Feedforward Structure Research Articles

Model-Adaptive Feedforward Structure for Enhanced Dynamic Response in SRM Current Controller Based on Operating Point Transformation

Model-Adaptive Feedforward Structure for Enhanced Dynamic Response in SRM Current Controller Based on Operating Point Transformation

A modified feedforward-feedback time-frequency domain hybrid algorithm for multichannel active road noise control system

Multi-channel Active Road Noise Control (ARNC) systems require numerous adaptive filters to reduce noise at multiple positions inside the vehicle. However, road noise conditions are more complex, and the correlation between the reference signals and the target noise is often poor. When the driving conditions become slightly complex, the correlation between the reference signal and the target noise is often poor. As a result, traditional road noise control algorithms applied to multi-channel ARNC systems often suffer from high computational cost and poor noise reduction performance in non-steady-state and high-speed driving conditions. To address this issue,a reference signal selection method based on the genetic algorithm (GA) is proposed to obtain a reference signals combination with stronger coherence. On this basis, we introduce the time–frequency domain structure into the forward-feedback hybrid algorithm and a multi-channel time–frequency domain hybrid algorithm with a novel variable step size method (VSS-TF-HANC) is proposed in this paper.The feedforward part adopts a time–frequency domain structure and introduces a novel variable step-size strategy to reduce computational complexity and improve noise reduction performance. The feedback part adopts a control system based on the Internal Model Control (IMC) architecture to attenuate the interior noise components that have a strong response but poor correlation with the reference signals. Simulation results demonstrate that the proposed algorithm achieves excellent control results under complex driving conditions, with significantly better noise reduction performance compared to the control algorithms that solely employ the feedforward structure, while also exhibiting low computational cost.

A 2nd-order noise-shaping SAR-assisted pipeline ADC with order-boosted gain-error-shaping

This paper presents a 4th-order interstage gain error shaping (GES) technique in pipeline successive approximation register (SAR) analog-to-digital converters (ADCs), which can substantially suppress the in-band quantization leakage error induced by the gain error. It is realized by simply arranging a low-order cascaded-integrator feed-forward (CIFF) structure in the first stage. In addition, the comparator noise and quantization error can be shaped together with the gain error in the proposed architecture. Verified by simulation in a 28-nm CMOS process, the prototype achieves a signal-to-noise-and-distortion ratio (SNDR) of 77.8 dB over 25-MHz bandwidth (BW) with oversampling ratio (OSR) of 8. Within a gain error range of −33 % to +33 %, the SNDR of the ADC deviates less than 3 dB. Under a 1 V supply voltage, the ADC consumes 3.75 mW and exhibits a Scherier figure of merit (FoMs) of 176 dB.

Design Principles for Perfect Adaptation in Biological Networks with Nonlinear Dynamics.

Establishing a mapping between the emergent biological properties and the repository of network structures has been of great relevance in systems and synthetic biology. Adaptation is one such biological property of paramount importance that promotes regulation in the presence of environmental disturbances. This paper presents a nonlinear systems theory-driven framework to identify the design principles for perfect adaptation with respect to external disturbances of arbitrary magnitude. Based on the prior information about the network, we frame precise mathematical conditions for adaptation using nonlinear systems theory. We first deduce the mathematical conditions for perfect adaptation for constant input disturbances. Subsequently, we translate these conditions to specific necessary structural requirements for adaptation in networks of small size and then extend to argue that there exist only two classes of architectures for a network of any size that can provide local adaptation in the entire state space, namely, incoherent feed-forward (IFF) structure and negative feedback loop with buffer node (NFB). The additional positiveness constraints further narrow the admissible set of network structures. This also aids in establishing the global asymptotic stability for the steady state given a constant input disturbance. The proposed method does not assume any explicit knowledge of the underlying rate kinetics, barring some minimal assumptions. Finally, we also discuss the infeasibility of certain IFF networks in providing adaptation in the presence of downstream connections. Moreover, we propose a generic and novel algorithm based on non-linear systems theory to unravel the design principles for global adaptation. Detailed and extensive simulation studies corroborate the theoretical findings.

Differences in the consolidation by spontaneous and evoked ripples in the presence of active dendrites.

