Search

• Broad learning system (BLS) was introduced to pattern recognition of hand movements and investigated for different feature sets and numbers of training samples. • MMG of one-time collection and sensor doffing and donning is studied with classification algorithms, feature sets, and numbers of training data. • In the condition of one-time collecting MMG within a day, the classification accuracy of hand movements is higher than 99% by using BLS. • The negative effect on classification accuracy caused by sensor doffing and donning can be alleviated by increasing the subsets of training data. Pattern classification of hand movements based on mechanomyography (MMG) has specific application value in the development of human–machine interaction and wearable devices. In the condition of one-time collection, high classification accuracy can be acquired. However, sensor doffing and donning unavoidably change the site and contact pressure of sensors, having a negative effect on classification accuracy. In the condition of sensor doffing and donning, eight-channel MMG of the forearms from participants when they were performing four classes of hand movements were collected for 12 days, and pattern recognition of hand movements were investigated for one-time collection and sensor doffing and donning. After feature extraction and a combination of feature subsets, a broad learning system (BLS) was introduced to pattern recognition of hand movements based on MMG, which was further compared with three other algorithms. In the condition of the one-time collection, recognition rates of each class are higher than 99 %, which is better than that by using a support vector machine (SVM), which demonstrates the excellent learning ability of BLS, whereas the SVM shows a better performance than the BLS in the condition of sensor doffing and donning. When using SVM, BLS, or extreme learning machine, the classification accuracy gradually increases with the number of data subsets, which illustrates that the negative effect on classification accuracy caused by sensor doffing and donning can be alleviated by increasing the amount of training data. [ABSTRACT FROM AUTHOR]

• A new experiment that involves upper limb motions and muscle forces is designed. • The classification of upper limb motions and muscle forces based on mechanomyography is proposed. • This study provide a basis for further research in which rehabilitation training is evaluated based on MMG. • This study has potential advantages in development on the human-computer interaction system that assist rehabilitation training. Rehabilitation training is essential for patients who have a history of certain illnesses, such as stroke. As a crucial part of rehabilitation training, upper limb training involves such key factors as upper limb motions and forces. This study investigated three upper limb motions (elbow flexion of 135°, Motion 1; shoulder flexion of 90°, Motion 2; and shoulder abduction of 90°, Motion 3) and various forces (muscle Force 0, no force; holding one 1.4 kg dumbbell, muscle Force 1; holding one 2.4 kg dumbbell, muscle Force 2) in combination to evaluate nine motion patterns. These patterns were completed by twelve healthy volunteers. Mechanomyography (MMG) measurements of the biceps brachii (Channel 1), triceps (Channel 2), and deltoid (Channel 3) muscles were collected. These were subsequently divided into signal segments corresponding to each of the motions using a segmentation method based on average energy. After extracting time-domain features and wavelet packet energy features, support vector machine analysis (SVM) was used for the classification of the upper limb motions and forces based on the MMG measurements. Channel 2 and Channel 3 were shown to play an important role in the classification of upper limb motions, and Channel 1 played a role in the classification of the forces. These results demonstrate that collection of MMG measurements from the three muscles is feasible and suggest a foundation for further studies in which rehabilitation training is evaluated based on MMG measurements. [ABSTRACT FROM AUTHOR]

Parameters of blood flow measured by ultrasound in radial and ulnar arteries, such as flow velocity, flow rate and wall shear rate, are widely used in clinical practice and clinical research. Investigation of these measurements is useful for evaluating accuracy and providing knowledge of error sources. A method for simulating the spectral Doppler ultrasound measurement process was developed with computational fluid dynamics providing flow-field data. Specific scanning factors were adjusted to investigate their influence on estimation of the maximum velocity waveform, and flow rate and wall shear rate were derived using the Womersley equation. The overestimation in maximum velocity increases greatly (peak systolic from about 10% to 30%, time-averaged from about 30% to 50%) when the beam-vessel angle is changed from 30° to 70°. The Womersley equation was able to estimate flow rate in both arteries with less than 3% error, but performed better in the radial artery (2.3% overestimation) than the ulnar artery (15.4% underestimation) in estimating wall shear rate. It is concluded that measurements of flow parameters in the radial and ulnar arteries with clinical ultrasound scanners are prone to clinically significant errors. [ABSTRACT FROM AUTHOR]

