Search

Considering its unique operational characteristics, the autonomous mobility-on-demand (AMoD) system is playing a more and more critical role in the future transportation system. In this article, we consider a joint planning problem of autonomous fleet (i.e., vehicle type and fleet size) and charging infrastructure (i.e., the size of charging piles with various charging rates, distributed renewable resources) while accounting for the spatial–temporal demand-side flexibility of the AMoD system. We adopt an augmented network flow model to capture the autonomous fleet's charging, order serving, and rebalancing decisions. The proposed planning problem is modeled as a two-stage stochastic program and solved by multicut Benders decomposition techniques. Simulation experiments based on real-world data are performed to demonstrate the effectiveness of the proposed planning scheme. [ABSTRACT FROM AUTHOR]

This article proposes a high-gain transformerless three-port converter (TPC) for standalone photovoltaic (PV) systems. The TPC is designed and developed based on a dual-inductor high-gain two-port converter by utilizing one of the buffer capacitors to derive the third part for PV input. A hybrid pulse-frequency modulation scheme unified with a pulsewidth modulation control strategy is adopted, which realizes the maximum power point tracking control, load voltage regulation, and bidirectional energy flow at the battery port. The proposed TPC offers the unique advantages of high voltage gain, continuous battery port current, reduced power semiconductors, lower voltage stresses, common ground shared by all the ports, low-cost gate driver, and small size of the rear-end inductor. The working principle, steady-state characteristics, small-signal models, and the control method, including design conditions, are comprehensively analyzed. The correctness of the theoretical analysis is verified by developing a 300-W experimental prototype, which shows that the maximum efficiency is 97.7%. [ABSTRACT FROM AUTHOR]

Human brain cortex acts as a rich inspiration source for constructing efficient artificial cognitive systems. In this paper, we investigate to incorporate multiple brain-inspired computing paradigms for compact, fast and high-accuracy neuromorphic hardware implementation. We propose the TripleBrain hardware core that tightly combines three common brain-inspired factors: the spike-based processing and plasticity, the self-organizing map (SOM) mechanism and the reinforcement learning scheme, to improve object recognition accuracy and processing throughput, while keeping low resource costs. The proposed hardware core is fully event-driven to mitigate unnecessary operations, and enables various on-chip learning rules (including the proposed SOM-STDP & R-STDP rule and the R-SOM-STDP rule regarded as the two variants of our TripleBrain learning rule) with different accuracy-latency tradeoffs to satisfy user requirements. An FPGA prototype of the neuromorphic core was implemented and elaborately tested. It realized high-speed learning (1349 frame/s) and inference (2698 frame/s), and obtained comparably high recognition accuracies of 95.10%, 80.89%, 100%, 94.94%, 82.32%, 100% and 97.93% on the MNIST, ETH-80, ORL-10, Yale-10, N-MNIST, Poker-DVS and Posture-DVS datasets, respectively, while only consuming 4146 (7.59%) slices, 32 (3.56%) DSPs and 131 (24.04%) Block RAMs on a Xilinx Zynq-7045 FPGA chip. Our neuromorphic core is very attractive for real-time resource-limited edge intelligent systems. [ABSTRACT FROM AUTHOR]

For embedded, mobile and edge-computing intelligent applications, this brief proposes a low-cost real-time neuromorphic hardware system of spiking Extreme Learning Machine (ELM) equipped with on-chip triplet-based reward-modulated spike-timing-dependent plasticity (R-STDP) learning capability. Our design employs a time-step pipelined dual-core architecture consisting of parallel computing unit arrays to improve processing speed, as well as a trace-assisting learning mechanism and on-the-fly hidden layer weight re-generators to significantly reduce hardware resource costs. Our architecture is scalable to different spiking ELM sizes under different tradeoffs among processing speed, recognition accuracy and resource costs. Tests showed that the on-chip triplet R-STDP learning capability can help to achieve relatively high recognition accuracies on our hardware system. An FPGA prototype with low logic and memory resource consumption was implemented, achieving 93% and 78.5% recognition accuracies on the MNIST and Fashion-MNIST image datasets, respectively, at a speed of 30 frames per second (fps) for inference and 22.5 fps for on-chip learning. [ABSTRACT FROM AUTHOR]

A family of transformer-less single-switch dual-inductor high voltage gain boost converters is proposed in this article. The proposed configurations can realize high voltage gain with low voltage and current stresses. The voltage gain of the proposed converters is analogous to the switched inductor boost converter; therefore, the proposed boost converters are suitable for high gain applications. The proposed family of converters has low component counts and reduced voltage stress on devices compared to three typical transformer-less single-switch high voltage gain converters including switched inductor boost converter, quadratic boost converter, and quasi-Z-source boost converter. Moreover, the current stress of the front-end diode and the rear-end inductor is also relatively low. Therefore, the conversion efficiency is enhanced while keeping the cost low. The operation principles and steady-state characteristics analysis of the proposed converters under the inductor current continuous conduction mode and continuous bidirectional conduction mode are discussed in detail. A prototype of 300 V, 250 W/100 kHz is developed to verify the performance of the proposed converter. The experimental results substantiate the effectiveness and advantages. The prototype achieves a peak efficiency of 97.5 $\%$. [ABSTRACT FROM AUTHOR]

