基于SIFT与SVM的应急救援图像检测方法研究*

针对公共安全突发事件应急救援的封闭空间场景（Confined Space Scene,CSS）信息采集不通畅、不全面而影响救援等问题,提出1种基于不变特征转换（Scale invariant feature transform,SIFT）和支持向量机（Support Vector Machine,SVM）的图像信息异常检测方法。在场景内外信息传递“断环”情况下,该方法可利用物联网技术采集的图片,或以网络图片信息、历史类似场景数据等作为补充,通过SIFT特征提取、K means聚类处理以及SVM分类,实现场景的智能识别。经仿真分析,该方法能实现封闭空间内外部图像信息互通,“接补”因无法了解事件内部情况而产生的救援环节链条的“断环”,为救援提供决策参考。     

Streamflow Hydrology Estimate Using Machine Learning (SHEM)

Continuity and accuracy of near real‐time streamflow gauge (streamgage) data are critical for flood forecasting, assessing imminent risk, and implementing flood mitigation activities. Without these data, decision makers and first responders are limited in their ability to effectively allocate resources, implement evacuations to save lives, and reduce property losses. The Streamflow Hydrology Estimate using Machine Learning (SHEM) is a new predictive model for providing accurate and timely proxy streamflow data for inoperative streamgages. SHEM relies on machine learning (“training”) to process and interpret large volumes (“big data”) of historic complex hydrologic information. Continually updated with real‐time streamflow data, the model constructs a virtual dataset index of correlations and groups (clusters) of relationship correlations between selected streamgages in a watershed and under differing flow conditions. Using these datasets, SHEM interpolates estimated discharge and time data for any indexed streamgage that stops transmitting data. These estimates are continuously tested, scored, and revised using multiple regression analysis processes and methodologies. The SHEM model was tested in Idaho and Washington in four diverse watersheds, and the model's estimates were then compared to the actual recorded data for the same time period. Results from all watersheds revealed a high correlation, validating both the degree of accuracy and reliability of the model.     

应用支持向量机评价土壤环境质量

基于野外采样和室内分析相结合的方法,采用电感耦合等离子体质谱法(ICP-MS)对羊草沟煤矿研究区表层土壤样品中的Cd、Cr、Zn、Pb和Cu含量进行测定,应用非线性支持向量机模型中的分类支持向量机,选用sigmoid核函数,利用MATLAB编写程序,进行土壤环境质量评价,并利用模糊综合评判法对评价结果进行验证.在此基础上,运用对应分析方法对样品和变量进行了关联分析,进一步了解重金属污染特征.评价结果表明,研究区土壤环境质量多为Ⅰ类,与模糊综合评判法的相同率达到91.67%,将支持向量机用于土壤环境质量评价是可行的.相比于传统的评价方法,支持向量机采用结构风险最小化原则,将复杂的非线性问题转化为线性问题,成功的解决了多分类、高维运算等问题.     

基于机器学习和非参数估计的PM2.5风险评估

为开展区域风险评估,融合手机信令、气象和地理信息等多源数据,引入随机森林机器学习、非参数估计分位数图示法和非监督学习K-mean等方法,构建了区域PM_2.5风险评估及特征识别评价框架,在南京市区以0.3km分辨率网格为基础单元开展了案例研究.结果表明,该技术既可有效模拟PM_2.5浓度时空分布,十折交叉验证R²达到0.76,证明了准确度较高,并基于此识别出4种主要污染特征;也可有效捕捉短期人口流动导致的风险,在污染浓度不变的情况下短期人口流动会导致风险增加0.30~0.97倍.综合PM_2.5浓度和人口分布,识别出4种主要暴露风险模式,其中,研究区域6.5%的面积为高风险地区,23.0%的面积为低风险地区.“十四五”期间应加快现代科学技术在环境保护领域的应用,实施网格化和差异化的风险控制政策,维护人群健康.     

