Journal of University of Chinese Academy of Sciences

Reconstruction density estimation based on sequential algorithms

HUANG Siyuan, XIE Tianfa, XIONG Shifeng

2025, 42 (6): 721-728. DOI: 10.7523/j.ucas.2024.018

In this paper, for the density estimation given by the reconstruction approach, an algorithm based on the sequential idea is proposed to solve the node selection problem in the reconstructed density estimation. Since density estimation can be regarded as an unsupervised learning problem, i.e., there is no response variable y, the node sequential selection approach for regression is not applicable here. We regard the node as a parameter and select the next node by minimising the loss function, then determine the entire set of nodes using a greedy algorithm. This algorithm is simple to operate, further improves the estimation effect, and can reduce the impact on density estimation due to different node selection. In addition, in this paper, the prior is given according to the actual meanings of the parameters in the reconstruction approach, the samples of the posterior distribution are obtained using the Metropolis algorithm, so that the interval estimation of the density function point by point is constructed by approximating the overall quartile through the sample quartiles. Finally, we validate the sequential reconstruction density estimation and its interval estimation on several datasets.

Charged boson-fermion system as cosmic dark energy

XIAO Weinan, YE Xuan, ZHANG Yang, A. Marcianò

2025, 42 (6): 729-737. DOI: 10.7523/j.ucas.2024.029

Abstract ( 223 ) PDF (0KB) ( 0 )

The physical origin of the observed acceleration of the universe has not been well understood. Within the framework of general relativity, dark energy is commonly introduced as being independent of other cosmic components, but such models have certain arbitrariness. In this paper, we consider a low-temperature system for dark energy based on the standard model of particle physics. This system consists of a charged boson condensate, a degenerate fermion gas, and a non-electromagnetic U(1) gauge field that couples both components and is constrained by charge conservation. The charged boson condensate has negative pressure and acts as dark energy, while the degenerate Fermi gas with opposite charge is a secondary component. With this boson-fermion system, we find that the accelerating cosmic expansion occurs for a broad range of model parameters. The dust component decreases with the expansion, the energy density of the boson-fermion component remains nearly constant, and the acceleration occurs at a redshift z ~0.7. Within the reasonable parameter space, this model provides a good explanation for the observational data of type Ia supernova and baryon acoustic oscillations.

Numerical simulation for penetrative Rayleigh-Bénard convection in a rotating system

WANG Song, CAO Yuhui

2025, 42 (6): 738-746. DOI: 10.7523/j.ucas.2024.020

Abstract ( 186 ) PDF (0KB) ( 0 )

The rotating penetrative convection in the fields of Earth science and engineering has attracted extensive attention. Due to the density inversion property of water near 4℃, cold water is used as the working fluid in the present paper to study the rotating penetrative Rayleigh-Bénard convection in a vertical annulus. Direct numerical simulation is performed to analyze the convective heat transfer of cold water under various parameter conditions, with the density inversion parameter θ_m=0.0,0.5, the inverse Rossby number 1/Ro and the Rayleigh number Ra changing in the ranges 0≤1/Ro≤10 and 10⁴≤Ra≤10⁸. The present results show that in the non-rotating cases (i.e. 1/Ro=0), the penetrative convection of cold water with θ_m=0.5 exhibits significant up-down asymmetry, with the top thermal boundary layer thickness δ_top^θ greater than the bottom one δ_bottom^θ. The scaling exponents of the Nusselt number Nu and δ^θ versus Ra are approximately ±0.3. In the rotating cases (i.e. 1/Ro>0), the flow changes with increasing the rotation rate (i.e. 1/Ro), leading to the transition of flow regime from thermal plumes to vortex columns at moderate 1/Ro. Particularly noteworthy is that for θ_m=0.0 both the cold and hot plumes are strong enough to form vortex columns in a certain range of 1/Ro, while the density inversion property at θ_m=0.5 renders the cold plumes weak so that only hot plumes can be converted into vortex columns. As a result, the augmentation of heat transfer, induced by the formation of vortex columns, for θ_m=0.5 is not as significant as that for θ_m=0.0. For the rotating penetrative convection of cold water with θ_m=0.5, at moderate to high 1/Ro, the thermal boundary layer thickness δ^θ exhibits a scaling law δ^θ~1/Ro^1/2, while the velocity boundary layer thickness δ^u still follows δ^u~1/Ro^-1/2.