• A Vision Transformer-based architecture for lower limb motion recognition from MMG and kinematic signals was proposed. • The model can recognize 8 motions with an accuracy of 94.62%, much higher than machine learning methods and CNN. • Fusion signals of MMG and kinematic data improve the classification accuracies of the model. • The proposed ViT-LLMR maintains high recognition accuracy when undersampling and only use part of signals. One of the key problems in lower limb-based human–computer interaction (HCI) technology is to use wearable devices to recognize the wearer's lower limb motions. The information commonly used to discriminate human motion mainly includes biological and kinematic signals. Considering that unimodal signals do not provide enough information to recognize lower limb movements, in this paper, we proposed a Vision Transformer (ViT)-based architecture for lower limb motion recognition from multichannel Mechanomyography (MMG) signals and kinematic data. Firstly, we applied the self-attention mechanism to enhance each input channel signal. Then the data was fed into ViT model. Vision Transformer-based Lower Limb Motion Recognition (ViT - LLMR) architecture proposed in this paper can avoid the model training problems such as autonomous feature extraction and feature selection for machine learning, and the model can recognize eight lower limb motions containing six subjects with an accuracy of 94.62%. In addition, we analyzed the generalization ability of the model when undersampling and only collecting fragment signals. In conclusion, the proposed ViT - LLMR architecture could provide a basis for practical applications in different HCI fields. [ABSTRACT FROM AUTHOR]

• Repetitive wear and multiple collection will cause non-identical distribution. • The traditional model has poor recognition effect on non-identical signals. • Incremental learning has a good recognition effect on non-identical signals. • Deep learning can construct sample incremental learning well. • Incremental learning can be used to build real-time evaluation training systems. Aiming at the problem that the traditional mechanomyography (MMG) pattern recognition model for upper limb rehabilitation action has poor recognition effect on non-identical distribution test data, this research proposes an incremental learning method for deep learning based on MMG. By collecting 8-channels of MMG of 4 types of hand movements (wrist flexion, wrist extension, wrist ulnar flexion, and wrist radial flexion), after signal preprocessing, feature extraction and dimensionality reduction, 12 groups of non-identical distributed data were obtained. The model was trained by using deep neural network (DNN), and five commonly used machine learning algorithms were used as comparison for incremental training. Finally, the recognition rate of DNN was 88.25%, and the final recognition rates of Passive Aggressive (PA), Incremental Support Vector Machine (ISVM), Perceptron, Bernoulli Naive Bayes (BNB) and Multinomial Naive Bayes (MNB) were 85.94%, 84.82%, 81.01%, 73.07% and 61.20% respectively. Among them, both DNN and PA had better upward trends, and DNN had the highest final recognition rate. By comparing the confusion matrices before and after DNN incremental learning, it could be seen that DNN incremental training can significantly improve the accuracy and precision in confusion matrix. The experimental results demonstrate that the method of incremental learning can not only achieve the recognition of non-identically distributed test data, but also improve the recognition rates and the generalization performance of the model. [ABSTRACT FROM AUTHOR]

The aim of this study was to evaluate the errors in measurement of volumetric flow rate and wall shear rate measured in radial and ulnar arteries using a commercial ultrasound scanning system. The Womersley equations were used to estimate the flow rate and wall shear rate waveforms, based on the measured vessel diameter and centerline velocity waveform. In the experiments, each variable (vessel depth, diameter, flow rate, beam-vessel angle and different waveform) in the phantom was investigated in turn, and its value was varied within a normal range while others were fixed at their typical values. The outcomes revealed that flow rate and wall shear rate were overestimated in all cases, from around 13% to nearly 50%. It is concluded that measurements of flow rate and wall shear rate in radial and ulnar arteries with a clinical ultrasound scanner are vulnerable to overestimation. [ABSTRACT FROM AUTHOR]

• A novel wireless multi-channel MMG acquisition system is designed. • Four swarm intelligence algorithms (GA, PSO, WOA, SSA) and three methods (ReliefF, PCA, SFS) for feature selection. • By adding the weight factor to improve the three algorithms (IPSO, IWOA, ISSA), better classification results were achieved. • The classification accuracy of eight types of lower limb movements is over 90%. Mechanomyography (MMG) is a kind of biomedical signal with great research value. This paper designed a novel wireless MMG signal acquisition system composed of modular nodes and core board. The modular nodes were worn on different muscle of the lower limbs, collecting movement data of the corresponding part, and transmitted it to the core board wirelessly. The core board was connected to the computer through the USB to achieve the wireless collection and real-time display of MMG signals. In order to improve the real-time performance, this paper adopted four swarm intelligence algorithms (GA, PSO, WOA, SSA) and three improved algorithms (IPSO, IWOA, ISSA) for feature selection to reduce the redundancy of information. In this study, the different effects of different feature selection methods on the recognition of eight types of lower limb movements based on MMG signals were discussed by comparing the classification accuracy, number of iterations and calculation time. The results show that swarm intelligence algorithms have merits in this type of feature selection problem, and GA has the most obvious effect in improving classification accuracy, but it requires more iterations and calculation time, while the classification accuracy of IPSO is close to that of GA, and it has advantages in time consumption. [ABSTRACT FROM AUTHOR]