Ripples are a typical form of neural activity in hippocampal neural networks associated with the replay of episodic memories during sleep as well as sleep-related plasticity and memory consolidation. The emergence of ripples has been observed both dependent as well as independent of input from other brain areas and often coincides with dendritic spikes. Yet, it is unclear how input-evoked and spontaneous ripples as well as dendritic excitability affect plasticity and consolidation. Here, we use mathematical modeling to compare these cases. We find that consolidation as well as the emergence of spontaneous ripples depends on a reliable propagation of activity in feed-forward structures which constitute memory representations. This propagation is facilitated by excitable dendrites, which entail that a few strong synapses are sufficient to trigger neuronal firing. In this situation, stimulation-evoked ripples lead to the potentiation of weak synapses within the feed-forward structure and, thus, to a consolidation of a more general sequence memory. However, spontaneous ripples that occur without stimulation, only consolidate a sparse backbone of the existing strong feed-forward structure. Based on this, we test a recently hypothesized scenario in which the excitability of dendrites is transiently enhanced after learning, and show that such a transient increase can strengthen, restructure and consolidate even weak hippocampal memories, which would be forgotten otherwise. Hence, a transient increase in dendritic excitability would indeed provide a mechanism for stabilizing memories.

A 65 nm 75 MHz–260 MHz sixth-order Active-RC bandpass filter

-A sixth-order Chebyshev-I active-RC bandpass filter was fabricated in a 65 nm CMOS process. An operational amplifier (OPA) with an improved feedforward compensation structure is achieved, which significantly increases bandwidth and gain while reducing power consumption. The tuning module enables the filter to measure a center frequency of 75 MHz–260 MHz, and the bandwidth, Q-factor and gain can be configured according to different application requirements. The effective area of the filter chip is 0.64 mm2, and it can operate at a voltage of 2.5 V with a current dissipation of 34.3 mA.

Prediction of pressure evolution in non-venting self-pressurized liquid hydrogen tanks using artificial neural network approach

Liquid hydrogen (LH2) is a future energy carrier compared to most conventional fuels. One of the practical bottlenecks in non-venting LH2 tanks is to accurately predict the pressure evolution. Accordingly, many thermal models and correlations were developed; however significant deviations arose due to inherent complex thermal phenomena which have not been addressed yet. A general artificial neural network (ANN) model has been first developed based on 448 datasets amassed from different literature sources to predict the pressure evolution under different heat fluxes, initial filling ratios, as well as tank volumes and shapes. The model has been configured with a typical multi-layer feed-forward structure using a back-propagation learning algorithm and Levenberg-Marquardt training algorithm. For the ANN model, a preliminary optimization process has been conducted for reaching the optimal values for training-to-testing datasets ratio, learning parameters and optimum network structure. Besides, some criteria were set to evaluate the ANN model performance. The ANN model with the configuration of 5-11-11-1 was found to be adopted for such a prediction task. Compared to experimental results and literature models and correlations, the ANN model has been proven to predict the pressure evolution in non-venting self-pressurized LH2 tanks with a mean square error of 1.1779 × 10−5 s2, root mean square error of 0.0034 s, mean absolute error of 6%, mean relative error of 0.2% as well as a satisfactory correlation coefficient of 0.99934 between the network outputs and corresponding targets. This research work provides an effective modeling approach based on artificial neural network to solve one complex problem related with LH2 tanks with fast, accurate and consistent results. Also, the results of the present work can offer some practical guidance for the design of LH2 tanks. Further, the present work gives some insights into new application areas of ANN which will be the emphasis of future research work.

Low-complexity and low-latency equalization technique - probabilistic noise cancellation.

Both inside data centers (DCs) and in short optical links between data centers (DC campuses), intensity-modulation and direct-detection (IMDD) systems using four-level pulse amplitude modulation (PAM4) will dominate this decade due to low transceiver price and power consumption. The next DC transceiver generation based on 100 Gbaud PAM4 will require advanced digital signal processing (DSP) algorithms and more powerful forward error correction (FEC) codes. Because of bandwidth limitations, the conventional DC DSP based on a few-tap linear feed-forward equalizer (FFE) is likely to be upgraded to more complex but still low-complexity Volterra equalizers followed by a noise whitening filter and either a maximum likelihood sequence estimation (MLSE) or a maximum a posteriori probability (MAP) algorithm. However, stringent power consumption and latency requirements may limit the use of complex algorithms such as decision feedback equalizer (DFE) or MLSE/MAP in DC networks (DCN). In this paper, we introduce a low-complexity, low-latency algorithm based on a feedforward structure, yielding a performance between DFE and MLSE. We call the novel equalization algorithm probabilistic noise cancellation (PNC), since it weights noise patterns based on their probabilities in the presence of bandwidth limitations. The probabilistic weighting is efficiently exploited in correcting correlated errors caused by noise coloring in the FFE.

