Output Samples Research Articles

Pyrough: A tool to build 3D samples with rough surfaces for atomistic and finite-element simulations

Natural samples are characterized by surface roughness which is intrinsically multi-scale as depicted by the well known concept of fractal dimension. Nevertheless, surface asperities are barely taken into account in simulations and modeling where flat surfaces and sharp corners or edges are generally preferred for the sake of simplicity. In this context, we propose here a versatile Python program called Pyrough that aims at building virtual samples characterized by configurable surface roughness for numerical applications such as atomistic and finite-element simulations. The program is open source and relies on the classical roughness theory that integrates the concept of self-affine surface. Several basic shapes including basic blocks, spheres, grains and wires with self-affine surface asperities are implemented and the object-oriented structure of the program simplifies the implementation of more complex objects. Virtual sample design is improved using Pyrough, which enables more realistic simulations to be made. Several application examples including e.g., the design of wavy grain boundaries or nanoindentation testing using a roughened indenter tip are presented. Program summaryProgram Title: PyroughCPC Library link to program files:https://doi.org/10.17632/x7jdtrbf4s.1Developer's repository link:https://github.com/jamodeo12/PyroughLicensing provisions: GNU General Public License 3Programming language: PythonExternal routines/libraries: Gmsh, Meshio, Wulffpack, ASE, Atomsk, cv2Nature of problem: 3D virtual samples used for atomistic or finite-element simulations generally rely on simplified geometries and surfaces for the sake of design simplicity. However, the influence of surface roughness play a crucial role in various fields of applications (e.g., mechanics, catalysis, lubrication) and must be taken into account.Solution method: Pyrough allows for the design of 3D virtual objects with rough surfaces by means of the classical roughness theory. The user can easily tune the morphology of surfaces and shape 3D objects. Output samples can be used in finite-element or atomistic simulations according to the user's needs.Additional comments including restrictions and unusual features: The program documentation is available at https://jamodeo12.github.io/Pyrough/

The method of GPS monitoring of small-scale inhomogeneities of the ionosphere and its application for predicting the noise immunity of satellite communication systems

A method has been developed for GPS monitoring of small-scale fluctuations in the total electronic content of the ionosphere and using its results to predict changes in the noise immunity of satellite communication systems under ionospheric disturbances. To this end, the software of the GPStation-6 receiver of the global navigation satellite system was modified in the direction of replacing the code measurements of the total electronic content of the ionosphere with combined (code-phase) measurements, which practically eliminates the "noise" component of the measurement error. To separate small-scale fluctuations of the total electronic content from large-scale ones, it is proposed to use a 6th-order digital Butterworth filter, which in the transmission frequency range (1-10 Hz) at a sampling frequency of 50 Hz provides an amplitude-frequency response close to ideal and introduces a delay not exceeding the permissible value. An expression is obtained for calculating the output samples of this digital filter. The obtained results of estimating small-scale fluctuations of the total electron content of the ionosphere at the output of the digital filter allow us to estimate the standard deviation of these fluctuations. On this basis, the results of calculating the ionospheric flicker index of the received signals of satellite communication systems are obtained. This makes it possible to predict the probability of their erroneous reception at a given average value of the signal-to-noise ratio at the input of the satellite receiver. Analysis of the obtained results of predicting the noise immunity of satellite communication systems shows that in the conditions of an undisturbed mid-latitude ionosphere, the probability of erroneous reception may exceed the permissible value for 30 seconds. This confirms the assumption that with small-scale disturbances of the ionosphere, the probability of error in satellite communication systems may exceed the permissible value for a long time.