Carbon intensity analysis of Chinese cities based on feature optimization Bayesian classification algorithm

SONG Wenming, ZOU Jialing, TANG Zhipeng

2025, 42 (6): 747-757. DOI: 10.7523/j.ucas.2023.090

Abstract ( 240 ) PDF (0KB) ( 0 )

Cities serve as the primary hubs for human activities, and the successful realization of China’s “Dual Carbon” goals critically hinges on the effective reduction of carbon emissions in urban areas. However, due to the lack of detailed disaggregated data on energy consumption by source, urban carbon emission accounting has emerged as a crucial research area. This study, based on an enhanced Bayesian classification algorithm, leverages provincial-level energy consumption data from 2005 to 2019. It combines this data with various multi-dimensional attributes, including socioeconomic indicators, to determine carbon intensity types. The approach involves training on optimized attributes corresponding to provincial-level carbon intensity and then downscaling them to identify carbon intensity types at the city level. Comparative analysis with data from the carbon emission assessment database system (CEADs) and traditional methods highlights the advantages of the proposed feature-optimized Bayesian classification method. Furthermore, this method unveils the carbon intensity evolution of 282 major Chinese cities from 2005 to 2019, revealing a notable shift from high to low carbon intensity in the majority of cities. Notably, significant disparities persist in carbon intensity types and improvement trends between cities in the northern and southern regions. In the future, special attention should be paid to carbon intensity reduction efforts in resource-rich cities in central and western China. Additionally, the feature-optimized Bayesian classification method proposed in this study exhibits strong scalability, holding promise for applications at smaller scales, including county-level carbon intensity assessments.

Effects of naphthalene stress on treatment performance and microbial community in SBR reactor

GUO Xiaoxiao, LI Zong, GUO Qiucui, LIU Bingxin, CHANG Zhankun, CAO Bing, LIU Xinchun

2025, 42 (6): 758-768. DOI: 10.7523/j.ucas.2024.015

Abstract ( 309 ) PDF (0KB) ( 0 )

This study built a sequencing batch reactor (SBR) in the laboratory, and added different concentrations of naphthalene to form gradient concentration stress. The reactor ran for a total of 166 d. The effects of naphthalene stress on the operating condition of the SBR, extracellular polymeric substances (EPS), and microbial community were studied. The results showed that: 1) Naphthalene stress improved the denitrification performance and total nitrogen removal rate of the reactor; 2) Naphthalene stress promoted the production of EPS, especially protein Ⅰ, fulvic acid, and humic acid; 3) Naphthalene stress affected the diversity and composition of microbial community and increased the abundance of naphthalene degrading genes.

A method for SAR-to-optical image synthesis based on bi-temporal features

WENG Yongchun, MA Yong, CHEN Fu, SHANG Erping, YAO Wutao, ZHANG Shuyan, YANG Jin, LIU Jianbo

2025, 42 (6): 769-780. DOI: 10.7523/j.ucas.2024.007

Abstract ( 231 ) PDF (0KB) ( 0 )

The robust optical image time series are of great value in many applications of remote sensing. However, due to the effects of weather conditions like clouds and rain, it is very difficult to obtain such robust time series of optical images in many regions. Using the all-weather imaging capability of synthetic aperture radar (SAR) to generate optical images from SAR images is an effective solution to the missing data of optical images. But there is still a problem that the quality of generated images in complicated scenarios is much worse than that in simple scenarios. In this paper, we build bi-temporal datasets of different scenarios based on Sentinel imagery and propose an improved generator of conditional generation adversarial network. The encoder-decoder-based generator learns to extract and fuse the bi-temporal polarized SAR features and the additional optical features from the source time phase. In addition, a strategy to balance the weights of SAR and optical features is adopted. Comparison experiments show that our method is the best on FID and PSNR among all evaluated methods. The proposed method significantly reduces the gap in the quality of generated images between simple scenario and complicated scenario. The ablation study shows that our method outperforms the baseline model by 46 in FID, 6.6dB in PSNR and 0.44 in SSIM. Our method efficiently improves the quality of generated images in different scenarios.