Efficient crowd density estimation with edge intelligence via structural reparameterization and knowledge transfer

Crowd stampedes and related incidents pose significant dangers to public safety and have resulted in numerous fatalities over the past few decades. Estimating crowd density in real-time can help avoid stampedes by providing early warning systems to prevent overcrowding and manage the flow of people. With the increasing prevalence of Internet of Things (IoT), the application of edge computing for field pedestrian density estimation can help to enhance security and efficiency of system. To deal with the issue of crowd scale variation, most previous works rely on heavy backbone networks or complex module, which require high runtime consumption and severely limits the deployment scope of their work. To overcome this issue, we propose multi-branch model Repmobilenet. For multi-scale spatial feature extraction. Repmobilenet equips with lightweight multi-branch depthwise separable convolutional block (DSBlock), which can effectively extract multi-scale feature of dense crowd to cope with scale variation. In the inference phase, the multi-branch structure can be transformed into a single-branch feed-forward structure through structural reparameterization. By this way, Repmobilenet can use multi-branch over-parameterized topology to capture more dense spatial features during training stage and decrease inference latency during inference stage. We also added dilated convolutions in the backend to expand the receptive field of the model. Comparison to state-of-the-art methods, proposed Repmobilenet is able to achieve comparable counting performance while maintaining small model size and low inference latency in ShanghaiTech benchmark. At last, we introduce a multi-layer knowledge distillation method to further increase the model’s fitting capability. By imitating the feature of multiple intermediate layers and final output soft ground truth of the teacher model, the student model can learn compact and efficient knowledge without increasing model size and inference latency. The code can be found in https://github.com/lcxxxiii/Repmobilenet.

Artificial neural network and genetic algorithm coupled fermentation kinetics to regulate L-lysine fermentation

Fermentation plays a pivotal role in the industrialization of bioproducts, yet there is a substantial lag in the fermentation process regulation. Here, an artificial neural network (ANN) and genetic algorithm (GA) coupled with fermentation kinetics were employed to establish an innovative lysine fermentation control. Firstly, the strategy of coupling GA with ANN was established. Secondly, specific lysine formation rate (qp), specific substrate consumption rate (qs), and specific cell growth rate (μ) were predicted and optimized by ANN-GA. The optimal ANN model adopts a three-layer feed-forward back-propagation structure (4:10:1). The optimal fermentation control parameters are obtained through GA. Finally, when the carbon to nitrogen ratio, residual sugar concentration, ammonia nitrogen concentration, and dissolved oxygen were [2.5, 4.5], [6.5, 9.5] g·L−1, [1.0, 2.0] g·L−1 and [20, 30] %, respectively, the lysine concentration reaches its peak at 213.0 ± 5.10 g·L−1. The novel control strategy holds significant potential for optimizing the fermentation of other bioproducts.

A Low-Distortion Low-Power MASH Architecture ΣΔ Modulator with Double-Sampling

This paper presents a low-distortion and low-power multi-stage noise-shaping (MASH) architecture Sigma-Delta ([Formula: see text] modulator with double-sampling. The modulator uses a cascade of integrator feed-forward (CIFF) structure as a cascaded modulator in a MASH architecture. Since the CIFF structure provides a feedforward path from the input signal to the quantizer, the input signal is directly transmitted to the output without loop processing, relaxing the linearity requirement of the modulator, reducing performance requirements for the OTA and harmonic distortion. Double-sampling technique is adopted in the first integrator to double the sampling rate, improving the accuracy of the modulator without the increase of power consumption. In order to eliminate the influence of capacitor mismatch in the sampling network, a fully floating differential switched capacitor (SC) feedback digital-to-analog (DAC) is used in the integrator. The simulation results show that the [Formula: see text] modulator achieves 113[Formula: see text]dB signal-to-noise and distortion ratio (SNDR) and the power consumption is 3.15[Formula: see text]mW from a 5[Formula: see text]V power supply.

Towards practical single-shot phase retrieval with physics-driven deep neural network.