Adaptive DLMS equalizer based on modern FPGA

A simple and effective way to implement adaptive equalization schemes in digital communication systems is to implement them on general-purpose microcontrollers. However, increasing the order of the equalizer, which leads to increase the time to obtain the output sample, at the same time leads to decreasing maximum allowable symbol rate. It makes the implementation based on microcontrollers impractical for use in modern high-speed communication protocols. A more suitable computing platform for implementing LMS algorithms is FPGA, where there is no dependence of output sample obtaining time on equalizer order. In order to derive the error signal at the equalizer output from the last (actual) output sample, the estimation of the error signal in the FPGA is required to occur during the same clock period as the calculation of the output signal sample. Trying to fit all the necessary operations into one clock period leads to decrease maximum allowable clock frequency and inefficient using of computing resources within the FPGA. The solving of this problem is to use the Delayed LMS (DLMS) algorithm, invented in the late 80s. The essence of DLMS algorithm is to pipeline both the blocks for calculating the equalizer coefficients and the output samples. The meaning of pipelining is to install synchronous delay elements (D-flip-flops) between the elements of combinatorial logic, which makes it possible to increase the clock and symbol frequency of the equalized signal. The disadvantage of this method is the presence of the synchronous delay of several clock cycles and, as consequence, using of irrelevant error signal for updating coefficients. Retiming procedure is used to optimally distribute synchronous delays to maximize system clock frequency. This article describes the DLMS equalizer and shows the effectiveness of DLMS algorithms implementation on FPGA. The effectiveness of algorithms, which have been used for the retiming procedure is shown. Diagrams of BER(SNR) and BER (Doppler frequency shift) of LMS and DLMS equalization circuits are presented. Obtained results facilitate transfer of demodulator circuit of existing modem from the Zynq microcontroller part built into the Xilinx XC7Z020 FPGA to its programmable logic part in order to increase the performance of the modem receiver to solve current and future challenges.

Embedded Device Energy Consumption Prediction System based on Machine Learning Optimization

Accurately predicting the energy consumption of embedded devices has important implications for improving system performance and stability, resolving energy consumption anomalies and faults, and developing energy management strategies. In addition, it can promote the rational use of energy, reduce carbon emissions and environmental pollution, and achieve sustainable development. This article establishes an energy consumption prediction model for embedded devices using machine learning algorithms, and adopts a physical verification system to verify the accuracy of the prediction model for embedded devices. Firstly, divide the input and output samples obtained from CFD(Computational Fluid Dynamics, CFD) into training and testing sets, and use the SVM(Support Vector Machines, SVM) algorithm to train the training set to obtain an energy consumption prediction model. Then, the testing set is simulated and validated using the model. Finally, a related embedded hardware circuit system is designed to collect temperature, humidity, air conditioner voltage, and current data using sensor modules, and the energy consumption prediction model is validated using physical verification. The results showed that the relative error percentage between the energy consumption prediction model obtained through SVM and the simulation results was small. The percentage error between the actual energy consumption value and the predicted energy consumption value was also relatively small, verifying the feasibility of the research in this article.

A Method for Extrapolating Continuous Functions by Generating New Training Samples for Feedforward Artificial Neural Networks

The goal of the present study is to find a method for improving the predictive capabilities of feedforward neural networks in cases where values distant from the input–output sample interval are predicted. This paper proposes an iterative prediction algorithm based on two assumptions. One is that predictions near the statistical sample have much lower error than those distant from the sample. The second is that a neural network can generate additional training samples and use them to train itself in order to get closer to a distant prediction point. This paper presents the results of multiple experiments with different univariate and multivariate functions and compares the predictions made by neural networks before and after their training with the proposed iterative algorithm. The results show that, having passed through the stages of the algorithm, artificial neural networks significantly improve their interpolation performance in long-term forecasting. The present study demonstrates that neural networks are capable of creating additional samples for their own training, thus increasing their approximating efficiency.

Tracking consensus for uncertain nonlinear multi‐agent systems under cloud‐based event‐triggered coordination

AbstractThis paper considers the cloud‐based event‐triggered coordination problem of uncertain high‐order nonlinear multi‐agent systems. The communication among agents is realized by indirect and asynchronous access to a shared cloud database. A self‐triggered extended state observer (ESO) is first designed to estimate the total uncertainty. Based on the output of the ESO, an event‐triggered controller is proposed. It is proved that under the proposed output feedback controller, the multi‐agent system can track a virtual leader. We also show that Zeno behavior from output sampling or controller update will never happen. Finally, a numerical example and a vehicle platoon example are provided to corroborate the obtained theoretical results.

Output Power Computation and Adaptation Strategy of an Electrosurgery Inverter for Reduced Collateral Tissue Damage.