An improved on-satellite-processing real-time digital formation technique of multi-zero beams for satellite-based SAR in elevation

CUI Zekai, XIAO Dengjun, QIU Jinsong

2025, 42 (6): 781-791. DOI: 10.7523/j.ucas.2024.013

Abstract ( 323 ) PDF (0KB) ( 0 )

The echoes of the nadir and the range ambiguity point will degrade the quality of high-resolution wide mapping band satellite-based synthetic aperture radar (SAR) imaging, and the use of digital beam forming (DBF) can make the antenna directional map to form a null in the corresponding direction for suppression. It is difficult to implement the existing multi-zero-points DBF processing technology on the satellite. In order to engineer DBF-SAR imaging technology, this paper proposes a set of improved engineering-achievable real-time multi-zero DBF processing method in elevation, which weights the echo signals in real time, and solves the inverse line velocity matrix problem faced in the real-time multi-zero weight generation by using the LDLT decomposition iteration to inhibit the Nadir echoes and range ambiguity in order to increase the range of the wave-position selection. The effectiveness of this real-time processing scheme is verified by multi-point target simulation.

Radar high-speed target detection method based on motion parameter estimation

GUO Zhenfang, SUN Jili, WANG Shuai

2025, 42 (6): 792-805. DOI: 10.7523/j.ucas.2024.011

Abstract ( 227 ) PDF (0KB) ( 0 )

Coherent accumulation over a long period of time has been an effective means to enhance radar echo energy. However, for high-speed moving targets, within a long coherent processing interval, across range unit (ARU) and Doppler frequency migration (DFM) caused by their velocity and acceleration in the radar radial direction cannot be ignored. Additionally, the velocity ambiguity problem caused by high speed also needs to be addressed. This paper proposes a high-speed target detection method based on motion parameter estimation. By segmenting coherent accumulation, the ARU under each segment is suppressed. The peak position of the echo after accumulation under each segment is used as a sample point for parameter estimation input. A filter is constructed using the target radial acceleration obtained through searching to compensate for nonlinear components in the echo phase. Finally, keystone transform is used in conjunction with velocity ambiguity term compensation to correct ARU in radar echoes, and coherent accumulation of the high-speed target echo signal is achieved by performing FFT along the azimuth.

Dual networks with hierarchical attention for fine-grained image classification

YANG Tao, WANG Gaihua

2025, 42 (6): 806-813. DOI: 10.7523/j.ucas.2024.008

Abstract ( 238 ) PDF (0KB) ( 0 )

In this paper, we propose hierarchical attention dual network (DNet) for fine-grained image classification. The DNet can randomly select pairs of inputs from the dataset and compare the differences between them through hierarchical attention feature learning, which are used simultaneously to remove noise and retain salient features. In the loss function, it considers the losses of difference in paired images according to the intra-variance and inter-variance. In addition, we also collect the disaster scene dataset from remote sensing images and apply the proposed method to disaster scene classification, which contains complex scenes and multiple types of disasters. Compared to other methods, experimental results show that the DNet with hierarchical attention is robust to different datasets and performs better.

Image dense matching algorithm combining superpixel segmentation and guided filtering

ZHANG Zheng, ZHANG Wenyi, XU Shu

2025, 42 (6): 814-822. DOI: 10.7523/j.ucas.2023.081

Abstract ( 253 ) PDF (0KB) ( 0 )

In order to solve the problem that the existing local stereo matching method has low matching accuracy in the discontinuous region of parallax, a dense matching method combining superpixel segmentation and guided filtering is proposed in this paper. Firstly, a feature matching method is used to determine the disparity range, and the zero-mean normalized cross correlation is combined with gray-level and gradient information to construct the cost function. Secondly, the label map after superpixel segmentation is used to constrain the adaptive changes of the guided filtering window shape, and the cost is aggregated. Finally, the aggregation cost is used as the data item to construct the global energy function, and the disparity map is solved by graph cut algorithm, and multi-step disparity optimization is performed on the disparity map. Experimental results show that the average mismatching rate of the proposed method is 4.8% on the standard test image set provided by Middlebury website, which is significantly better than the traditional guided filtering dense matching method and semi-global matching method.

