Search
Adv Search
  
20 August 2020
Volume 50 Issue 4
Machine Learning & Data Mining
An integrated learning approach for O3 mass concentration prediction model
Yan PENG,Tingting FENG,Jie WANG
2020, 50(4):  1-7.  doi:10.6040/j.issn.1672-3961.0.2019.423
Abstract ( 89 )   HTML( 32 )   ( 14 )   PDF (2691KB) ( 32 )   Save
Figures and Tables | References | Related Articles | Metrics

In order to accurately predict O3 mass concentration and development trend and to analyze inducing factors, an O3 mass concentration prediction model based on integrated learning was proposed. A multilayer FS-IL model for the O3 pollutant mass concentration was established in accordance with the data of O3 pollutant mass concentration and meteorological factors from 2015 to 2016 in Beijing, on the basis of missing value filling and outlier analysis, Pearson correlation analysis and Lasso regression analysis were used to select features of the cleaned meteorological data to eliminate data redundancy and improve prediction accuracy; an integrated learning algorithm based on self-organizing featuremap (SOFM)-Elman neural network (ENN) was proposed. After clustering sample data with SOFM to realize reasonable distribution of samples, ENN was used for simulation training to predict O3 mass concentration. The experimental results showed that the accuracy of ENN-based O3 pollutant mass concentration prediction was improved from 74.6% to 82.1% after the preliminary processing of data with Pearson-Lasso feature selection and SOFM sample clustering.

Visual sentiment analysis based on spatial attention mechanism and convolutional neural network
Guoyong CAI,Xinhao HE,Yangyang CHU
2020, 50(4):  8-13.  doi:10.6040/j.issn.1672-3961.0.2019.422
Abstract ( 69 )   HTML( 22 )   ( 1 )   PDF (1354KB) ( 22 )   Save
Figures and Tables | References | Related Articles | Metrics

Existing visual sentiment analysis based on deep learning mainly ignored the intensity differences of emotional presentation in different parts of the image. In order to solve this problem, the convolutional neural network based on spatial attention (SA-CNN) was proposed to improve the effect of visual sentiment analysis. The affective region detection neural network was designed to discover the local areas of sentiment induced in images. The spatial attention mechanism was used to assign attention weights to each location in the sentiment map, and the sentiment features of each region were extracted appropriately, which was helpful for sentiment classification by using local information. The discriminant visual features were formed by integrating local region features and global image features, and were used to train the neural network classifier of visual sentiment. Classification accuracy of the method achieved 82.56%, 80.23% and 79.17% on three real datasets Twitter Ⅰ, Twitter Ⅱ and Flickr, which proved that the method could improve the visual emotion classification effect by making good use of the difference of emotion expression in the local area of the image.

A scheduling algorithm based on multi-objective container cloud task
Xiaolan XIE,Qi WANG
2020, 50(4):  14-21.  doi:10.6040/j.issn.1672-3961.0.2018.210
Abstract ( 52 )   HTML( 15 )   ( 0 )   PDF (1365KB) ( 15 )   Save
Figures and Tables | References | Related Articles | Metrics

In order to solve the unrealistic, unfair, inefficient and unbalanced problems caused by container cloud scheduling model facing isomorphic tasks, isomorphic resources and single objectives, a tree scheduling objective model with constraint repair was proposed. Based on heterogeneous tasks and resources, constraint repair was adopted to avoid the impracticability of mapping scheme, and then priority to synthesize multiple sub-goals and attributed them to sub-spaces under different tree branches, and eventually achieved a fair, efficient, economical and balanced scheduling model among multiple upper application frameworks. The experimental results showed that the tree scheduling objective model with constrained repair was not inferior to other single-objective models in fairness, which could meet more tasks, and had higher resource utilization and load balancing under the preceding conditions. It was superior to the single-objective model in practicability, fairness, efficiency and balancing and ensured fair allocation of resources, which increased the benefits of container services, decreased the cost of physical resources, increased the stability and availability.