This paper investigates two ways of output-power computation, namely, sparse- and multi-sampling-based methods, to overcome sampling speed limitation and arcing nonlinearity for electrosurgery. Moreover, an impedance-based power adaptation strategy is explored for reduced collateral tissue damage. The efficacy of the proposed power computation and adaptation strategy are experimentally investigated on a gallium-nitride (GaN)-based high-frequency inverter prototype that allows electrosurgery with a 390 kHz output frequency. The sparse-sampling-based method samples output voltage once and current twice per cycle. The achieved power computing errors over 1000 cycles are 1.43 W, 2.54 W, 4.53 W, and 4.89 W when output power varies between 15 W and 45 W. The multi-sampling-based method requires 28 samples of both outputs, and the corresponding errors are 0.02 W, 0.86 W, 1.86 W, and 3.09 W. The collateral tissue damage gauged by average thermal spread is 0.86 mm, 0.43 mm, 1.11 mm, and 0.36 mm for the impedance-based power adaptation against 1.49 mm for conventional electrosurgery. Both power-computation approaches break sampling speed limitations and calculate output power with small errors. However, with arcing nonlinearity presence, the multi-sampling-based method yields better accuracy. The impedance-based power adaptation reduces thermal spreads and diminishes sensor count and cost. This paper exemplifies two novel power-computation ways using low-end industrial-scale processors for biomedical research involving high-frequency and nonlinearly distorted outputs. Additionally, this work is the first to present the original impedance-based power adaptation strategy for reduced collateral damage and it may motivate further interdisciplinary research towards collateral-damage-less electrosurgery.

Research on building material design in circular economy based on improved neural network

Material management technology has always been the basic and core technology of traditional construction projects. But now, because traditional material management consumes more resources and has low efficiency, it has gradually been replaced by modern material management systems represented by computer technology, thus saving a lot of enterprise input costs and reducing the impact of traditional material management technology on enterprise operating income. Based on this background, this paper introduces improved neural network technology to adapt to the circular economy environment, so as to complete the construction and improvement of the building materials data management system. The system can be used to complete building materials management, basic data management, etc., and can be optimized for the system itself. Through network technology, the client can be connected to the server switch, and Ajax technology and JSP programming language can be used for compatibility processing, so that the system can run better in the Windows environment. According to the simulation experiment designed for the system, the output sample value of the system is close to the measured value, with good prediction accuracy, small error and high stability. Therefore, the application of AdaBoost ANN algorithm can effectively realize the prediction of building materials. In addition, other functions of the system, such as interactive page design, operation interface and database, can better adapt to the circular economy environment and actual user needs. In this paper, the improved neural network is applied to the field of building material design under the environment of circular economy, thus completing the system optimization.

\(\boldsymbol{\mathcal{L}_2}\)-Optimal Reduced-Order Modeling Using Parameter-Separable Forms

We provide a unifying framework for -optimal reduced-order modeling for linear time-invariant dynamical systems and stationary parametric problems. Using parameter-separable forms of the reduced-model quantities, we derive the gradients of the cost function with respect to the reduced matrices, which then allows a nonintrusive, data-driven, gradient-based descent algorithm to construct the optimal approximant using only output samples. By choosing an appropriate measure, the framework covers both continuous (Lebesgue) and discrete cost functions. We show the efficacy of the proposed algorithm via various numerical examples. Furthermore, we analyze under what conditions the data-driven approximant can be obtained via projection.

The Discrete Load Frequency Control System Using a Robust Periodic Output Feedback Controller

The load frequency control (LFC) is a most important tool for the frequency regulation mechanism in the widely spread modern power system. The LFC system consists of a communication structure to transmit the measurement and control signal. Usually the controllers in LFC systems are designed and implemented in continuous mode of operation. This article investigates the discrete mode load frequency control (LFC) mechanism, by employing the concept of the periodic output feedback (POF)-based controller with varying input and output sampling frequencies. Both the optimal sampling frequency and the optimal POF controller gain matrix are found by using the particle swarm optimization (PSO) method. The POF-based controller is intended for usage in two-area multisource LFC systems, with varying input and output sampling frequencies. The performance analysis takes into account a variety of scenarios, including those without a conventional stabilizer, with conventional continuous and corresponding discrete mode PSS, and a proposed discrete mode POF controller. Furthermore, the efficacy of the discrete mode POF controller is evaluated on the MATLAB/Simulink platform.

Homogeneity Enforced Calibration of Stage Nonidealities for Pipelined ADCs

Pipelined analog-to-digital converters (ADCs) are fundamental components of various signal processing systems requiring high sampling rates and a high linearity. Over the past years, calibration techniques have been intensively investigated to increase the linearity. In this brief, we propose an equalization-based calibration technique which does not require knowledge of the ADC input signal for calibration. For that, a test signal and a scaled version of it are fed into the ADC sequentially, while only the corresponding output samples are used for calibration. Several test signal sources are possible, such as a signal generator (SG) or the system application (SA) itself. For the latter case, the presented method corresponds to a background calibration technique. Thus, slowly changing errors are tracked and calibrated continuously. Because of the low computational complexity of the calibration technique, it is suitable for an on-chip implementation. Ultimately, this brief contains an analysis of the stability and convergence behavior as well as simulation results.

