Abstract:In recent years,deep learning has been a hot research topic in traffic flow prediction.Graph convolutional networks outperform traditional convolutional neural networks in spatial feature modeling,in view of their powerful capabilities in processing non-Euclidean data such as topological map,distance map and flow similarity map.Therefore,graph convolutional network and its variants have become a research hotspot in traffic flow prediction,and many attractive research results have been obtained.This article classifies and summarizes traffic flow prediction models based on graph convolutional networks in recent years.First,the graph convolution is elaborated by combining the definitions of spatial convolution and spectral convolution.Second,in view of the network structure of the prediction model,the graph convolutional network based traffic flow prediction models are divided into two major categories of combined type and improved type,each of which are analyzed and discussed in detail with representative model structures.In addition,typical datasets commonly used in traffic flow prediction for model performance comparison are reviewed,and a simulation test is conducted using one real dataset to demonstrate the prediction performance of four traffic flow prediction models based on graph convolutional networks.Finally,the future research hotspots and challenges in traffic flow prediction based on graph convolutional networks are prospected.

Abstract:To address the low prediction accuracy in agricultural commodity futures due to their nonlinear and non-smooth features resulting from various influencing factors,this paper proposes a decomposition and ensemble forecasting approach based on CEEMDAN and Transformer-Encoder-TCN.First,the Complementary Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) is used to decompose the time series into multiscale Intrinsic Mode Function (IMF) and residuals,reducing the complexity of series modeling.Second,each subseries is predicted via Temporal Convolutional Network (TCN) incorporating multi-stage self-attention unit (Transformer-Encoder),which optimizes the modeling weights of significant features.Finally,the prediction results of each subseries are linearly summed and integrated to obtain the final prediction results.The soybean futures revenue index in the agricultural commodity index of South China Futures Company is used as the research object.The model is retrained by time-series cross-validation and parameter transfer.The ablation and comparison experimental results show that the proposed model has superiority in RMSE,MAE and DS,verifying its effectiveness in predicting agricultural commodity futures.

Abstract:A prediction model based on ikPCA-FABAS-KELM is proposed to improve the short-term wind power prediction by traditional data-driven machine learning models.First,the principal component analysis is improved and the reversible kernel Principal Component Analysis (ikPCA) is proposed to reduce the complexity of input data while ensuring data features,with the purpose to advance the model in running speed.Second,the individual attraction strategies for Firefly Algorithm (FA) are used to improve the Beetle Antennae Search (BAS) thus a FABAS algorithm is proposed.Finally,the FABAS algorithm is used to optimize the regularization parameter C and kernel parameters γ of the Kernel Extreme Learning Machine (KELM),which can reduce the impact of manual parameter setting on blind model training thus improve model prediction accuracy.The simulation results show that the proposed model effectively improves the short-term wind power prediction accuracy.

Abstract:To address the problems of stock price prediction due to its non-static,highly complex and random fluctuations,a combination model based on Variational Mode Decomposition (VMD)-Circle Sparrow Search Algorithm (CSSA)-Long Short-Term Memory (LSTM) neural network is established.The original stock closing data is decomposed into several Intrinsic Mode Function (IMF) components by VMD,and then the CSSA is used to optimize the parameters of hidden layer neurons,iteration number and learning rate of LSTM,and the optimal parameters are fitted into the LSTM,where each IMF component is modeled and predicted,and the prediction results of IMF component are superimposed to obtain the final result.Experiments show that the RMSE,MAE and MAPE of the proposed model are minimized on multiple stock datasets,the error of the predictied closing prices of individual stocks fluctuates around 0,which is more stable with better fitting and higher accuracy.

Abstract:Illegally parked vehicles reduce road traffic efficiency,and cause traffic congestion even traffic accidents.Traditional vehicle detection methods are perplexed by a large number of parameters and low accuracy.Here,we propose a method using the improved YOLOv5 model and ray method to detect illegally parked vehicles.First,a lightweight feature extraction module is designed to reduce the amount of model parameters.Second,the attention mechanism is added to the model to enhance its feature extraction ability from both channel dimension and spatial dimension to ensure the model's accuracy.Then,the mixed data is used to enhance and enrich the dataset samples thus improve the detection performance in complex backgrounds,and EIoU is selected as the loss function to improve the model's positioning performance.Experiments show that the mean accuracy of the improved YOLOv5 model reaches 91.35%,which is 1.01 percentage points higher than that of the original YOLOv5s,and the number of parameters is reduced by 35.79%.Finally,the improved YOLOv5 model is combined with the ray method,which can reach real-time inspection speed of 28 frames per second on the embedded platform of Jetson Xavier NX.