Depth segment classification algorithm based on convolutional neural network
ZHAO Ningning, TANG Xuesong, ZHAO Mingbo
2020, 50(4):  22-27.  doi:10.6040/j.issn.1672-3961.0.2019.416
Abstract ( 46 )   PDF (7208KB) ( 20 )   Save
References | Related Articles | Metrics
In order to solve the problem that redundant pixels in monocular images influenced depth information detection, a depth segment classification algorithm based on the convolutional neural network was proposed. We used NYU-Depth dataset to detect the segment-based features. Afterward, depth information was represented by line segments and its labels by the data preprocessing. The convolutional neural network was designed for considering the characteristics of the segments, and the classification of depth segments in monocular images was realized. By conducting several multi-group comparison experiments on different hyper-parameters, the accuracy of depth segment classification reached 73.50%. This experimental results proved the implement ability of the depth segment classification based on convolutional neural network, which was helpful to deep estimation using geometric features of images.
Image caption generation method based on class activation mapping and attention mechanism
LIAO Nanxing, ZHOU Shibin, ZHANG Guopeng, CHENG Deqiang
2020, 50(4):  28-34.  doi:10.6040/j.issn.1672-3961.0.2019.454
Abstract ( 47 )   PDF (7422KB) ( 14 )   Save
References | Related Articles | Metrics
Class activation mapping-attention mechanism was introduced to soft attention based image caption framework. The class activation mapping mechanism introduced the position information to convolutional features with richer semantic information, where there was a better alignment between convolutional features and description words, so that the generated description could describe the image content more completely. Improved the attention mechanism with double layer of long short-term memory network made the attention mechanism suitable for global and local information for generating words with specific features. The experiments showed that the improved model could generate more accurate description and outperformed the performance of models such as the soft attention mechanism in many evaluation criteria, specially the bleu-4 result on the MSCOCO dataset increased 16.8% compared with the soft attention-based model, which showed class activation mapping-attention could align the word and the convolutional feature, and generate more accurate descriptions with less key information lost.
Emotional EEG recognition based on Bi-LSTM
LIU Shuai, WANG Lei, DING Xutao
2020, 50(4):  35-39.  doi:10.6040/j.issn.1672-3961.0.2019.679
Abstract ( 56 )   PDF (1303KB) ( 10 )   Save
References | Related Articles | Metrics
To solved a challenging task of emotional electroencephalogram(EEG)recognition, this study proposed a bidirectional long short-term memory(Bi-LSTM)EEG classification model and explored the emotional mechanism of the brain, with the highest arousal accuracy of 76.78% and the highest valence accuracy of 77.28%. The Bi-LSTM model, compared with other models, had excellent performances in the recognition of emotional EEG. The Bi-LSTM model was used to explore the brain emotion mechanism by comparing the accuracy of different frequency bands, brain regions and feature density, and the results showed that the frequency bands, brain regions and feature density with the highest emotional correlation in the brain were respectively the α and β regions, Parietal Lobe and Frontal Lobe, 50 and 15.
Sliding mode synchronization of fractional-order Rucklidge systems with unknown parameters based on a new type of reaching law
WANG Chunyan, DI Jinhong, MAO Beixing
2020, 50(4):  40-45.  doi:10.6040/j.issn.1672-3961.0.2019.282
Abstract ( 36 )   PDF (903KB) ( 9 )   Save
References | Related Articles | Metrics
The problem of self-adaptive sliding mode synchronization of uncertainty fractional-order Rucklidge systems based on a new reaching law was studied with the fractional-order calculus theory. The sufficient conditions were concluded for drive-response systems to get sliding mode chaos synchronization by sliding mode approach. It was proved that drive-response systems were chaos synchronization under proper controllers and sliding mode function. Numerical simulations results verified the feasibility and effectiveness of the proposed method.
Three control schemes of chaos synchronization for fractional-order Brussel system
CHENG Chunrui
2020, 50(4):  46-51.  doi:10.6040/j.issn.1672-3961.0.2019.475
Abstract ( 42 )   PDF (1017KB) ( 11 )   Save
References | Related Articles | Metrics
Based on the fractional calculus theory, three synchronous control schemes were proposed to make the error system of the fractional Brussel system converge the error system states to the equilibrium point. An appropriate controller was designed in the first control scheme and the convergence of the error system was obtained by using Mittag-Leffler function. In the second control scheme, the fractional sliding mode surface was introduced, and the chaos synchronization of the fractional order Brussel master-slave systems was achieved based on the fractional version of the Lyapunov stability and the sliding mode control method. The effects of model uncertainties and external disturbances were fully taken into account in the third control scheme. A new sliding mode reaching law was designed and the fast convergence of the error system to the equilibrium point was obtained based on fractional order terminal sliding mode control. It was proved that master-slave systems were chaos synchronization under proper controllers. Numerical simulations were presented to illustrate the effectiveness and applicability of the proposed schemes and to validate the theoretical results of the paper.
Civil Engineering
Review on smart highways critical technology
Jianqing WU,Xiuguang SONG
2020, 50(4):  52-69.  doi:10.6040/j.issn.1672-3961.0.2020.149
Abstract ( 79 )   HTML( 214 )   ( 14 )   PDF (2579KB) ( 214 )   Save
Figures and Tables | References | Related Articles | Metrics