A Recursive System Identification With Non-Uniform Temporal Feedback Under Coprime Collaborative Sensing

Abstract We present a system identification method based on recursive least-squares (RLS) and coprime collaborative sensing, which can recover system dynamics from non-uniform temporal data. Focusing on systems with fast input sampling and slow output sampling, we use a polynomial transformation to reparameterize the system model and create an auxiliary model that can be identified from the non-uniform data. We show the identifiability of the auxiliary model using a Diophantine equation approach. Numerical examples demonstrate successful system reconstruction and the ability to capture fast system response with limited temporal feedback.

Time-delay metric for TDE of SISO delay systems

The time-delay intrinsic in the time series sampled from dynamic systems results in the chaotic correspondence between the elements of the time series. Therefore, accurate estimation of the time-delay is essential in time series-based tasks. In this paper, we propose two time-delay metrics for time-delay estimation (TDE) of single-input single-output (SISO) delay systems with unknown structures. The two proposed metrics are based on the ellipse shape ratio (ESR) and called the linear ellipse shape ratio (LESR), and the nonlinear ellipse shape ratio (NESR). In the devised TDE process employing the proposed metrics, an iteratively updated ellipse is used to filter out the outliers from the system input and output samples. Simulation experiments confirm the efficiency of the proposed time-delay metrics.

Update on the CIC Multistage Decimation Filter With a Minimum Number of Additions per Output Sample (APOS): Can We Still Decrease the Number of APOS? [Tips & Tricks

This article explores some tips and tricks on how to decrease the number of additions per output sample (APOS) in a cascaded-integrator-comb (CIC) multistage decimation filter, ensuring the smallest number of APOS for a given decimation factor; the desired minimum worst-case attenuation (WCA) in dB; and the desired passband edge frequency <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">${\omega}_{p}$</tex-math></inline-formula> . We assume that the decimation factor <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">M</i> is a product of prime factors. In the last stage, we propose modifying the optimal multistage structure with the minimum number of APOS presented in the literature.

Data-Driven Certification of Neural Networks With Random Input Noise

Methods to certify the robustness of neural networks in the presence of input uncertainty are vital in safety-critical settings. Most certification methods in literature are designed for adversarial or worst-case inputs, but researchers have recently shown a need for methods that consider random input noise. In this article, we examine the setting where inputs are subject to random noise coming from an arbitrary probability distribution. We propose a robustness certification method that lower-bounds the probability that network outputs are safe. This bound is cast as a chance-constrained optimization problem, which is then reformulated using input–output samples to make the optimization constraints tractable. We develop sufficient conditions for the resulting optimization to be convex, as well as on the number of samples needed to make the robustness bound hold with overwhelming probability. Case studies on synthetic, MNIST, and CIFAR-10 networks experimentally demonstrate that this method is able to certify robustness against various input noise regimes over larger uncertainty regions than prior state-of-the-art techniques.

An intrusion detection method based on stacked sparse autoencoder and improved gaussian mixture model

The analysis of a substantial portion of network data is a requirement for almost any machine learning-based network intrusion detection method. High dimension features, a lack of labelled datasets, and inflexible feature selection are some of the difficulties it encounters. This paper proposes an intrusion detection method based on stacked sparse autoencoder and improved Gaussian mixture model (SIGMOD). The method utilizes the stacked sparse autoencoder to produce non-linear dimensionality reduction after first using the Pearson correlation coefficient to achieve linear dimensionality reduction. It can lessen redundant data, obtain low-dimensional features and generate reconstruction errors of the samples, which are subsequently fed into the improved Gaussian mixture model. The method finally judges whether it is abnormal according to the output sample energy threshold of the improved Gaussian mixture model. Unlike the traditional anomaly detection technology that performs dimensionality reduction and clustering in two separate steps, SIGMOD jointly optimizes the parameters of the stacked sparse autoencoder and the Gaussian mixture model in an end-to-end manner, avoiding the key feature loss of cluster analysis or probability density estimation in dimensionality reduction by the two-step method. The experiment results on the UNSW-NB15 dataset demonstrate that SIGMOD performs significantly better than the conventional anomaly detection approach.

