Convolutional neural network for ultrasonic weldment flaw classification in noisy conditions

Munir, Nauman; Kim, Hak-Joon; Park, Jinhyun; Song, Sung‐Jin; Kang, Sung-Sik

doi:10.1016/j.ultras.2018.12.001

Cited by 127 publications

(61 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The CNNs consisted of 2 convolutional layers [26,39]. The first was either a 2D or 3D convolutional layer containing 8 5 × 5 pixel filters for each sensor input image depending on whether one or two sensor signals were being used as inputs.…”

Section: Convolutional Neural Networkmentioning

confidence: 99%

“…This approach of using manually engineered features is termed shallow ML. Ultrasonic measurements have been combined with shallow ML algorithms such as Artificial Neural Networks (ANNs) [22][23][24][25][26][27][28][29] and Support Vector Machines (SVMs) [23,25,30,31], using waveform features from the time domain [23,25,27,31,32] and frequency domain [24,27,31,32] after analyses such as wavelet transforms [22,24]. These have been used for applications such as predicting sugar concentration during fermentation [33], measuring particle concentration in multicomponent suspensions [34], and classification of heat exchanger fouling in the dairy industry [23,25].…”

Section: Introductionmentioning

confidence: 99%

“…CNNs use convolutional filters to measure the spatial relationship data values and have found application in image recognition tasks [37,38]. CNNs have also been used to improve ML prediction from ultrasonic signals in both the time [26] and frequency domain after the wavelet transform [39]. The focus of this study is to compare different feature engineering methods and ML algorithms to classify the mixture state and predict the time remaining until mixing completion for two model mixing systems.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Monitoring Mixing Processes Using Ultrasonic Sensors and Machine Learning

Bowler

Bakalis

Watson

2020

Sensors

View full text Add to dashboard Cite

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.

show abstract

Section: Convolutional Neural Networkmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Monitoring Mixing Processes Using Ultrasonic Sensors and Machine Learning

Bowler

Bakalis

Watson

2020

Sensors

View full text Add to dashboard Cite

show abstract

“…random forests [50] or Support Vector Machines SVM [51,52]) with advanced feature engineering are used with the aim to develop computationally lightweight models that can be implemented on-the-fly to aid the inspector in manual inspection. Secondly, data-driven DL approaches using convolutional neural networks (CNNs), U-Nets, etc., [53][54][55][56][57][58][59] are used to learn from raw NDE signals without the need for explicit features engineering. The recent work on deep models takes full advantage of advances in models developed for other applications and shows good results across different NDE fields.…”

Section: Applied Ai For Automated Data Analysis/evaluation Of Nde40 mentioning

confidence: 99%

Next Generation NDE Sensor Systems as IIoT Elements of Industry 4.0

Valeske

Osman

Römer

et al. 2020

Research in Nondestructive Evaluation

View full text Add to dashboard Cite

“…In this proposed NN method, the convolution and pooling are the most important and unique operations. Generally, convolution is an operation on two real-valued functions, such as [13]:…”

Section: Figurementioning

confidence: 99%

A study of supervised combined neural-network-based ultrasonic method for reconstruction of spatial distribution of material properties

Dapkus

Mažeika

Sliesoraitis³

2020

ITC

View full text Add to dashboard Cite

This paper examines the performance of the commonly used neural-network-based classifiers for investigating a structural noise in metals as grain size estimation. The biggest problem which aims to identify the object structure grain size based on metal features or the object structure itself. When the structure data is obtained, a proposed feature extraction method is used to extract the feature of the object. Afterwards, the extracted features are used as the inputs for the classifiers. This research studies is focused to use basic ultrasonic sensors to obtain objects structural grain size which are used in neural network. The performance for used neural-network-based classifier is evaluated based on recognition accuracy for individual object. Also, traditional neural networks, namely convolutions and fully connected dense networks are shown as a result of grain size estimation model. To evaluate robustness property of neural networks, the original samples data is mixed for three types of grain sizes. Experimental results show that combined convolutions and fully connected dense neural networks with classifiers outperform the others single neural networks with original samples with high SN data. The Dense neural network as itself demonstrates the best robustness property when the object samples not differ from trained datasets.

show abstract

Convolutional neural network for ultrasonic weldment flaw classification in noisy conditions

Cited by 127 publications

References 16 publications

Monitoring Mixing Processes Using Ultrasonic Sensors and Machine Learning

Monitoring Mixing Processes Using Ultrasonic Sensors and Machine Learning

Next Generation NDE Sensor Systems as IIoT Elements of Industry 4.0

A study of supervised combined neural-network-based ultrasonic method for reconstruction of spatial distribution of material properties

Contact Info

Product

Resources

About