Phase retrieval (PR), a long-established challenge for recovering a complex-valued signal from its Fourier intensity-only measurements, has attracted considerable attention due to its widespread applications in optical imaging. Recently, deep learning-based approaches were developed and allowed single-shot PR. However, due to the substantial disparity between the input and output domains of the PR problems, the performance of these approaches using vanilla deep neural networks (DNN) still has much room to improve. To increase the reconstruction accuracy, physics-informed approaches were suggested to incorporate the Fourier intensity measurements into an iterative estimation procedure. Since the approach is iterative, they require a lengthy computation process, and the accuracy is still not satisfactory for images with complex structures. Besides, many of these approaches work on simulation data that ignore some common problems such as saturation and quantization errors in practical optical PR systems. In this paper, a novel physics-driven multi-scale DNN structure dubbed PPRNet is proposed. Similar to other deep learning-based PR methods, PPRNet requires only a single Fourier intensity measurement. It is physics-driven that the network is guided to follow the Fourier intensity measurement at different scales to enhance the reconstruction accuracy. PPRNet has a feedforward structure and can be end-to-end trained. Thus, it is much faster and more accurate than the traditional physics-driven PR approaches. Extensive simulations and experiments on an optical platform were conducted. The results demonstrate the superiority and practicality of the proposed PPRNet over the traditional learning-based PR methods.

Robust Symbol Timing Synchronization for Initial Access under LEO Satellite Channel.

This paper proposes a robust symbol timing synchronization scheme for return link initial access based on the Digital Video Broadcasting-Return Channel via Satellite 2nd generation (DVB-RCS2) system for the Low Earth Orbit (LEO) satellite channel. In most cases, the feedforward estimator structure is considered for implementing Time Division Multiple Access (TDMA) packet demodulators such as the DVB-RCS2 system. More specifically, the Non-Data-Aided (NDA) approach, without using any kind of preamble, pilot, and postamble symbols, is applicable for fine symbol timing synchronization. However, it hinders the improvement in estimation accuracy, especially when dealing with short packet lengths during the initial access from the User Terminal (UT) to the Gateway (GW). Moreover, when a UT sends a short random access packet for initial access or resource request to the LEO satellite channel, the conventional schemes suffer from a large Doppler error depending on UT's location in a beam and satellite velocity. To ameliorate these problems, we propose a novel symbol timing synchronization algorithm for GW, and its advantage is confirmed through computer simulation.

A 111dB-SNDR 20kHz-BW Delta-Sigma Modulator with Auto-Zeroing and Class-AB Amplifier

This paper describes the design of a high-precision audio Delta-Sigma modulator. To guarantee the modulator linearity and resolution with a limited oversampling rate (OSR), 4th-order 1-bit architecture is chosen. Auto-zeroing technique is used to reduce offset and low-frequency noise. To optimize power consumption, the switched-capacitor loop filter is implemented by feedforward structure with a passive adder and class-AB two-stage amplifiers. A clock generator with optimized non-overlapping pulse widths is designed to improve precision. Simulation results in 180nm CMOS show that with a 10.24MHz sampling rate, the presented modulator achieves 111dB signal-to-noise-and-distortion ratio (SNDR) at 20kHz bandwidth (BW) and consumes 17.9mW power under 3.3V supply, corresponding to an excellent figure-of-merit (FoM) of 171.5dB. Simulation results in process corners and temperatures show that the modulator is robust over variations.

An intermediate frequency reconfigurable bandpass noise‐shaping SAR ADC for IoT and mobile application

AbstractThis letter proposes a reconfigurable bandpass noise‐shaping successive approximation register analog‐to‐digital converter (ADC) with a cascade of integrators with feedforward (CIFF) structure. Unlike conventional CIFF structures that use a passive feedforward path, the new design incorporates an amplifier within the filter, providing effective bandpass noise transfer function regardless of passband location, and facilitating the use of a unity gain 2‐input comparator for low noise and high stability. The loop filter includes amplification by a low power Gm‐R amplifier and two‐step charge sharing. The proposed ADC supports three different intermediate frequencies (IFs), and 12‐bit resolution is achieved for all IFs in a 28‐nm complementary metal‐oxide‐semiconductor (CMOS) process.

A highly accurate model for prediction of thermal conductivity of carbon-based nano-enhanced PCMs using an artificial neural network

Nano-PCMs, which contain nanostructured materials, can enhance the low thermal conductivity of phase change materials (PCMs). It is crucial to predict the precise thermal conductivity of nano-PCM to assess the heat transfer during phase change procedures such as melting and solidification. In this study, artificial neural network (ANN) and response surface method (RSM) were used to develop a model for predicting the thermal conductivity of carbon-based nano-enhanced phase change materials (NEPCMs) using 482 experimental samples collected from various datasets. The carbon-based nano-particles were SWCNTs, MWCNTs, graphite, graphene, and (CNFs). Six input parameters were considered with varying temperature, mass fractions, sizes, thermal conductivities of PCMs and nano-particles, and the phase of PCM. The ANN was designed using a multi-layered feed-forward structure and Levenberg–Marquardt back-propagation algorithm with one hidden layer consisting of eight neurons. The transfer function was varied to achieve optimal performance. It was revealed that the predictive capacity of the ANN model is greater than that of the RSM model based on their corresponding the coefficient of determination (R2) and root mean square error (RMSE) values. The developed ANN achieved a R2 value of 0.99, while the RSM method model achieved an R2 value of 0.79.

