A review of ultrasonic sensing and machine learning methods to monitor industrial processes

Supervised machine learning techniques are increasingly being combined with ultrasonic sensor measurements owing to their strong performance. These techniques also offer advantages over calibration procedures of more complex fitting, improved generalisation, reduced development time, ability for continuous retraining, and the correlation of sensor data to important process information. However, their implementation requires expertise to extract and select appropriate features from the sensor measurements as model inputs, select the type of machine learning algorithm to use, and find a suitable set of model hyperparameters. The aim of this article is to facilitate implementation of machine learning techniques in combination with ultrasonic measurements for in-line and on-line monitoring of industrial processes and other similar applications. The article first reviews the use of ultrasonic sensors for monitoring processes, before reviewing the combination of ultrasonic measurements and machine learning. We include literature from other sectors such as structural health monitoring. This review covers feature extraction, feature selection, algorithm choice, hyperparameter selection, data augmentation, domain adaptation, semi-supervised learning and machine learning interpretability. Finally, recommendations for applying machine learning to the reviewed processes are made.

Combined Longitudinal and Surface Acoustic Wave Analysis for Determining Small Filling Levels in Curved Steel Containers

Measurement of the small filling levels in closed steel container systems is still a challenge. Ultrasound, however, is a sensitive and non-invasive technique and is suitable for online monitoring. This study describes a new ultrasonic sensor system for sensing small filling levels using longitudinal and surface acoustic wave analysis. The sensor system consists of one transducer for the longitudinal wave analysis and two transducers for the longitudinal and surface acoustic wave analysis. All transducers were mounted to the outer wall of the steel container, ensuring non-invasiveness, and a filling level ranging from 0 to 5 cm was investigated. Combining both approaches, a consistent determination of small filling levels was achieved for the entire measuring range (R² = 0.99).

Comparison of various classification techniques for supervision of milk processing

Detecting the types of anomalies that can occur throughout the milk processing process is an important task since it can assist providers in maintaining control over the process. The Raman spectrometer was used in conjunction with several classification approaches—linear discriminant analysis, decision tree, support vector machine, and k nearest neighbor—to establish a viable method for detecting different types of anomalies that may occur during the process—temperature and fat variation and added water or cleaning solution. Milk with 5% fat measured at 10°C was used as the reference milk for this study. Added water, cleaning solution, milk with various fat contents and different temperatures were used to detect abnormal conditions. While decision trees and linear discriminant analysis were unable to accurately categorize the various type of anomalies, the k nearest neighbor and support vector machine provided promising results. The accuracy of the support vector machine test set and the k nearest neighbor test set were 81.4% and 84.8%, respectively. As a result, it is reasonable to conclude that both algorithms are capable of appropriately classifying the various groups of samples. It can assist milk industries in determining what is wrong during milk processing.

Transfer learning for process monitoring using reflection-mode ultrasonic sensing

The fourth industrial revolution is set to integrate entire manufacturing processes using industrial digital technologies such as the Internet of Things, Cloud Computing, and machine learning to improve process productivity, efficiency, and sustainability. Sensors collect the real-time data required to optimise manufacturing processes and are therefore a key technology in this transformation. Ultrasonic sensors have benefits of being low-cost, in-line, non-invasive, and able to operate in opaque systems. Supervised machine learning models can correlate ultrasonic sensor data to useful information about the manufacturing materials and processes. However, this requires a reference measurement of the process material to label each data point for model training. Labelled data is often difficult to obtain in factory environments, and so a method of training models without this is desirable. This work compares two domain adaptation methods to transfer models across processes, so that no labelled data is required to accurately monitor a target process. The two method compared are a Single Feature transfer learning approach and Transfer Component Analysis using three features. Ultrasonic waveforms are unique to the sensor used, attachment procedure, and contact pressure. Therefore, only a small number of transferable features are investigated. Two industrially relevant processes were used as case studies: mixing and cleaning of fouling in pipes. A reflection-mode ultrasonic sensing technique was used, which monitors the sound wave reflected from the interface between the vessel wall and process material. Overall, the Single Feature method produced the highest prediction accuracies: up to 96.0% and 98.4% to classify the completion of mixing and cleaning, respectively; and R² values of up to 0.947 and 0.999 to predict the time remaining until completion. These results highlight the potential of combining ultrasonic measurements with transfer learning techniques to monitor industrial processes. Although, further work is required to study various effects such as changing sensor location between source and target domains.