Giving highway engineering "wisdom" and establishing new generation five-in-one system of "Internet+" design, construction, management, monitoring and operation, namely, the smart highway, was the hot issue of the interdisciplinary study of civil engineering, control engineering, mechanical engineering, transportation engineering, and computer science. To comprehensively understand the smart highway, this review focused on the critical technology in the integrated system in full life-cycle of the smart highway as well as systematically investigated the relevant previous efforts, critical common technologies, and future scopes on multi-function pavement material, smart construction, smart detection, autonomous vehicles, connected vehicles, and internet of things technology.

Numerical simulation of rock fragmentation process by TBM cutter in double-joint rock mass
SHI Xuesong, GUAN Qingzheng, WANG Wenyang, XU Zhenhao, LIN Peng, WANG Xiaote, LIU Jie
2020, 50(4):  70-79.  doi:10.6040/j.issn.1672-3961.0.2019.744
Abstract ( 47 )   PDF (16856KB) ( 18 )   Save
References | Related Articles | Metrics
Effects of joint characteristics on rock fragmentation process by full-face rock tunnel boring machine(TBM)disc cutter was studied by numerical simulation using the particle discrete element method. A numerical model including double-joint rock mass and a disc cutter was established. 30 sets of numerical simulation tests with different joint spacing and joint angle were carried out to explore the failure modes of rock mass, study the variations of peak value of vertical contact force of cutter with joint spacing and joint angle, and analyze the characteristics of microscopic cracks growth during the rock fragmentation process. Results showed that four rock fragmentation modes were proposed as the joint characteristics changes; the peak value of vertical contact force of cutter first decreased and then increased with joint angle, and the minimum value was at 30° or 45° and the maximum value was at 90°; the peak value of vertical contact force of cutter generally increased with joint spacing; the amount of microscopic cracks was related to the vertical contact force of cutter and the relation was divided into three stages. These conclusions had certain guiding significance for the rational design of TBM cutter and TBM construction.
The KPI design method of performance assessment of hydraulic engineering construction enterprise based on entropy method
CHENG Sen
2020, 50(4):  80-84.  doi:10.6040/j.issn.1672-3961.0.2019.785
Abstract ( 42 )   PDF (547KB) ( 13 )   Save
References | Related Articles | Metrics
On the basis of “28 principle” and “SMART principle”, with the advantage that entropy could reflect the order degree of system information, the weight calculation method of entropy method was improved, and the weight calculation method of KPI index of entropy method was established. This method eliminated the influence of human subjective factors on index weight. Based on the statistical data of hydraulic engineering construction enterprise and the company's objectives, combined with the causal analysis method and the improved entropy method, the KPI index selection and weight calculation method suitable for hydraulic engineering construction enterprise was established. And this method was applied to the performance evaluation analysis of a hydraulic engineering construction enterprise.
Mechanical behavior of large-span and small spacing road tunnel with biased pressure
WANG Chunguo
2020, 50(4):  85-89.  doi:10.6040/j.issn.1672-3961.0.2019.307
Abstract ( 40 )   PDF (3123KB) ( 7 )   Save
References | Related Articles | Metrics
