-driven soft sensors, which provide real-time analytics for monitoring crucial process parameters traditionally measured through slower offline methods. This chapter explores the design and implementation of soft sensors, emphasizing the integration of artificial intelligence (AI) and machine learning (ML) in real-time quality monitoring within the petroleum refining industry. Over the past 15 years, these advanced sensors have significantly enhanced monitoring and quality assurance, enabling precise control over raw material composition and final product quality. By examining different soft sensor models and their applications, the chapter highlights their impact on industrial operations, as well as challenges such as managing large data streams and ensuring sensor reliability. Through real-world case studies, it demonstrates the practical benefits of soft sensors and their role in advancing the oil and gas sector toward Industry 4.0, paving the way for innovation and enhanced operational intelligence.
Development and comparison of neural network based soft sensors for online estimation of cement clinker quality
(Elsiever, 2013-01) Mohanta, Hare Krishna; Pani, Ajaya Kumar
The online estimation of process outputs mostly related to quality, as opposed to their belated measurement by means of hardware measuring devices and laboratory analysis, represents the most valuable feature of soft sensors. As of now there have been very few attempts for soft sensing of cement clinker quality which is mostly done by offline laboratory analysis resulting at times in low quality clinker. In the present work three different neural network based soft sensors have been developed for online estimation of cement clinker properties. Different input and output data for a rotary cement kiln were collected from a cement plant producing 10,000 tons of clinker per day. The raw data were pre-processed to remove the outliers and the resulting missing data were imputed. The processed data were then used to develop a back propagation neural network model, a radial basis network model and a regression network model to estimate the clinker quality online. A comparison of the estimation capabilities of the three models has been done by simulation of the developed models. It was observed that radial basis network model produced better estimation capabilities than the back propagation and regression network models.
Fault Detection of Complex Processes Using nonlinear Mean Function Based Gaussian Process Regression: Application to the Tennessee Eastman Process
(Springer, 2020-11-16) Pani, Ajaya Kumar
Process monitoring or fault detection and diagnosis have gained tremendous attention over the past decade in order to achieve better product quality, minimise downtime and maximise profit in process industries. Among various process monitoring techniques, data-based machine learning approaches have become immensely popular in the past decade. However, a promising machine learning technique Gaussian process regression has not yet received adequate attention for process monitoring. In this work, Gaussian process regression (GPR)-based process monitoring approach is applied to the benchmark Tennessee Eastman challenge problem. Effect of various GPR hyper-parameters on monitoring efficiency is also thoroughly investigated. The results of GPR model is found to be better than many other techniques which is reported in a comparative study in this work.
A hybrid soft sensing approach of a cement mill using principal component analysis and artificial neural networks
(IEEE, 2013) Pani, Ajaya Kumar; Mohanta, Hare Krishna
Soft sensors play an important role in predicting the values of unmeasured process variables from knowledge of easily measured process variables. Online estimation of particle size is vital for efficient control of a grinding circuit. Due to high energy consumption in cement grinding processes and unavailability of reliable hardware sensors for continuous monitoring, soft sensors have tremendous scope of application in cement mills. Modern cement plants are increasingly using vertical roller mills for clinker grinding. While there have been some works reported in the literature about modelling of ball mills, very few research work is available on vertical roller mill modelling. In the present work a PCA based neural network model of a cement mill is developed based on the actual plant data for estimation of cement fineness. Real time data for all process variables relevant to cement grinding process were collected from a cement plant having a clinker grinding capacity of 235 TPH. The collected raw industrial data were pre processed for outlier removal and missing value imputation. Principal component analysis of the input data was performed to transform the original variables to a less number of un correlated principal components. The selected principal component scores were divided to a training set and a validation set using Kennard-Stone subset selection algorithm. The training set was used to develop a back propagation neural network model which was subsequently tested with the validation set. Simulations results show satisfactory prediction capabilities of the developed model over that of linear regression and principal component regression models.
Independent component analysis application for fault detection in process industries: Literature review and an application case study for fault detection in multiphase flow systems
(Elsevier, 2023-03) Pani, Ajaya Kumar
In process industries, early detection and diagnosis of faults is crucial for timely identification of process upsets, equipment and/or sensor malfunctions. Machine learning techniques using process data can be used as efficient process monitoring tools and is an active research area in the past two decades. The technique of independent component analysis (ICA) is a viable alternative to the widely used principal component analysis method. In this article, the basic ICA technique, its advantages, limitations and the various improvements proposed over the years are reviewed. Further, a detailed survey of ICA based techniques for process monitoring is presented. Finally, the application of ICA along with selection of independent components by negentropy calculation and control limit and monitoring index calculation is illustrated by an industrial case study of multiphase flow system
Industrial process monitoring using support vector data description: A systematic review and application for fault detection in multiphase flow system
(Elsevier, 2025-12) Pani, Ajaya Kumar
In the era of Industry 4.0, machine learning based data-driven techniques are increasingly explored for industrial process monitoring. This article presents a review on application of support vector data description (SVDD) for industrial process monitoring followed by the design of SVDD model for fault detection in a multiphase flow system. In the review section, the basic technique, open design issues and a detailed survey on industrial applications are presented. In the application part, PRONTO benchmark multiphase flow dataset, is used to design SVDD model for detection of three faults: air leakage, air blockage and diverted flow. The Gaussian kernel parameter of the SVDD model is determined using particle swarm optimization (PSO) and the starting value for PSO, is obtained from literature provided analytical formula. Simulation of PSO-SVDD models shows promising results for fault detection in multiphase flow system.
