Supervised Deep Learning and Classification of Single-Event Transients

A radiation-hardened-by-design (RHBD) phase-locked loop (PLL) designed in 0.13-μm CMOS is demonstrated to effectively mitigate the single-event transient (SET) effect. In this paper, a novel Current Limiter circuit Controlled by Comparators (CLCBC) is proposed to improve SET tolerance of the Charge-Pump (CP). Further a SET tolerant Voltage-Controlled Oscillator (VCO) based on an interleaved architecture is also presented. The CLCBC features in being able to detect the abnormal current surged at the CP output node as a result of a SET strike and then to keep this abnormal current from flowing into the low pass filter (LPF) by a high resolution and fast compensation scheme. The VCO consists of a 4-stage interleaved delay buffer implemented to perform the Majority Decision Voting functions without incurring additional signal delay. Such a VCO can generate symmetric multi-phase outputs. Simulation results show that, with deposited charges of 2650fC, the output variation of the CP/ LPF hardened by the proposed CLCBC is reduced by 84.5%, in comparison to the previous work. For the RHBD PLL designed, the perturbation of the VCO control voltage is reduced by 95.6%, while the PLL recovery time is reduced by 68.9%, and the period variation range of the output is reduced by 97%, all with reference to the baseline PLL unhardened. Moreover, the jitter performance of the proposed RHBD PLL remains unaffected and is nearly the same as that of the baseline PLL unhardened.

This paper demonstrates the use of an RF power detector to sense the occurrence of single-event transients (SETs) in RF circuits and systems. The detector was connected to a low-noise amplifier (LNA), and the relationship between the detector output and the LNA transients was investigated via two-photon absorption, pulsed-laser testing, together with mixed-mode TCAD simulations. The response of the detector shows a strong correlation with the RF power of the generated SET. An analytical expression for the detector output is derived and utilized to fit the data. The effects of parameters affecting circuit performance on the detector response to SETs are explored using calibrated TCAD mixed-mode simulations, and design tradeoffs are presented. Finally, a solution to distinguish between transients generated in the detector from those generated in the circuit being monitored is proposed and verified in simulation. The use of detectors to monitor SETs in RF systems could lead to judicious deployment of detection-driven SET mitigation techniques.

Single-event transients (SET) are analyzed in integer- <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">N</i> frequency dividers configured internal and external to a closed-loop phase-locked loop (PLL) circuit. Simulations, corroborated by experimental results, indicate that the location and gain of the frequency divider in any PLL arrangement strongly influence both the error rate of the PLL circuit and the propagation of transients through the closed-loop. The probability of an ion-strike causing output phase displacement values on the order of the operating-period can be significantly reduced by increasing the divisor of the stand-alone output frequency divider. Conversely, increasing the feedback divisor is shown to magnify SETs propagating through the closed-loop PLL.

A probabilistic analysis technique for single event transient sensitivity evaluation of phase-lock-loops

When a high energy particle strikes an integrated circuit, the electron-hole pairs generated in the substrate get collected at source/drain regions. The net effect on the circuit is a transient current (single event transient or SET) injected into circuit nodes. This SET can propagate and cause an error in a state register, which is called a single-event upset (SEU). We perform a detailed characterization of the impact of an SET on a logic circuit. We observe that the impact of an SET can be understood only as a two cycle phenomenon, with several possible SEU outcomes. To understand the relative probabilities of these outcomes, we perform a detailed characterization of the impact of an SET using a Monte Carlo sampling scheme which uses post-layout circuit simulations. These simulations were run in 180nm and 65nm technologies with circuits from ISCAS'85 and ITC'99 benchmarks as test cases. Based on these simulations, we observe that all the SEU outcomes are represented to a substantial extent. Further, there is a substantial fraction of outcomes in which the SET affects multiple state registers. This indicates that the traditional view of impact of an SET on a circuit as a single bit-flip and a single cycle phenomenon may not be fully justified. The impact of an SET is in reality, distributed across two cycles and across multiple state register bits.