In order to analyze the mechanical behavior of large-span and small spacing tunnel with biased pressure, the finite difference numerical simulation software FLAC3D was used to analyze the stress-strain evolution law and to select the construction method by comparing in situ monitoring data with numerical simulation results. The results showed that due to the influence of the biased pressure, the settlement and stress values of the left-line tunnel were larger than those of the right-line tunnel. The principal stress of the tunnel crown and arch waist increased greatly after the excavation, which would increase the shear stress of the tunnel lining structure,at the same time, the principal stress and shear stress of the middle surrounding rock increased. So the stress concentration occurred near the arch waist of the two tunnels. Compared with the change of the axial force of the left-line tunnel, the excavation of the right-line tunnel had a greater influence on the bending moment of the left-line tunnel, especially on the bending moment of the arch waist of left-line tunnel near the side of the right-line tunnel. The results provided a scientific basis for the design and construction of the large-span small spacing road tunnel with biased pressure.
Bi-level planning of transmission network with solar-storage combination system based on learning theory
SUN Donglei, ZHAO Long, QIN Jingtao, HAN Xueshan, YANG Ming, WANG Mingqiang
2020, 50(4):  90-97.  doi:10.6040/j.issn.1672-3961.0.2019.341
Abstract ( 49 )   PDF (2249KB) ( 6 )   Save
References | Related Articles | Metrics
In order to address the solar power output uncertainty in transmission network planning, a bi-level planning model of transmission network was proposed in which the solar-storage combination system was modeled by learning theory. In the lower level, the scheduled power of solar-storage combination system submitted to the large power system was optimized by maximizing the long-term profit of the solar-storage combination system and minimizing the uncertainty of the planned power. Substituting the planned power of the solar-storage combination system obtained by the lower layer optimization into the upper transmission network planning model, then we minimized the transmission line investment cost, power system operating cost, and solar-shedding cost. The modified IEEE-118 bus system experimental results verified that the solar-storage combination system could reduce the uncertainty of the planned power and enhanced the credibility of planning result. The Q-learning controller established in this paper had good online learning ability and could effectively guide the planned output of the solar-storage combination system after learning a large amount of data.
Determination and control method on power operating area of modular multilevel converter considering the constraint of internal dynamics
ZHANG Feng, YANG Guixing, YUE Chenjing, HAO Quanrui, LI Dong
2020, 50(4):  98-107.  doi:10.6040/j.issn.1672-3961.0.2019.776
Abstract ( 49 )   PDF (3641KB) ( 6 )   Save
References | Related Articles | Metrics
This paper proposed a point-scanning method to obtain the internal dynamics of modular multilevel converter(MMC)and determine its operating area by solving the state-space equations. A steady-state model of modular multilevel converter based high voltage direct current(MMC-HVDC)system was developed, and the corresponding state variables of each power point were obtained by solving the state equations. A method to calculate the extreme value of non-sinusoidal periodic quantities was proposed to solve the amplitude of internal dynamics constraints. The power operating area of MMC was obtained when the internal dynamic constraints and the conventional constraints were considered by the point-scanning method. Based on the analysis results, a boundary control strategy was designed to ensure that the system satisfied the internal dynamic constraints. Simulation results in PSCAD/EMTDC verified the correctness of the MMC power operating area determination method and validity of the boundary control strategy.
Fabrication of Fe0/C composite and its application for acrylonitrile wastewater treatment
XIAO Fang, HUANG Deyi, YUE Qinyan, XU Xing, GAO Baoyu, WANG Wengang
2020, 50(4):  108-113.  doi:10.6040/j.issn.1672-3961.0.2019.175
Abstract ( 44 )   PDF (3221KB) ( 8 )   Save