Ultra-Low-Power Voice Activity Detection System Using Level-Crossing Sampling

This paper presents an ultra-low-power voice activity detection (VAD) system to discriminate speech from non-speech parts of audio signals. The proposed VAD system uses level-crossing sampling for voice activity detection. The useless samples in the non-speech parts of the signal are eliminated due to the activity-dependent nature of this sampling scheme. A 40 ms moving window with a 30 ms overlap is exploited as a feature extraction block, within which the output samples of the level-crossing analog-to-digital converter (LC-ADC) are counted as the feature. The only variable used to distinguish speech and non-speech segments in the audio input signal is the number of LC-ADC output samples within a time window. The proposed system achieves an average of 91.02% speech hit rate and 82.64% non-speech hit rate over 12 noise types at −5, 0, 5, and 10 dB signal-to-noise ratios (SNR) over the TIMIT database. The proposed system including LC-ADC, feature extraction, and classification circuits was designed in 0.18 µm CMOS technology. Post-layout simulation results show a power consumption of 394.6 nW with a silicon area of 0.044 mm2, which makes it suitable as an always-on device in an automatic speech recognition system.

Kernel-Based Identification of Incrementally Input-to-State Stable Nonlinear Systems

Methods based on Reproducing Kernel Hilbert Spaces (RKHS) have proven to be a valuable tool for the identification of linear time-invariant systems in both discrete- and continuous-time. In particular, unlike most other techniques, they enable to systematically confer a priori desirable properties, such as stability, on the estimated models. However, existing RKHS methods mainly target impulse responses and, hence, do not extend well to the context of nonlinear systems. In this work, we propose a novel RKHS-based methodology for the identification of discrete-time nonlinear systems guaranteeing that the identified system is incrementally input-to-state stable (δISS). We model the identified system using a predictor function that, given past input and output samples, yields the output prediction at the next time instant. The predictor is selected from an RKHS by solving a constrained optimization problem that guarantees its δISS properties. The proposed approach is validated via numerical simulations.

Deep MMSE Estimation for Data Detection

This letter develops minimum mean-squared error (MMSE) estimators based on deep neural networks for data detection. Since the optimal MMSE is analytically intractable, researchers usually resort to linear MMSE approximations, which often incur performance degradation. To overcome this loss, we develop a near-optimal estimator by exploiting the Donsker-Varadhan representation of mutual information (MI) and the recently discovered derivative relationship between MI and MMSE. The near-optimal MMSE estimator can be computed with a deep neural network (NN). We can train the NN using the mini-batches of input and output samples. We validate this estimator using two examples where the closed-from MMSE is available. We then use this estimator to design an end-to-end communication system. We compare this setup with several conventional techniques and show promising performance.

Generative Adversarial Neural Approach Towards Construction of Warli-stick Figure Aboriginal Artwork

Image Synthesis or generation is evolving at a high rate helping from diagnosis of diseases to traffic control. Deep Learning has touched a new height in which domains like Computer Vision have successfully developed Deep learning models for image generation. Recent improvements in deep learning enabled Generative Adversarial Networks or GANs to achieve a remarkable level of success in image synthesis. GANs help in generating significant and relevant outcomes from convoluted patterns as input. This piece of research aims at making a call towards conservation of tribal art culture through deploying neural models and computer vision techniques, in line with Warli tribal human synthetic stick figures of Asian tribes. This paper presents several methods for synthesizing human stick figures in Asian Tribal Art using modified GAN architecture. The modified GANs were quite different in terms of layers as well as the range and distribution of latent vector space. To provide input data for the training process, a customized data set was created based on spherical polar coordinates and the laws of coordinate geometry. The process of training the GAN architecture involves insertion of random vector points, and addition of neural layers, to generate outputs of higher and better resolution. The paper includes training of Generator and Discriminator following the concepts of Transfer learning and providing relevant affirmation to draw conclusions that the generated samples exhibit similarities and resemble the original input sample. Further work extends towards using of different color space data set images for generating outputs with better accuracy through the modified GAN models. The final generated output images along with other visualizations are also attached with this work to provide evidence for the fact that the output samples are analogous to the actual data. This piece of research tries to open a gate way of using artificial intelligence in terms of cultivating and protecting the traditional art cultures of various tribes across the globe. The proposed methodology performed quite well in terms of producing warli figures from given latent vector space, which is vindicated by the comparison of images at pixel level and image histograms.