Search

With a spurt of progress in deep learning techniques, convolutional neural network-based and transformer-based methods have yielded impressive performance on the hyperspectral image (HSI) classification tasks. However, pixel-level manual annotation is time-consuming and laborious, and the small amount of labeled HSI data brings challenges to deep learning methods. Existing methods use carefully designed network architectures combined with self-supervised or semi-supervised learning to deal with the lack of training samples. Those methods were designed for specific datasets and often needed to tune hyperparameters on new datasets carefully. To tackle this problem, a unified HSI masked autoencoder framework was proposed for HSI classification. Different from existing works, the hyperspectral image masked autoencoder (HSIMAE) framework was pretrained on a large-scale unlabeled HSI dataset, named HSIHybrid, which contained a large amount of HSI data acquired by different sensors. First, to handle the different spectral ranges of HSIs, a group-wise PCA was applied to extract features of HSI spectra and transform them into fixed-length vectors. Then, a modified masked autoencoder was proposed for large-scale pretraining. It utilized separate spatial–spectral encoders followed by fusion blocks to learn spatial correlation and spectral correlation of HSI data. Finally, to leverage the unlabeled data of the target dataset, a dual-branch finetuning framework that used an extra unlabeled branch for mask modeling learning was introduced. Extensive experiments were conducted on four HSI datasets from different hyperspectral sensors. The results demonstrate the superiority of the proposed HSIMAE framework over the state-of-the-art methods, even with very few training samples.

Underwater target detection has developed greatly in recent years. However, the accuracy of underwater target detection is limited by the complex underwater environment. Based on YOLOv7, we propose an underwater object detection algorithm model to improve precision and confidence (BiFusion Neck module and a MPDIoU loss function). Compared to traditional networks module, the Bifusion Neck module preserves more features from the lower layers by utilizing the output of the P2 feature layer. Moreover, the loss function was improved on the basis of IoU introducing Minimum Point Distance. Finally, the LSKA attention mechanism is introduced to enhance the feature extraction of targets at different scales. The experimental results demonstrate that the BFD-YOLO model proposed in this study achieves an average detection accuracy(mAP50) of 84.8% on a customized dataset, surpassing the performance of the YOLOv7 algorithm by 11.5% and outperforming other tested algorithms. Furthermore, the BFD-YOLO algorithm exhibits strong performance on various datasets and demonstrates superior generalization capabilities.

Low carbon power system with high penetration of clean energy is an effective way to realize the carbon emission target. Long term power system planning should consider both technical constraints and reasonable investment cost forecasting. For reasonable investment cost forecasting in long term, the effect on investment cost by technical progress and scale effect should be taken into consideration both. Investment cost can affect the planning results of installed capacity, while installed capacity affects the investment cost forecasting result mutually. There is no paper that takes technical effect into investment cost forecasting or analyzes the mutual effect between planned installed capacity and investment cost forecasting. Technical progress can be quantified by TRL (Technical Readiness Level) while scale effect can be quantified by installed capacity. The novelty of this paper is proposing the 3D curve function which qualified the relationship between investment cost and technical progress & installed capacities for the first time to realize investment forecasting in long term. And the 3D curve is combined with the GESTP model by feedback the forecasting investment cost to power system planning which reveals the mutual effect between installed capacity planning and investment cost forecasting for the first time. The study case indicates that accelerating of technical progress will decrease the investment cost and increase the installed capacity of this technology and affect other technologies with interaction in the whole power system.

This paper proposes a reverse procedure detection method for space target streaks based on motion parameter estimation. According to the phase shift characteristics of the spectrum, the interframe phase difference is used to obtain the displacement vectors of the target streak and image background, and then the morphological parameters of the target streak are estimated. In addition, a streak window is designed on the basis of the parameters, utilizing the local grayscale correlation in two image frames to carry out target searching and positioning with the interval of the target displacement vector. Images from different scenes are used to evaluate the adaptability of our method, and the experimental results suggest that the expected targets can be detected successfully with subpixel precision in all telescope operating modes. The proposed frequency-domain reverse procedure detection method avoids the problem of fake target confirmation caused by blind detection in traditional methods and overcomes the interference produced when the target streak passes through a bright star. Moreover, first target tracking and then positioning, which cannot be performed with time-domain methods, can be achieved. Our proposed method is attractive for multiple applications, such as space target cataloging, early warning and orbit prediction.

We present a structured overview of adaptation algorithms for neural network-based speech recognition, considering both hybrid hidden Markov model / neural network systems and end-to-end neural network systems, with a focus on speaker adaptation, domain adaptation, and accent adaptation. The overview characterizes adaptation algorithms as based on embeddings, model parameter adaptation, or data augmentation. We present a meta-analysis of the performance of speech recognition adaptation algorithms, based on relative error rate reductions as reported in the literature.