Abstract:Although traditional vision-based SLAM (VSLAM) technologies have achieved impressive results,they are less satisfactory in challenging environments.Deep learning promotes the rapid development of computer vision and shows prominent advantages in image processing.It's a hot spot to combine deep learning with VSLAM,which is promising through the efforts of many researchers.Here,we introduce the combination of deep learning and traditional VSLAM algorithm,starting from the classical neural networks of deep learning.The achievements of Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN) in depth estimation,pose estimation and closed-loop detection are summarized.The advantages of neural network in semantic information extraction are elaborated,and the future development of VSLAM is also prospected.

Abstract:To address the low segmentation accuracy of multifidus muscle lesion sites in MRI images of patients with lumbar disc herniation,this paper proposes a new model to improve the U2-Net network with the goal that the encoding and decoding subnetworks are interconnected by a series of nested jump paths.To reduce the semantic missing of feature maps in the encoding and decoding subnetworks,the jump connections in the middle of RSU-7,RSU-6,RSU-5,and RSU-4 in the U2-Net model are redesigned,while the RSU-4F part remains unchanged.In addition,the channel attention module is added to enable the net to focus on channels of higher contribution to task,thus extract high quality multifractal muscle features.The experiments on the multifidus muscle MRI image dataset show that the improved U2-Net outperforms U-Net,U2-Net and U-Net++ network in indicators of Dice,HD and MIoU.It can be concluded that the proposed approach has good performance on MRI image segmentation of multifidus muscle,which can assist doctors to make diagnosis.

Abstract:Southeast Asia holds great importance as an essential passage of the Belt and Road Initiative and a key node region for China's foreign trade.Here,the spatial and temporal characteristics of economic development of Southeast Asian countries from 2012 to 2021 are analyzed using NPP／VIIRS nighttime lighting data and spatial analysis methods such as centroid,standard deviation ellipse and Moran index.The results indicate a significant correlation between GDP and nighttime light data,and a generally northwest bound transfer of economic centers of Southeast Asian countries.Most of the nightlight clusters have moved 476 kilometers to the northwest of Southeast Asia (that is,to the land border between Southeast Asia and China),suggesting a relationship between the movement of economic centers and the Belt and Road Initiative.The overall economic volume of Southeast Asia has increased,and economic development has been clustering in the region with evident directional characteristics.In terms of the spatial features,high-high aggregation and low-low aggregation are the two most significant spatial aggregation characteristics.The high-high agglomeration areas have played a good role and driven the economic development of the surrounding areas,the low-high agglomeration areas have greater development potential,while the low-low agglomeration areas in northern Southeast Asia have decreased significantly during 2012-2021.

Abstract:Scenic spots are most distinctive nature reserves in China,and the main carrier of tourism resources.Here,the spatial distribution characteristics and main influencing factors of scenic spots in China's cities above prefecture level are deeply explored via ArcGIS spatial analysis and SPSS multiple regression statistical analysis,using data of POI (Point of Interest) of national scenic spots as well as urban construction statistical yearbook.The results show a general pattern of strong southeast and weak northwest of China's scenic spots,with contiguous distribution stretching around major urban agglomerations,belt distribution of scenic spots above A level,and point distribution of national scenic spots.Three low-quality distribution belts along Shanxi-Hubei-Guangdong,Inner Mongolia-Qinghai-Xizang and the southeast coast have formed in the scenic spots;and two low-quality gathering belts of Inner Mongolia-Qinghai-Xizang and the southeast coast hold strong construction advantages and great potential for landscape quality improvement.Urban construction scale,urban population,government financial support,tertiary industry development and urban economic status are the main influencing factors of the spatial distribution of urban scenic spots.

Abstract:Traditional drought indices mainly consider a single factor and often cannot comprehensively reflect the drought condition.Based on data of MODIS and CLDAS (CMA Land Data Assimilation System),a daily scale integrated drought monitoring model was established by Gradient Boosting Machine (GBM) with multiple influencing factors and drought index as independent variables and comprehensive meteorological drought index (CI) as dependent variable.It was researched by taking drought in North China from 2015 to 2018 as a case.The results show that the model monitoring results are significantly correlated with the calculated CI values of the observation stations.The coefficient of determination is 0.945 and 0.655,and the Root Mean Square Error (RMSE) is 0.033 and 0.082 for training and test sets,respectively,indicating the high accuracy of the proposed integrated drought monitoring model.The consistency rate between the model monitored CI and calculated CI values is above 65%,and the correlation coefficient with Standard Precipitation Evapotranspiration Index (SPEI) and Relative Soil Moisture (RSM) is 0.68 and 0.6,respectively,showing its capacity to reflect both the meteorological drought and the agricultural drought.Monitoring of typical drought condition shows that the integrated drought monitoring model can accurately identify the drought occurrence,and represent the situation of comprehensive drought via considering various drought influencing factors.