First measurements are presented from a newly commissioned two-photon absorption (TPA) setup at Nikhef. The characterisation of the various components of the system is discussed. Two planar silicon sensors, one being electron collecting and one hole collecting, are characterised with detailed measurements of the charge collection and time resolution. The TPA spot is determined to have a radius of 0.975(11) μm and length of 23 μm in silicon. The trigger time resolution of the system is shown to be maximally 30.4 ps. For both sensors, uniform charge collection is observed over the pixels, and the pixel side metallisation is imaged directly using the TPA technique. The best time resolution for a single pixel is found to be 600 ps and 560 ps for the electron and hole collecting sensors respectively, and is dominated by ASIC contributions. Further scans at different depths in the sensor and positions within the pixels have been performed and show a uniform response. It is concluded that the TPA setup is a powerful tool to investigate the charge collection and temporal properties of silicon sensors.

Seeing through disguise: Getting to know you with a deep convolutional neural network

In this paper, the partial discharge (PD) characteristic test is carried out for the four types of typical defective bushings, and the PD characteristics of different typical defective bushings are obtained. The deep convolutional neural network (DCNN) Inception-resnet-light is designed, and the PRPD grayscale image is constructed from the PD test data of different defective bushings as the model input. The method of image recognition is adopted to extract the non-linear characteristics of bushing defects and classify them, then realize the bushing fault diagnosis method based on DCNN. And the model verification is carried out by using examples of defective bushing. The results show that the equivalent frequency of the bushing PD is mainly distributed between 5. 6MHz and 9. 0MHz, and the equivalent time is between 113ns and 1200ns. The PD characteristics of different defective bushings have certain differences, but have a high degree of similarity. The bushing fault diagnosis model based on DCNN which established in this paper can effectively classify bushing defects, and the accuracy of identification on the instance defect bushing is 97.26%, which proves that the proposed method has good generalization performance and feasibility.

Rapid and accurate assessment of endotracheal tube (ETT) location is essential in the intensive care unit (ICU) setting, where timely identification of a mispositioned support device may prevent significant patient morbidity and mortality. This study proposes a series of deep learning-based algorithms which together iteratively identify and localize the position of an ETT relative to the carina on chest radiographs. Using the open-source MIMIC Chest X-Ray (MIMIC-CXR) dataset, a total of 16,000 patients were identified (8000 patients with an ETT and 8000 patients without an ETT). Three different convolutional neural network (CNN) algorithms were created. First, a regression loss function CNN was trained to estimate the coordinate location of the carina, which was then used to crop the original radiograph to the distal trachea and proximal bronchi. Second, a classifier CNN was trained using the cropped inputs to determine the presence or absence of an ETT. Finally, for radiographs containing an ETT, a third regression CNN was trained to both refine the coordinate location of the carina and identify the location of the distal ETT tip. Model accuracy was assessed by comparing the absolute distance of prediction and ground-truth coordinates as well as CNN predictions relative to measurements documented in original radiologic reports. Upon five-fold cross validation, binary classification for the presence or absence of ETT demonstrated an accuracy, sensitivity, specificity, PPV, NPV, and AUC of 97.14%, 97.37%, 96.89%, 97.12%, 97.15%, and 99.58% respectively. CNN predicted coordinate location of the carina, and distal ETT tip was estimated within a median error of 0.46 cm and 0.60 cm from ground-truth annotations respectively. Overall final CNN assessment of distance between the carina and distal ETT tip was predicted within a median error of 0.60 cm from manual ground-truth annotations, and a median error of 0.66 cm from measurements documented in the original radiology reports. A serial cascaded CNN approach demonstrates high accuracy for both identification and localization of ETT tip and carina on chest radiographs. High performance of the proposed multi-step strategy is in part related to iterative refinement of coordinate localization as well as explicit image cropping which focuses algorithm attention to key anatomic regions of interest.

Alpha particles incident on CMOS integrated circuits deposit charges on circuit nodes resulting in single-event transients (SETs). These transient errors propagate through the circuit and reach a latch where they may get latched under proper conditions. This paper presents circuit design techniques to remove the effects of such SET pulses from the circuit. An optimized design is developed whose area, power, and speed performance is superior to other design methods for SET mitigation. Simulation results showing SET pulse elimination are presented.

Detection and classification of transients in data from gravitational wave detectors are crucial for efficient searches for true astrophysical events and identification of noise sources. We present a hybrid method for classification of short duration transients seen in gravitational wave data using both supervised and unsupervised machine learning techniques. To train the classifiers we use the relative wavelet energy and the corresponding entropy obtained by applying one-dimensional wavelet decomposition on the data. The prediction accuracy of the trained classifier on 9 simulated classes of gravitational wave transients and also LIGO's sixth science run hardware injections are reported. Targeted searches for a couple of known classes of non-astrophysical signals in the first observational run of Advanced LIGO data are also presented. The ability to accurately identify transient classes using minimal training samples makes the proposed method a useful tool for LIGO detector characterization as well as searches for short duration gravitational wave signals.