Abstract: Disturbances that propagate throughout a plant can have an impact on product quality and running costs. There is thus a motivation for the automated detection of plant-wide disturbances and for the isolation of the sources. A new application of independent component analysis (ICA), multi-resolution spectral ICA, is proposed to detect and isolate the sources of multiple oscillations in a chemical process. Its key feature is that it extracts dominant spectrum-like independent components each of which has a narrow-band peak that captures the behaviour of one of the oscillation sources. Additionally, a significance index is presented that links the sources to specific plant measurements in order to facilitate the isolation of the sources of the oscillations. A case study is presented that demonstrates the ability of spectral ICA to detect and isolate multiple dominant oscillations in different frequency ranges in a large data set from an industrial chemical process. [Copyright &y& Elsevier]

Abstract: Constrained, spectral, independent component analysis of perturbed controller output data is proposed to isolate multiple sources of plant-wide oscillations. The technique is described and applied to data pertaining to a simulated case study and to real data obtained from an industrial chemical plant. Results demonstrate its ability to isolate the sources of multiple oscillations at the loop level. [Copyright &y& Elsevier]

Loop status monitoring involves the declaration of deterministic trends, such as oscillations and drifting, in loops that are in multi-loop plant configurations. By analysing various time domain statistics pertaining to PI/PID control loops, a trend can be recognised as one of seven categories. The scientific basis for working with the particular statistics is given and the categorisation process is described. These statistics can be combined to produce an Overall Loop Performance Index for each loop, which can be compared to localise a single fault in a multi-loop arrangement. Estimation methods for these time domain statistics are outlined and the performance of the approach is demonstrated on both simulated and real plant data sets. [Copyright &y& Elsevier]

• In MMG-based hand movement recognition, novel swarm intelligence algorithms (BES, GWO, SSA) were introduced to optimize SVM. • Based on hand-crafted feature extraction methods and a deep learning method, the performances of classification methods are compared in accuracy and time consumption. • Adopting GWO-SVM with TD + FD features can obtain good classification accuracy of eight classes of hand movements, up to 93.55 %, and take relatively less training time. On the basis of extracting mechanomyography (MMG) signal features, the classification of hand movements has certain application values in human-machine interaction systems and wearable devices. In this paper, pattern recognition of hand movements based on MMG signal is studied with swarm intelligence algorithms introduced to optimize support vector machine (SVM). Time domain (TD) features, wavelet packet node energy (WPNE) features, frequency domain (FD) features, convolution neural network (CNN) features were extracted from each channel to constitute different feature sets. Three novel swarm intelligence algorithms (i.e., bald eagle search (BES), sparrow search algorithm (SSA), grey wolf optimization (GWO)) optimized SVM is proposed to train the models and recognition of hand movements are tested for each MMG feature extraction method. Using GWO as the optimization algorithm, time consumption is less than using the other two swarm algorithms. Using GWO with TD+FD features can obtain the classification accuracy of 93.55 %, which is higher than other methods while using CNN to extract features can be independent of domain knowledge. The results confirm GWO-SVM with TD + FD features is superior to some other methods in the classification problem for tiny samples based on MMG. [ABSTRACT FROM AUTHOR]

Abstract: In this study, a measurement method of parametric temperature model for industrial bioreactor is proposed. The system of industrial bioreactor is an integrating process with time delay. Obviously, it cannot hold a stable system response since the characteristic roots are located on the imaginary axis of the complex plane. In this paper, the temperature model of the bioreactor was obtained by maintaining the bioreactor continuous response to a monopulse signal. This proposed method has a powerful ability to steady system response. The bioreactor consists of 30L fermentor, two impellers and four baffles. By using the proposed method and Bierman algorithm, the parameters of the temperature model for bioreactor are successfully measured on-line. Simulation results are given to show the effectiveness of the measurement method. [Copyright &y& Elsevier]