Abstract:Phytoplankton is a major participant in the material and energy cycles of lake ecosystems,and the information of its community structure is of great significance in coping with and regulating lake ecosystems.In this study,the community characteristics of eukaryotic phytoplankton in the western Chaohu Lake in winter and summer were obtained via high-throughput sequencing.A total of 7 phyla and 71 genera of eukaryotic phytoplankton were detected during the survey,including 7 phyla and 59 genera in summer and 5 phyla and 27 genera in winter,dominated by Chlorophyta and Bacillariophyta,and the dominant genera varied greatly in winter compared with those in summer.The mean values of Shannon-Wiener index in summer and winter were 1.83 and 1.88 respectively,and the Pielou index were averaged 0.75 and 0.83 for summer and winter respectively.The results of water quality analysis indicated that TN and TP were relatively high in the western Chaohu Lake during the study period,and the physicochemical factors of the water body varied significantly between summer and winter (P≤0.05).Redundancy analysis showed that the eukaryotic phytoplankton community can be roughly explained by PO₄^--P,TN,TP and NH₄⁺-N,especially the PO₄^--P (P≤0.05).Mantel correlation analysis showed a close correlation between eukaryotic phytoplankton abundance and WT,DO,pH,NH₄⁺-N,TN and Chl.a.Variance partitioning analysis showed that seasonal factors explained most of the eukaryotic phytoplankton community dynamics.

Abstract:Though the reclaimed land is an important reserve land resource,it usually is poor in soil structure and low in organic matter and nutrient content.Organic fertilizer can quickly improve soil productivity,yet it will cause large emissions of greenhouse gases such as Nitrous Oxide (N₂O).It has been proved that the inoculation of Plant Growth-Promoting Rhizobacteria (PGPR) with N₂O reduction function not only reduces greenhouse gas emissions but also promotes crop growth.In this study,a PGPR denitrificans YSQ030 with N₂O reduction function was used as the test strain to clarify the effect of YSQ030 on N₂O emission and nitrogen cycling key functional genes in reclaimed soil with organic fertilizer application.Soil microcosmic experiments were set up for application of organic-inorganic compound fertilizer and sheep manure,then the soil N₂O emission fluxes after inoculation of YSQ030 were analyzed by gas chromatography.Meanwhile,soil chemical properties were analyzed at the end of the experiment,and the abundance of soil nitrification and denitrification functional genes (AOA amoA and AOB amoA;nirS,nirK,nosZ Ⅰ and nosZ Ⅱ ) were analyzed by real-time quantitative PRC.The results showed that YSQ030 significantly reduced the N₂O emission of reclaimed soil with organic-inorganic compound fertilizer or sheep manure,with the maximum reduction of N₂O emission reaching 91.5％ and 30.9％,respectively.The N₂O emissions of organic-inorganic compound fertilizer treatment were much higher than those of sheep manure treatment,which may be due to the low abundance of N₂O reductase genes of nosZ Ⅰ and nosZ Ⅱ in the former treatment.Furthermore,significant reduction of the abundance of nitrification and denitrification functional genes were observed only in organic-inorganic compound fertilizer treatments.This study shows that YSQ030 can reduce the N₂O emission in soil applied with organic fertilizer,which can provide a scientific basis for both soil fertility improvement and N₂O emission reduction,and also provide core strain resources for the research and development of new microbial fertilizers or bio-organic fertilizers.

Abstract:The emissions of methane (CH₄) and nitrous oxide (N₂O) from croplands in the Yangtze River Delta (YRD) in 2018 were estimated via CH4MOD model and emission factor method,and their temporal and spatial distributions were further analyzed to establish 1 km×1 km grid greenhouse gas emission inventory.The results show that,the CH₄ emission factor of paddy field was 348.54 kg／hm² and N₂O emission factor of croplands was 0.95 kg／hm² in the YRD,which were consistent with previous research results.In 2018,1.769 million tons of CH₄ (equivalent to 37.149 million tons of CO₂) were discharged from paddy fields in the YRD,with single-season paddy fields as the main contributor;while the croplands in the YRD emitted 15 114.9 tons of N₂O (equivalent to 4.504 million tons of CO₂) to the air,with wheat land as the biggest contributor.Jiangsu and Anhui provinces contributed most to the CH₄ and N₂O emissions,especially during April to August.It is suggested to reduce cropland nitrogen input thus mitigate agricultural greenhouse gas emissions.