Design of A Finite-Time Adaptive Controller for Image-Based Uncalibrated Visual Servo Systems with Uncertainties in Robot and Camera Models.

Aiming at the time-varying uncertainties of robot and camera models in IBUVS (image-based uncalibrated visual servo) systems, a finite-time adaptive controller is proposed based on the depth-independent Jacobian matrix. Firstly, the adaptive law of depth parameters, kinematic parameters, and dynamic parameters is proposed for the uncertainty of a robot model and a camera model. Secondly, a finite-time adaptive controller is designed by using a nonlinear proportional differential plus a dynamic feedforward compensation structure. By applying a continuous non-smooth nonlinear function to the feedback error, the control quality of the closed-loop system is improved, and the desired trajectory of the image is tracked in finite time. Finally, using the Lyapunov stability theory and the finite-time stability theory, the global finite-time stability of the closed-loop system is proven. The experimental results show that the proposed controller can not only adapt to the changes in the EIH and ETH visual configurations but also adapt to the changes in the relative pose of feature points and the camera's relative pose parameters. At the same time, the convergence rate near the equilibrium point is improved, and the controller has good dynamic stability.

APPLICATION OF NEURAL NETWORK IN PREDICTING THE OUTCOME OF CONSTRUCTION DISPUTES

Surat, India The Indian construction sector has become more organized after implementation of the Real Estate (Regulation and Development) Act, 2016. However, occurrence of disagreement between the stakeholders is still a raising question despite the well-established framework of the Real Estate Regulatory Authority (RERA). Project characteristics such as project cost, project type, project scale and anticipated completion time have an impact on causing these disagreements, which ultimately result in a dispute. The disputes might adversely impact the stakeholders associated with them and may lead the project to delay. In order to establish the interrelationship between the project characteristics causing delay due to disputes, an appropriate soft computing technique can be applied. Therefore, this study highlights the application of Artificial Neural Network (ANN) for forecasting the outcome of a dispute. The inputs for the ANN model are project costs, project type, land area acquired and anticipated completion time of disputed projects from RERA. The proposed ANN model of 4-6-1 feed-forward back propagation structure helps to satisfy the need for quick and accurate resolution of disputes with lowering the stress on the judicial system. Moreover, the stakeholders engaged in the dispute can potentially save a considerable amount of money and time if they have an insight of the potential outcome.

An application of artificial neural networks for solving fractional higher-order linear integro-differential equations

This ongoing work is vehemently dedicated to the investigation of a class of ordinary linear Volterra type integro-differential equations with fractional order in numerical mode. By replacing the unknown function by an appropriate multilayered feed-forward type neural structure, the fractional problem of such initial value is changed into a course of non-linear minimization equations, to some extent. Put differently, interest was sparked in structuring an optimized iterative first-order algorithm to estimate solutions for the origin fractional problem. On top of that, some computer simulation models exemplify the preciseness and well-functioning of the indicated iterative technique. The outstanding accomplished numerical outcomes conveniently reflect the productivity and competency of artificial neural network methods compared to customary approaches.

An 18-bit 2-2 MASH delta-sigma modulator for isolated amplifier

This paper presents an 18-bit 2-2 MASH delta-sigma modulator (DSM) for the isolated amplifier. Combining the cascade of integrators with feedforward (CIFF) structure and the cascade of integrators with feedback (CIFB) structure, the DSM effectively results in reduced harmonic distortion. Gain boost and slew rate (SR) boost techniques are used to enhance the performance of the first stage integrator. To reduce low-frequency noise, the chopper technique is adopted as well. The DSM was fabricated in a 0.18 μm CMOS process with an active area of 0.90 mm2, dissipating 29.70 mW under a 3.3 V supply with the Schreier figure-of-merit (FoMs) 170.00 dB. The 108.71 dB signal-to-noise and distortion ratio (SNDR) and 17.77 bits the effective number of bits (ENOB) are achieved by simulation in a 20 kHz bandwidth.