HRRP ship targets recognition based on double branches feature fusion convolutional neural network

ZHU Sijian, QI Xiangyang, FAN Huaitao

2025, 42 (6): 823-831. DOI: 10.7523/j.ucas.2024.012

Abstract ( 378 ) PDF (0KB) ( 0 )

To improve the accuracy of radar high resolution range profile ship target recognition, a ship target recognition method based on dual-branch feature fusion convolutional neural network model is proposed. Two branches are designed to extract features at different levels. The method designs a stacked convolutional detail branch with reduced downsampling to extract high resolution local features of ships. The global branch is composed of a modular structure used to extract low resolution global attitude features of ships. Based on the dimensional changes of the feature map after passing through two branches, the two features are changed in size separately in the feature fusion module, and the features are fused with each other to output recognition results. The experimental results show that the proposed method has faster convergence, fewer parameters, and higher accuracy compared to traditional recognition methods, verifying its effectiveness in HRRP ship classification.

Voiceprint recognition based on fused MGCC and CNN-SE-BiGRU features

FAN Tao, ZHAN Xu

2025, 42 (6): 832-842. DOI: 10.7523/j.ucas.2024.004

Abstract ( 377 ) PDF (0KB) ( 0 )

In order to solve the problems of single feature, weak representation ability and anti-noise ability in the field of voiceprint recognition, weak feature expression ability of traditional convolutional neural network (CNN) model and incomplete acquisition of temporal features, an acoustic feature fused with Mel frequency cepstral coefficient (MFCC) and Gamma frequency cepstral coefficient (GFCC) was proposed to carry out voiceprint recognition with a novel voiceprint recognition model based on enhanced CNN and bidirectional GRU networks (CNN-SE-BiGRU). Firstly, the extracted MFCC features and GFCC features are normalized, and according to the inter-class discrimination power, appropriate weights are designed to linearly combine the MFCC and GFCC features, and the Mel-gammatone cepstral coefficients (MGCC) with stronger speaker discrimination were obtained. Secondly, in order to improve the expression of CNN to features, an improved channel feature response SE-Block (squeeze and excitation block) model was introduced. Finally, building upon the spatial features extracted by the enhanced squeeze-and-excitation CNN (CNN-SE), the time series features are further extracted through the bidirectional gated recurrent unit network (BiGRU) to improve the performance of the whole network. Experimental results show that the acoustic features of MGCC show stronger characterization ability and better robustness under different noise backgrounds, while the average recognition rate of the CNN-SE-BiGRU model can be 96.05% under MGCC acoustic features, which fully proves the effectiveness and robustness of the proposed method.

Species identification of palm-leaf manuscripts based on Py-GC/MS

CHEN Qingle, YANG Yimin, HAN Bin, JIANG Hong

2025, 42 (6): 843-852. DOI: 10.7523/j.ucas.2024.078

Abstract ( 337 ) PDF (0KB) ( 0 )

The materials for making palm-leaf manuscripts mainly come from the leaves of two palms, Corypha umbraculifera and Borassus flabellifer. Applying a micro-destructive and rapid method to determine the plant species origin is conducive to in-depth cognition of the utilization of the palm-leaf manuscripts and provides a reference for their future conservation. In this study, a batch of standard samples was produced through simulation experiments to search for the criteria to distinguish the two palm species and validate them on the artifacts of the palm-leaf manuscripts of known plant species. Pyrolysis gas-chromatography/mass-spectrometry (Py-GC/MS) was utilized in combination with multivariate statistical analysis to differentiate the C. umbraculifera and B. flabellifer. The results revealed the presence of several triterpenoids in the samples, among which cycloeucalenol derivatives were found only in C. umbraculifera; two lupane-type triterpenoids and one cyclolaudenol and hopane-type triterpenoids were found only in B. flabellifer. The lupane-type triterpene and hopane-type compounds are stable in B. flabellifer. The relative contents of 10 lignin monomers were selected for principal component analysis (PCA), and the results showed a good differentiation between the C. umbraculifera group and the B. flabellifer group. The combination of the characteristic compounds and the PCA method successfully identified different types of plant sources of modern palm-leaf manuscripts, which provides a new idea for the rapid and trace identification study of the species and genera of ancient palm leaf manuscripts.

Current Issue