References | Related Articles | Metrics
Un-sintered type Fe0/C composite was prepared in this study to be used as the micro-electrolysis fillers for acrylonitrile wastewater treatment. Effects of influent pH, hydraulic retention time(HRT)and aeration rate on chemical oxygen demand(CODCr)and acrylonitrile removal efficiency were investigated to determine the optimal conditions. The results indicated that the strength of Fe0/C composite met the water treatment fillers requirements. Acrylonitrile wastewater was treated by the un-sintered type Fe0/C composite at optimal operating conditions(influent pH of 3, HRT of 6 h and aeration rate of 0.2 L/min), and the removal efficiencies of CODCr and acrylonitrile were calculated to be 65.8% and 70.4%, respectively. During the successive running of micro-electrolysis reactor, the treating performance for acrylonitrile wastewater was stable without the harden of Fe0/C composite.
Water footprint of lead-acid battery throughout the whole life cycle
ZHANG Ruirui, MA Xiaotian, HONG Jinglan, CHANG Jingcai
2020, 50(4):  114-118.  doi:10.6040/j.issn.1672-3961.0.2019.670
Abstract ( 32 )   PDF (2551KB) ( 11 )   Save
References | Related Articles | Metrics
To quantify the water footprint impacts throughout the whole life cycle of lead-acid batteries from raw materials extraction to comprehensive recycling and reuse, a water footprint analysis based on life cycle assessment method was used to comprehensively analyze the water footprint generated by life cycle. SimaPro 8.4 software was used to analyze the water footprint impacts on lead-acid battery preparation, transportation, use, production of consumed electricity, and waste battery recycling processes. Results showed that lead-acid battery production and generation of consumed electricity were the two most important processes in water footprint analysis, among which the former had significant impact on aquatic eutrophication, carcinogens, freshwater ecotoxicity and water scarcity categories, while the latter had considerable contributions to acidification and non-carcinogens. In addition, key substances such as copper, chromium, phosphates discharged into water and sulfur dioxide to the atmosphere were mainly derived from these two processes. Optimizing these two main processes would help to reduce the water footprint of the lead-acid battery production chain effectively.
Aerosol optical properties on hazy days and clear days at rural and background sites in Shandong Province of China
ZHANG Wan, YANG Lingxiao, ZHANG Xiongfei, YAN Weida, WANG Xinfeng, WEN Liang,ZHAO Tong, WANG Wenxing
2020, 50(4):  119-126.  doi:10.6040/j.issn.1672-3961.0.2019.541
Abstract ( 33 )   PDF (4998KB) ( 9 )   Save
References | Related Articles | Metrics
To study the aerosol optical properties of rural and background areas in Shandong Province, aerosol was collected from June, 2014 to August, 2014 in Yucheng and Mt. Tai. We measured aerosol scattering coefficient and absorption coefficient by nephelometer and aethalometer. It was shown that the average of the single scattering albedo(ω)of haze days in Yucheng and Mt. Tai were both higher than that in clean days, which could be caused by the formation of sulphate and nitrate aerosols. The diurnal variation of scattering coefficient and absorption coefficient of Yucheng showed two peak patterns, which were mainly caused by morning and evening rush hour. However, the diurnal variation of scattering coefficient and absorption coefficient of Mt. Tai showed a unimodal pattern, which was attributed to the diurnal evolution of the atmospheric boundary layer. Indicated by the 48h backward trajectories, long-distance transmission from the southeast coastal areas via polluted cities such as Jinan, as well as local biomass burning emissions had an important influence on the formation of haze at Yucheng. However, during haze periods at Mt. Tai, the most frequently observed cluster moved slowly and could be considered a local source.
    [an error occurred while processing this directive]