Learning from demonstrations with Probabilistic Movement Primitives (ProMPs) has been widely used in robot skill learning, especially in human-robot collaboration. Although ProMP has been extended to multi-task situations inspired by the Gaussian mixture model, it still treats each task independently. ProMP ignores the common scenario that robots conduct adaptive switching of the collaborative tasks in order to align with the instantaneous change of human intention. To solve this problem, we proposed an alternate learning-based parameter estimation method and an empirical minimum variation-based decomposition strategy with projection points, combining with linear interpolation strategy for weights, based on a Gaussian mixture model framework. Alternate learning of weights and parameters in multi-task ProMP (MTProMP) allows the robot to obtain a smooth composite trajectory planning which crosses expected via points. Decomposition strategy reflects how the desired via point state is projected onto the individual ProMP component, rendering the minimum total sum of deviations between each projection point with the respective prior. Linear interpolation is used to adjust the weights among sequential via points automatically. The proposed method and strategy are successfully extended to multi-task interaction ProMPs (MTiProMP). With MTProMP and MTiProMP, the robot can be applied to multiple tasks in industrial factories and collaborate with the worker to switch from one task to another according to changing intentions of the human. Classical via points trajectory planning experiments and human-robot collaboration experiments are performed on the Sawyer robot. The results of experiments show that MTProMP and MTiProMP with the proposed method and strategy perform better.

The explicit expression of orbital angular momentum (OAM) spiral spectrum of Laguerre-Gaussian (LG) beam propagating in anisotropic plasma turbulence is derived based on Rytov theory. The relationship between the detection probability of OAM and the crosstalk with the anisotropy parameters, topological charge, radial index, wavelength, beam width, receiving aperture, inner and outer scales are discussed in detail. In order to quantify the effect of plasma turbulence, we compared it with atmospheric turbulence based on restriction function. To optimize OAM detection probability, we also compared the effects of circular aperture method (CAM) with focusing mirror method (FMM) on the propagation characteristics of the spiral spectrum. The results show that even in the strong turbulence region of atmospheric turbulence, the radial dimension of LG beam in plasma turbulence is nearly twice that of atmospheric turbulence and the modes crosstalk has almost reached its upper bound. Besides, by comparing the two optimization methods, we found the ability to reduce crosstalk is related to the choice of beam width and receiving aperture. Therefore, we give a method to optimally select between the CAM and the FMM. We found the corresponding relationship and gave an analytical formula. The FMM has a greater ability to reduce crosstalk than the CAM when the beam width ≥4 cm and the aperture ≥3 cm. We quantified the effect of plasma turbulence and provide a guideline for the selection of beam width and receiving aperture under these two optimization methods. The research results may be helpful in the field of optical communication.

The renewable energy sources (RESs) dominated power grid is an envisaged infrastructure of the future power system, where the commonly used grid following (GFL) control for grid-tied converters suffers from lacking grid support capability, low stability, etc. Recently, emerging grid forming (GFM) control methods have been proposed to improve the dynamic performance and stability of grid-tied converters. This paper reviews existing GFM control methods for the grid-tied converters and compares them in terms of control structure, grid support capability, fault current limiting, and stability. Considering the impact of fault current limiting strategies, a comprehensive transient stability analysis is provided. In addition, this paper explores the typical applications of GFM converters, such as AC microgrid and offshore wind farm high-voltage direct current (OWF-HVDC) integration systems. Finally, the challenges to the GFM converters in future applications are discussed.

In industrial production processes, the data usually have high-dimensional characteristics. When a support vector machine (SVM) is used for fault detection, it takes a long time to run. For high-dimensional data, principal component analysis (PCA) and independent component analysis (ICA) are very effective dimensionality reduction algorithms. Since the data follow different distributions in industrial processes, PCA is commonly used to process Gaussian-distributed data, and ICA is used to process non-Gaussian distributed data. Addressing these limitations, a novel fault detection method of SVM based on Gaussian and non-Gaussian double subspace (DSSVM) is proposed. The Kolmogorov-Smirnov (KS) test is used to determine the normal distribution characteristics of the process variables in the original data. The process variables are divided into a Gaussian subspace and a non-Gaussian subspace. Fault detection models of the Gaussian subspace are established based on PCA, and models of the non-Gaussian subspace are based on ICA. To reduce the effect of autocorrelation among process variables on the SVM, the delay and time difference input characteristics are introduced into the principal components obtained in the Gaussian subspace and independent components obtained in the non-Gaussian subspace. Finally, the time delay matrix and time difference matrix are combined, and the SVM model is used for fault detection and monitoring. The DSSVM method reduces the data dimensions and eliminates the effect of autocorrelation among process variables on the detection results. The proposed method is applied to a multivariable numerical simulation and the Tennessee-Eastman industrial process. Comparisons with simulation results for the kernel principal component analysis (KPCA), kernel independent component analysis (KICA), SVM and the statistical process monitoring method based on variable distribution characteristics (VDSPM) further verify the effectiveness of the algorithm.

Daugman’s design of IrisCode continues to be fascinating in the research world with its practicality, efficiency, and outstanding performance. The limits of Daugman’s recognition system, however, remain a topic of active discussion. Multiple approaches to scale performance have been explored in the past. Despite them, the problem of finding the maximal population of IrisCode remains open. Because of this, we appeal to Rate-Distortion theory (limits of error-correction codes) to establish bounds on the maximum possible population of iris classes that IrisCode can support under the constraint of a minimum Hamming Distance (HD) between any two codewords. This approach considers the distribution of iris data within and across iris classes and the quality of iris data. We first present the Hamming, Plotkin, and Elias-Bassalygo upper bounds and the Gilbert-Varshamov lower bound on the population of IrisCode. The bounds relate the number of iris classes that the IrisCode algorithm can sustain and the quality of iris data expressed in terms of HD. Then, we analyze our results and draw conclusions regarding the relationship of IrisCode population size and the level of quality that enrolled data must have to ensure a particular population coverage. By applying the theory presented here, researchers can better understand what maximum population is achievable based on the quality of their iris dataset.