Inferential Sensing of Output Quality in Petroleum Refinery Using Principal Component Regression and Support Vector Regression
(IEEE, 2017) Pani, Ajaya Kumar
In this research, linear regression (ordinary least square and principal component) and non-linear regression (standard and least square support vector) models are developed for prediction of output quality from sulphur recovery unit. The hyper parameters associated with standard SVR and LS-SVR are determined analytically using the guidelines proposed in the literature. The relevant input-output data for process variables are taken from open source literature. The training set and validation set are statistically designed from the total data. The designed training data were used for design of the process model and the remaining validation data were used for model performance evaluation. Simulation results show superior performance of the standard SVR model over other models.
An integrated approach combining randomized kernel PCA, Gaussian mixture modeling and ICA for fault detection in non-linear processes
(IOP, 2024) Pani, Ajaya Kumar
Principal component analysis (PCA) and independent component analysis (ICA), as well as their kernel extensions, have been widely applied in the past for industrial fault detection with Gaussian or non-Gaussian process data with linear or non-linear characteristics. Kernel-based techniques lead to computational complexity due to the high dimensionality of the dataset in the feature space. In this work, a randomization approach is used to obtain a low-rank approximation of the high-dimensional kernel matrix. A hybrid machine learning technique is proposed that integrates randomized kernel PCA (RKPCA) with ICA and Gaussian mixture modeling (GMM). The proposed approach, ICA-RKPCA-GMM, addresses the Gaussian and non-Gaussian characteristics of non-linear process data. Another hybrid algorithm combining three basic techniques of ICA, PCA and GMM is also developed (ICA-PCA-GMM). The fault detection performances of the proposed techniques (ICA-RKPCA-GMM and ICA-PCA-GMM) are compared with PCA, ICA, KPCA and combined ICA-PCA techniques by applying the techniques to two benchmark systems. Monitoring performances were evaluated by determining the false alarm rate and fault detection rate for different types of process and sensor faults. The simulation results show that the proposed ICA-RKPCA-GMM approach yields better results than individual ICA, PCA and KPCA techniques, the combined ICA-PCA and the proposed ICA-PCA-GMM technique.
Machine learning approaches for data-driven process monitoring of biological wastewater treatment plant: A review of research works on benchmark simulation model No. 1(BSM1)
(Springer, 2023-07) Pani, Ajaya Kumar
In the past decade, machine learning techniques have seen wide industrial applications for design of data-based process monitoring systems with an aim to improve industrial productivity. An efficient process monitoring system for wastewater treatment process (WWTP) ensures increased efficiency and effluents meeting stringent emission norms. Benchmark simulation model No. 1 (BSM1) provides a simulation platform to researchers for developing efficient data-based process monitoring, quality monitoring, and process control systems for WWTPs. The present article presents a review of all research works reporting applications of various machine learning techniques for sensor and process fault detection of BSM1. The review focuses on process monitoring of biological wastewater treatment process, which uses a series of aerobic and anaerobic reactions followed by secondary settling process. Detailed information on various parameters monitored, different machine learning techniques explored, and results obtained by different researchers are presented in tabular and graphical format. In the review, it was observed that principal component analysis (PCA) and its variants account for the maximum number of research works for process monitoring in WWTPs and there are very few applications of recently developed deep learning techniques. Following the review and analysis, various future scopes of research (such as techniques yet to be explored or improvement of results for a particular fault) are also presented. These information will assist prospective researchers working on BSM1 to take forward the research.
Neural Network Soft Sensor Application in Cement Industry: Prediction of Clinker Quality Parameters
(IEEE, 2011) Pani, Ajaya Kumar; Mohanta, Hare Krishna
A soft sensor tries to estimate difficult to measure quality parameters from the knowledge of easy to measure online process variables. Empirical approach of soft sensor development has gained much popularity recently due to availability of huge quantity of actual process data stored in the industrial database. In this work a soft sensor based on back propagation neural network has been developed for rotary cement kiln. For this purpose, data for all variables associated with rotary cement kiln were collected over a period of one month from a cement industry having a capacity of 10000 tons of clinker production per day. Data preprocessing of the raw data has been performed to remove the anomalies present in the original data. The processed data was used to develop the neural network model of the kiln. Model simulation produced quite satisfactory prediction of free lime, C 3 S, C 2 S and C 3 A.