Intelligent Industrial Cleaning: A Multi-Sensor Approach Utilising Machine Learning-Based Regression

Effectively cleaning equipment is essential for the safe production of food but requires a significant amount of time and resources such as water, energy, and chemicals. To optimize the cleaning of food production equipment, there is the need for innovative technologies to monitor the removal of fouling from equipment surfaces. In this work, optical and ultrasonic sensors are used to monitor the fouling removal of food materials with different physicochemical properties from a benchtop rig. Tailored signal and image processing procedures are developed to monitor the cleaning process, and a neural network regression model is developed to predict the amount of fouling remaining on the surface. The results show that the three dissimilar food fouling materials investigated were removed from the test section via different cleaning mechanisms, and the neural network models were able to predict the area and volume of fouling present during cleaning with accuracies as high as 98% and 97%, respectively. This work demonstrates that sensors and machine learning methods can be effectively combined to monitor cleaning processes.

Monitoring Mixing Processes Using Ultrasonic Sensors and Machine Learning

Mixing is one of the most common processes across food, chemical, and pharmaceutical manufacturing. Real-time, in-line sensors are required for monitoring, and subsequently optimising, essential processes such as mixing. Ultrasonic sensors are low-cost, real-time, in-line, and applicable to characterise opaque systems. In this study, a non-invasive, reflection-mode ultrasonic measurement technique was used to monitor two model mixing systems. The two systems studied were honey-water blending and flour-water batter mixing. Classification machine learning models were developed to predict if materials were mixed or not mixed. Regression machine learning models were developed to predict the time remaining until mixing completion. Artificial neural networks, support vector machines, long short-term memory neural networks, and convolutional neural networks were tested, along with different methods for engineering features from ultrasonic waveforms in both the time and frequency domain. Comparisons between using a single sensor and performing multisensor data fusion between two sensors were made. Classification accuracies of up to 96.3% for honey-water blending and 92.5% for flour-water batter mixing were achieved, along with R2 values for the regression models of up to 0.977 for honey-water blending and 0.968 for flour-water batter mixing. Each prediction task produced optimal performance with different algorithms and feature engineering methods, vindicating the extensive comparison between different machine learning approaches.

Characterization of the Evolution of Crystallization Fouling in Membranes

Liquid-to-air membrane energy exchangers (LAMEEs) are promising in heating, ventilating, and air-conditioning applications because they are able to use semipermeable membranes to transfer heat and moisture between air and liquid desiccant streams. However, the development of crystallization fouling in membranes may pose a great risk to the long-term performance of LAMEEs. The main aim of this paper is to characterize the evolution of crystallization fouling in membranes through the use of both noninvasive and invasive methods. Noninvasive methods are used to study the development of fouling in the LAMEE by monitoring the changes in moisture flux through the membrane and overall moisture-transfer resistance of the LAMEE. On the other hand, invasive methods are implemented to characterize fouled membranes by using optical microscopy and scanning electron microscopy (SEM) to depict the morphology of crystal deposits and energy-dispersive X-ray spectroscopy (EDX) to identify the composition of the deposits. Experiments are performed by using air to dehydrate MgCl₂(aq) at two operating conditions of low and high fouling rates. The results show that the moisture flux decreases and the moisture-transfer resistance increases more considerably during the test at the high fouling rate than in the test at the low fouling rate. SEM micrographs show that cake crystal deposits cover the membrane surface in the test at the high fouling rate, whereas only few crystal particles are observed on the membrane in the test at the low fouling rate. Furthermore, the crystal deposits undergo more structural changes in the tests at the high fouling rate than in the tests at the low fouling rate, possibly because of the higher moisture transfer rate through the membrane in the tests at the high fouling rate. Finally, the SEM-EDX analysis confirms that the crystal deposits primarily consist of Mg, Cl, and O elements.