With the rise of deep learning, there has been renewed interest within process industries to utilize data on large-scale nonlinear sensing and control problems. We identify key statistical machine learning techniques that have seen practical success in industries. To do so, we start with hybrid modeling provide a methodological framework underlying core application areas: soft sensing, optimization, control. Soft contains wealth industrial applications methods. quantitatively research...

The polymerization process produces industrially important products; hence, its monitoring and control are of paramount importance. However, the nonavailability real-time (on-demand) measurement quality variables gives rise to difficulties in achieving effective control. To overcome this hurdle, a novel multioutput soft sensor algorithm is proposed for simultaneous estimation four [rate esterification, degree polymerization, average molecular weight, melt viscosity index (MVI)] industrial...

Over the last ten years, we have seen a significant increase in industrial data, tremendous improvement computational power, and major theoretical advances machine learning. This opens up an opportunity to use modern learning tools on large-scale nonlinear monitoring control problems. article provides survey of recent results with applications process industry.

In this paper, a novel and effective stiction detection method is proposed by combining K-means clustering the moving window approach. As byproduct, offers an estimation for band in sticky control valves. The tested industrial case studies consisting of benchmark loops from oil sands industry. loops, results are compared with some existing methods. This comparison shows superior performance method. It noticed through simulation study that not only provides but also can detect severe valve or...

Oscillation in control loops is a frequent problem the process industries. These oscillations directly impact product quality, leading to decreased plant profit. Additionally, increase energy consumption and waste raw materials pose significant restriction on performance of operational unit. Therefore, it essential isolate that exhibit oscillations. In this work, two practical effective methods are proposed detect loops. Both aimed at detecting presence triangle-like shape "D vs variable...

Data-driven soft sensors have been widely used in industrial processes for over two decades. Industrial often exhibit nonlinear and time-varying behavior due to complex physical chemical mechanisms, feedback control, dynamic noise. Lately, variational autoencoder (VAE) has arisen as one of the most prevalent methods unsupervised learning intricate distributions. Despite being successful deep feature extraction uncertain data modeling, it still suffers from instability reconstruction error...

Abstract Control valve stiction is an industrial problem that often causes oscillations in process control loops. Oscillating loops are not capable of maintaining key variables near or at their desired values, thus yielding low‐quality products, inducing economic loss, and increasing environmental impacts. Therefore, it vital importance to detect valves. In this regard, the present work proposes a new method based on Markov transition field convolutional neural network (CNN) identify sticky...

Efficient control loop performance is pivotal in process industries to ensure optimal production, maintain product quality, and adhere regulatory standards. Poorly tuned controllers can disrupt these objectives, necessitating accurate detection methods. This paper introduces a novel approach for detecting poor controller tuning through advanced techniques: the Gramian Angular Field (GAF) Stack Auto-Encoder (SAE). Unlike manual methods, this automated system promptly identifies poorly...

Control valve, affected by stiction, causes closed-loop signals to experience oscillations, which ultimately leads a decrease in product quality, reduced plant throughput, and increased environmental footprint. Therefore, it is indispensable detect quantify stiction control valves. To accomplish this objective, the present work, four noninvasive practical simple methods are developed with help of statistical tests such as F-test, t-test (Student’s t-test), modified Hotelling T2-test, reverse...

A supervised multi-class classification method based on learning vector quantization (LVQ) neural network was proposed to classify tea samples of five commercial brands; Brook bond, Double-Diamond, Lipton, Lipton-Darjeeling and Marvel. Data required for classifier design were obtained by performing laboratory experiments with electronic tongue. Multi-class classifiers multilayer perceptron, weighted k-nearest neighbors Mahalanobis distance developed compare the results LVQ classifier. The...

Present study addresses the monitoring of drum boiler process. Methodologies; based on clustering time series data and moving window pattern matching have been proposed for detection fault in chosen Design databases were created process by simulating developed model. A modified k-means algorithm using similarity measure as a convergence criterion has adopted discriminating among pertaining to various operating conditions. The distance PCA combined along with approach used discriminate normal...

Control valve stiction is constantly encountered by virtually all process industries on a regular basis. Valve menace to the safe and economically optimal operation of industries. In present work, novel data-driven method developed detect sticky control valves in industrial loops. The rudimentary concepts statistics are foundation method. Rigorous testing proposed loops pertaining wide variety demonstrates superior detection capability over most existing methods. Apart from detection, offers...

In the present study, neural network (NN) based multivariable controllers were designed as a series of single input-single output (SISO) or multi variable SISO (MVSISO) utilizing classical decoupled process models. Multilayer feed forward networks (FFNN) used direct inverse (DINN) controllers, which dynamics process. To address disturbance rejection problems, IMC control architecture was proposed with suitable choice filter and transfer function. Multi input - (MIMO) non-linear processes...

Sticky control valve induced oscillations are the most common source of deteriorated loop performance. Although there many stiction detection algorithms, complexity remains as top reason for their restricted applications. This work proposes a new method to detect malfunctioning valves (due stiction) in process loops. The proposed embroils fitting sigmoid function (or logistic function) OP (controller output) and ΔPV (change variable (PV)). Effectiveness is assessed by application industrial...

Partial least squares technique has been in use for identification of the dynamics & control multivariable distillation process.Discrete input-output time series datawere generated by exciting non-linear process models with pseudo random binary signals.Signal to noise ratio was set 10 adding white data.The ARX as well FIR combination were used build up dynamic inner relations among scores data, which logically built framework PLS based controllers.In this work, also identified latent...

In the present work, a multifunction sensor is developed using convolutional neural network (CNN) and multiple linear regression to measure temperature pressure simultaneously. First, hardware built piezo-resistive material. When excited by electric signals, it produces an aggregated output (emf). A hybrid CNN MLR based approach devised extract values from output.