面向驾驶员辅助系统的换道意图辨识方法研究

为提高先进驾驶员辅助系统工作效能,通过差异性分析选取了人-车-路特征参数,建立了支持向量机换道意图辨识模型.基于驾驶模拟器采集的数据,运用ROC曲线对比分析了不同参数组合、时窗、数据表达形式下模型的分类效果,结果表明以方向盘转角、加速度、车辆距车道线距离和驾驶人头动信息为特征参数,时窗为1.5s,以时窗内信息均值和方差表述特征参量可获得最优辨识效果.研究表明,模型在保证1 s前瞻性的同时辨识率可达93％.     

电除尘在烧结机尾使用中的几个问题

对烧结机机尾电除尘器在使用过程中经常出现的几种故障进行分析,并提出相应的解决方法。     

Model-based direct search optimization of the broke recirculation system in a newsprint mill

This work is concerned with the dynamic modelling of the papermaking section of an integrated newsprint mill, and the subsequent application of two direct search methods in a proof-of-concept optimization study of broke recirculation strategies. The effect of the current broke recirculation policy on the mixed pulp properties at the paper machine headboxes was quantified using a dynamic simulation generating data for an objective function which reflected the rate of change in measurable (flow, consistency and temperature) and immeasurable (total dissolved solids and fibre length distribution) parameters. Genetic Algorithm and the Nelder–Mead simplex methods were then linked dynamically to the simulation and used to find improved ways of recirculating broke pulp. Specifically, the profile of the changes to the broke ratio was modelled using a quadratic function and a second-order transfer function, and the coefficients of these functions were optimized. The optimized broke recirculation resulted in a significant reduction in the value of the objective function, thus, suggesting that the fluctuations in the properties of the mixed pulp stream could be dampened and paper machine headbox stability improved. It is expected that this process enhancement strategy would contribute to an increase in production yield (via a reduction in the occurrence of paper machine breaks), and thus an overall reduction in the waste of materials and energy.     

基于支持向量机回归和小波变换的O3预报方法

使用南京工业区2016年6月1日~8月15日的臭氧（O₃）、O₃前体物及常规气象数据，结合多元线性回归（MLR）方法和小波变换（WT）改进支持向量机回归（SVR）对O₃小时浓度的预报精度.结果表明，通过WT方法将一个高变异性的序列转化为多个低变异性的序列后再处理可提高预报精度，M-WT-SVR预报的决定系数（R²）达到0.90，平均绝对误差（MAE）、平均绝对百分误差（MAPE）和均方根误差（RMSE）分别为3.86×10^-9、28.26%和5.57×10^-9，优于M-SVR和SVR.低层细节序列主要与NO、NO₂和芳香烃有关，而更高层的近似序列受到气象条件、前体物和O₃前期浓度共同影响.与经典的MLR方法相比，M-WT-SVR对O₃小时浓度的预报有明显优势.     

水泥机立窑窑尾烟气水雾喷淋除尘治理技术

机立窑窑尾烟气因含尘浓度高,温度、湿度变化大等原因,治理的难度较大,很多厂家均采用沉降室来治理窑尾烟气,但效果都不够理想.而铜山县第九水泥厂采用了水雾喷淋除尘技术治理窑尾烟气,从一段时间的运行状况看,取得了明显的成效.     

PREDICTION OF TWO-YEAR PEAK STREAM-DISCHARGES USING NEURAL NETWORKS1

ABSTRACT: The cascade correlation neural network was used to predict the two-year peak discharge (Q₂) for major regional river basins of the continental United States (US). Watersheds ranged in size by four orders of magnitude. Results of the neural network predictions ranged from correlations of 0.73 for 104 test data in the Souris-Red Rainy river basin to 0.95 for 141 test data in California. These results are improvements over previous multilinear regressions involving more variables that showed correlations ranging from 0.26 to 0.94. Results are presented for neural networks trained and tested on drainage area, average annual precipitation, and mean basin elevation. A neural network trained on regional scale data in the Texas Gulf was comparable to previous estimates of Q₂ by regression. Our research shows Q₂ was difficult to predict for the Souris-Red Rainy, Missouri, and Rio Grande river basins compared to the rest of the US, and acceptable predictions could be made using only mean basin elevation and drainage areas of watersheds.