Much research from open literature shows that polarization diversity can provide another dimension which may be exploited to improve the performance in ground moving target indication (GMTI), compared with space–time adaptive processing (STAP). In this article, we report the multichannel synthetic aperture radar (SAR)-based GMTI (SAR-GMTI) experiment and its preliminary results with a N-SAR system which is an airborne quadrature-polarimetric (quad-pol) radar system. First, the joint polarization-space adaptive processing (JPolSAP) is performed in the image level, but two suboptimal versions of JPolSAP, where the polarimetric matched filter (PMF) vector and the full-one vector are exploited to substitute for the polarimetric steering vector, respectively, are evaluated since the polarimetric steering vector of the moving target is unknown precisely in practical applications. Then, considering the computational complexity and lack of secondary data in a inhomogeneous environment due to high degrees of freedom of the JPolSAP processor, two cascade processors are evaluated, including the polarization enhancement that uses PMF and noncoherence integration detection (NCID) technique. Furthermore, we utilize the polarization information to accomplish SAR terrain classification, and subsequently secondary data from the same scattering type clutter can be obtained for clutter suppression under the guidance of polarization classification results as a priori knowledge. Finally, the experimental results demonstrate that: 1) the suboptimal JPolSAP processor with PMF steering vector can effectively enhance GMTI performance about 13 dB (or even up to 5 dB) relative to the worst (or best) single-polarization (S-pol) processor case and has the best robustness compared with the one with full-one steering vector; 2) polarization enhancement using PMF also obtains a good output gain of polarization filter, especially for quad-pol enhancement, which gains up to 2–3 dB with respect to the best output of S-pol processor, and the NCID technique can obtain good performance of moving-target detection; and 3) under the guidance of polarization classification results, the capability of clutter suppression can improve even up to 15 dB with respect to the one without classification. [ABSTRACT FROM AUTHOR]

Synthetic aperture radar (SAR) operating in an along-track interferometric (ATI) mode has the advantage of minimum detectable velocity (MDV) in active marine target detection. However, most of the conventional ATI detectors fail to identify the marine targets cruising with blind speeds, in which case the interferometric phases of these targets are closely distributed around those of the sea background due to phase wrapping and aliasing through ATI processing. To address this issue, we propose a method to detect the active marine targets based on complex interferometric dissimilarity for dual-channel ATI-SAR systems. First, an interferometric bilateral filter is designed to smooth the random noises and locally adapt to the spatial structure of the interferogram for measuring the interferometric magnitude and phase. Then, the target detection metric is constructed based on the complex interferometric dissimilarity between the marine targets and the sea background. By adaptively regressing into a magnitude-based test toward the blind-speed targets, the target detection metric can mitigate the blind-speed detection problem and thus yield a satisfactory detection result. Furthermore, this metric is of a constant false-alarm rate (CFAR), and its probability density function (pdf) is derived to facilitate the detection threshold computation. Finally, both the simulated and real-data processing results are given to validate the superiorities of the proposed method. [ABSTRACT FROM AUTHOR]

To address the problems of location and suppression of inshore azimuth ambiguous clutter for the azimuth multichannel synthetic aperture radar (SAR) systems, a multifeature autosegmentation-based approach is developed in this paper. This proposed method can segment a SAR image automatically according to the distinctions among main land clutter, ambiguous land clutter, and sea clutter in the features of interferogram's phase and magnitude. First, the finite mixture clutter model for a multilook covariance matrix (MLCM) is built, where the off-diagonal elements of the MLCM contain the information of magnitude and interferogram's phase between azimuth channels. Then, SAR image autosegmentation is carried out by using the expectation maximum algorithm with combination of the aforementioned mixture model, and the isolated points that are segmented incorrectly can be eliminated via exploiting the Markov random field smoothing technique. Finally, azimuth ambiguous clutter can be suppressed by means of the clutter covariance matrix, which is constructed by the training samples of segmented ambiguities. The experiments on simulated data and real data measured by TerraSAR-X demonstrate that the proposed approach can obtain the more accurate position information and good cancellation performance for the azimuth ambiguous clutter, without the accurate system parameters and the information of the sources account for azimuth ambiguities. [ABSTRACT FROM AUTHOR]

Simultaneous measurement of hydrogen concentration and temperature with high sensitivity and high precision was realized by integrating a photonic crystal fiber (PCF) modal interferometer and a high-birefringence fiber loop mirror (HBFLM). The PCF was coated with sol-gel Pd/WO3 coating, whose refractive index and volume would be changed along with the variation of surrounding hydrogen concentration or temperature, and then the resonant dip of PCF modal interferometer would shift accordingly. Meanwhile, the resonant dip of the HBFLM has a high sensitivity to surrounding temperature. By integrating PCF modal interferometer and HBFLM together, the two independent resonant dips that can be simultaneously monitored at the output spectrum of the integrated interferometers would all shift with the change of external hydrogen concentration or temperature. Finally, combined with the dual-wavelength matrix method, the hydrogen concentration and temperature could be simultaneously measured. Experimental results showed that the resonant dip of the PCF modal interferometer moved in the short wavelength direction with the increase of hydrogen concentration, and the sensitivity of the hydrogen concentration was −1.12 nm/% within the concentration range from 0% to 1%. In the process of temperature change, the HBFLM realized a higher temperature sensitivity of −1.84 nm/°C. The sensor had high sensitivity, low cost, and simple structure. [ABSTRACT FROM PUBLISHER]