Heart diseases require advanced diagnostic techniques for early and accurate detection. This paper combines multi-modal data sources, such as CT images, MRI scans, and ECG signals to provide a hybrid deep learning architecture for accurate cardiac disease identification. The system uses specific feature extraction methods, such as 3D-UNet for 3D MRI and CT images and Temporal Convolutional Graph Neural Networks (TC-GNN) for ECG, and then uses genetic algorithms to optimize the features. Autoencoders, which are 1D for ECG and 3D for MRI and CT, are employed for non-linear dimensionality reduction in order to handle the high dimensionality of fused information. A Convolutional Neural Network (CNN) processes the fused compact features for the final classification. The proposed model achieved a 97.1% accuracy, outperforming known models. Accuracy, recall, F1-score, and ROC-AUC scores support its generalizability and robustness. This multi-modal and feature-aware approach significantly increases classification accuracy, reduces false positives and false negatives, and provides a scalable clinical decision support solution for cardiovascular diagnostics.

Recently, automatic vehicle identification system has great significance in practical life and with modern technology now widely available, the research on automatic vehicle recognition based on computer vision is becoming increasingly thorough and extensive. However, automatic vehicle identification also faces significant challenges, such as changes in lighting, weather conditions and distortion of captured images in natural road scenes. Since 2012, as a deep learning in image classification, the convolution neural network has become a major subject in many fields, so this study would focus on the vehicle identification in terms of the convolutional neural network. The Cars196 data set and an improved Convolutional Neural Network (CNN) network capable of multi-attribute recognition are applied to study the optimization network problem of the convolutional neutral network and to realize the multi-attribute classification and identification of automobile brands, models and colors. The data analysis shows that the improved CNN recognition network has a recognition accuracy in automatic vehicle recognition, but the overall recognition network is affected by many factors. According to the results of data analysis, in practice, alongside the improvement of traditional convolutional neural network, the improved new neural network structure should also be used for vehicle recognition, so as to improve the accuracy of vehicle identification to better serve the community.

Accurate identification and classification of atmospheric particulates can provide the basis for their source apportionment. Most current research studies mainly focus on the classification of atmospheric particles based on the energy spectrum of particles, which has the problems of low accuracy and being time-consuming. It is necessary to study the classification method of atmospheric particles with higher accuracy. In this paper, a convolutional neural network (CNN) model with attention mechanism is proposed to identify and classify the scanning electron microscopy (SEM) images of atmospheric particles. First, this work established a database, Qingdao 2016–2018, for atmospheric particles classification research. This database consists of 3469 SEM images of single particulates. Secondly, by analyzing the morphological characteristics of single particle SEM images, it can be divided into four categories: fibrous particles, flocculent particles, spherical particles, and mineral particles. Thirdly, by introducing attention mechanism into convolutional neural network, an Attention-CNN model for the identification and classification of the four types of atmospheric particles based on the SEM images is established. Finally, the Attention-CNN model is trained and tested based on the SEM images database, and the results of identification and classification for four types of particles are obtained. Under the same SEM images database, the classification results from Attention-CNN are compared with those of CNN and SVM. It is found that Attention-CNN has higher classification accuracy and reduces significantly the misclassification number of particles, which shows the focusing effect of attention mechanism.

Due to the high cost and difficulty of traffic data set acquisition and the high time sensitivity of traffic distribution, the machine learning-based traffic identification method is difficult to be applied in airborne network environment. Aiming at this problem, a method for airborne network traffic identification based on the convolutional neural network under small traffic samples is proposed. Firstly, the pre-training of the initial model for the convolutional neural network is implemented based on the complete data set in source domain, and then the retraining of the convolutional neural network is realized through the layer frozen based fine-tuning learning algorithm of convolutional neural network on the incomplete dataset in target domain, and the convolutional neural network model based feature representing transferring(FRT-CNN) is constructed to realize online traffic identification. The experiment results on the actual airborne network traffic dataset show that the proposed method can guarantee the accuracy of traffic identification under limited traffic samples, and the classification performance is significantly improved comparing with the existing small-sample learning methods.