Non-linear process monitoring using kernel principal component analysis: A review of the basic and modified techniques with industrial applications
(Springer, 2021-07-04) Pani, Ajaya Kumar
Timely detection and diagnosis of process abnormality in industries is crucial for minimizing downtime and maximizing profit. Among various process monitoring and fault detection techniques, principal component analysis (PCA) and its different variants are probably the ones with maximum applications. Because of the linearity constraint of the conventional PCA, some non-linear variants of PCA have been proposed. Among different non-linear variants of PCA, the kernel PCA (KPCA) has gained maximum attention in the field of industrial fault detection. This article revisits the basic KPCA algorithm along with different limitations of KPCA and the crucial open issues in design of KPCA based monitoring system. Different modifications proposed by different researchers are reviewed. Strategies adopted by various researchers for optimal selection of kernel parameter and number of principal components are also presented.
Non-negative matrix factorization combined with Fuzzy C-means enhanced k-nearest neighbor for fault detection and diagnosis in process industries
(Elsevier, 2026-01) Mohanta, Hare Krishna; Garg, Girish Kant; Pani, Ajaya Kumar
In the context of Industry 4.0, modern industrial processes generate high-dimensional, non-negative and potentially non-linear data streams, posing significant challenges for effective fault detection and diagnosis. Traditional statistical and multivariate techniques mostly assume restrictions such as a Gaussian distribution and linear relationships, which limit their use in real-world problems. This paper proposes a novel hybrid technique, Non-negative Matrix Factorization (NMF)–Enhanced Local Weighting Fuzzy C-Means (FCM) with Distance-Based k-Nearest Neighbors (NEFkNN), for fault detection. Initially, NMF is applied for dimensionality reduction. This is followed by FCM clustering, where cluster centers were refined with an enhanced local weighting (ELW) strategy. Detection threshold is determined by calculating the Euclidean distance between each sample and the enhanced cluster centers. A cluster-sensitive feature attribution method called Cluster-Aware Residual Contribution Analysis (CARCA) is proposed for fault diagnosis, which adjusts each feature's contribution to a fault by accounting for the local variance within its assigned cluster, enhancing interpretability. The NEFkNN technique was evaluated on two benchmark systems of a wastewater treatment plant(WWTP) and a continuous stirred tank reactor(CSTR) and achieved high fault detection rates and low false alarm rates. The diagnosis indicates that the fault is highly localized and attributable to a single process variable.
Online monitoring and control of particle size in the grinding process using least square support vector regression and resilient back propagation neural network
(Elsiever, 2015-05) Mohanta, Hare Krishna; Pani, Ajaya Kumar
Particle size soft sensing in cement mills will be largely helpful in maintaining desired cement fineness or Blaine. Despite the growing use of vertical roller mills (VRM) for clinker grinding, very few research work is available on VRM modeling. This article reports the design of three types of feed forward neural network models and least square support vector regression (LS-SVR) model of a VRM for online monitoring of cement fineness based on mill data collected from a cement plant. In the data pre-processing step, a comparative study of the various outlier detection algorithms has been performed. Subsequently, for model development, the advantage of algorithm based data splitting over random selection is presented. The training data set obtained by use of Kennard–Stone maximal intra distance criterion (CADEX algorithm) was used for development of LS-SVR, back propagation neural network, radial basis function neural network and generalized regression neural network models. Simulation results show that resilient back propagation model performs better than RBF network, regression network and LS-SVR model. Model implementation has been done in SIMULINK platform showing the online detection of abnormal data and real time estimation of cement Blaine from the knowledge of the input variables. Finally, closed loop study shows how the model can be effectively utilized for maintaining cement fineness at desired value.
Online monitoring of cement clinker quality using multivariate statistics and Takagi-Sugeno fuzzy-inference technique
(Elsiever, 2016-12) Pani, Ajaya Kumar; Mohanta, Hare Krishna
This article addresses the issue of outlier detection in industrial data using robust multivariate techniques and soft sensing of clinker quality in cement industries. Feed-forward artificial neural network (back propagation, radial basis function and regression neural network) and fuzzy inference (Mamdani and Takagi-Sugeno (T-S)) based soft sensor models are developed for simultaneous prediction of eight clinker quality parameters (free lime, lime saturation factor, silica modulus, alumina modulus, alite, belite, aluminite and ferrite). Required input-output data for cement clinkerization process were obtained from a cement plant with a production capacity of 10000 t of clinker per day. In the initial data preprocessing activity, various distance based robust multivariate outlier detection techniques were applied and their performances were compared. The developed soft-sensors were investigated for their performance by computing various statistical model performance parameters. Results indicate that the accuracy and computation time of the T-S fuzzy inference model is quite acceptable for online monitoring of clinker quality.
Optimizing process monitoring efficiency through control limit adjustment in multivariate ewma charts
(IEEE, 2024-10) Pani, Ajaya Kumar
Statistical process control charts have proven to be helpful in process monitoring. The majority of previous research on SPC charts has been on univariate scenarios. This study builds a multivariate exponentially moving average (MEWMA) chart to perform process monitoring in a continuous stirred tank reactor (CSTR). The normal and faulty data were obtained from the Simulink model of CSTR. Smoothing parameter of EWMA was optimized to maximize process monitoring efficiency. False alarm rate (FAR) and fault detection rate (FDR) were used for calculating the monitoring efficiency. A novel control limit calculation combining T2 and square prediction error (SPE) is proposed to increase the accuracy of MEWMA technique.