Nondestructive testing of external defects on Nanguo pear

In view of the common pests and diseases and irregularly shaped fruits of Nanguo pear, this paper fused the spectral information and image features to realize the rapid nondestructive testing and recognition of the external defects on Nanguo pear by hyperspectral imaging technology. Backpropagation neural network and support vector machine model was established to identify external defects, which are commonly used in classification and pattern recognition. The testing results show that recognition effect of support vector machine is better than backpropagation neural network. Among them, the recognition accuracy of fruits damaged by insects and rotten fruits of Nanguo pears reaches 100%. This study provides a theoretical basis for developing online grading system and quality detection of Nanguo pear based on multispectral imaging technique.     

基于不均衡数据的小企业信用风险评价

小企业信用风险评价既是银行风险管理问题,又事关经济社会稳定。针对小企业贷款实践中,违约样本远少于非违约样本、且违约客户误判对银行影响较大的现实,采用不均衡支持向量机对小企业信用风险评价指标进行赋权,进而构建了能有效区分违约客户、非违约客户的评价模型。根据有无特定评价指标、特定评价指标数值变化对贷款小企业违约状态的影响程度赋权;反映了对违约状态影响越大、评价指标权重越大的赋权思路。将违约样本正确识别率、违约样本的准确率与查全率等因素作为支持向量机赋权模型中客户识别率的度量标准,改变了样本数据不均衡所导致的样本总体精度很高、违约样本精度反而不高的现象。研究结果表明:行业景气指数、资本固定化比率、净利润现金含量、恩格尔系数、营业利润率等评价指标对小企业信用风险的影响较大。     

荧光光谱结合支持向量机测定食用色素纯度

采用一种由原点矩法改造所得的特征压缩算法对荧光光谱数据进行预处理,将处理后的数据与加权最小二乘支持向量机(WLS-SVM)算法结合,建立鲁棒回归模型,用以预测实际食用色素粉末的纯度。以亮蓝和胭脂红这2种色素为例论述该方法对实际食用色素粉末纯度的预测效果。首先,利用FLS920荧光光谱仪测量获得两种色素的标准样本和实际样本在最佳激发波长下的荧光发射光谱数据,利用由原点矩法改造所得的特征压缩算法对获取的荧光光谱数据进行压缩和变换,一方面缩短了算法的运算时间,另一方面也提高了模型的预测精度。将预处理后的荧光光谱数据输入加权最小二乘支持向量机中建立浓度预测模型,该模型对亮蓝、胭脂红实际样本溶液给出的预测光谱与它们的实测光谱吻合程度好,半高峰宽区间内的平均决定系数分别为0.662和0.931。所有亮蓝、胭脂红溶液的预测浓度和标称浓度之间具有良好的线性关系,相关系数分别为0.997和0.992。由此通过多项式拟合得到的亮蓝、胭脂红粉末的预测纯度分别为61.0%和72.3%。     

一维热传导方程热源反问题基于最小二乘法的正则化方法

研究一维热传导方程热源反问题.给出基于最小二乘支持向量机（LS-SVM）求解的半解析表达式,此外还给出一种参数调节方法以及算法稳定性的证明.数值实验表明该方法具有较高的数值精度和稳定性.     

Pedestrian choice behavior analysis and simulation of vertical walking facilities in transfer station

Considering the interlayer height, luggage, the difference between queuing pedestrians, and walking speed, the pedestrian choice model of vertical walking facilities is established based on a support vector machine. This model is verified with the pedestrian flow data of Changchun light-rail transfer station and Beijing Xizhimen transfer station. Adding the pedestrian choice model of vertical walking facilities into the pedestrian simulation model which is based on cellular automata, the pedestrian choice behavior is simulated. In the simulation, the effects of the dynamic influence factors are analyzed. To reduce the conflicts between pedestrians in opposite directions, the layout of vertical walking facilities is improved. The simulations indicate that the improved layout of vertical walking facilities can improve the efficiency of pedestrians passing.     

通过载体提高燃料电池催化剂的稳定性

Pt/C是目前质子交换膜燃料电池使用最广泛的一种催化剂,制约其实际应用的关键技术之一是稳定性,增强燃料电池催化剂稳定性研究一直是该领域的热点. 本文结合作者课题组的研究工作,综述了通过改进载体材料来提高Pt基催化剂稳定性的一些方法,指出采用更稳定的载体材料以及发展与金属催化剂纳米颗粒有强相互作用的载体材料,是增强催化剂稳定性的努力方向.     

ITER 磁体支撑结构多维加载系统测控系统的设计

为了验证ITER 磁体支撑结构的可靠性,建立了多维加载测试平台对支撑部件进行机械性能检测。测量控制系统由7 台EDC 控制器、计算机系统、16 通道准动态应变仪、25 只位移传感器、7 套油缸中的负荷和位移传感器组成。基于可编程语言,利用EDC220 数字控制器实现了多通道协调加载的测量和控制过程,完成了 7 套双油缸高精度同步控制实验。针对典型的加载工况进行工程测试。结果表明,测控系统设计合理,控制系统可靠,能满足下一步的实验要求。     

应用近红外高光谱成像预测三文鱼肉的水分含量

应用近红外高光谱成像技术实现三文鱼肉水分含量的快速无损检测。采集来自不同部位的三文鱼肉共90个样本的高光谱图像,提取样本感兴趣区域(ROI)的平均光谱。随机取60个样本作为建模集,其余30个样本作为预测集。分别采用偏最小二乘回归(PLSR)和最小二乘支持向量机(LS-SVM)对全波段和水分含量建立相关性模型,并对预测集样本的水分含量进行预测。再用一种新的变量提取方法random frog选择特征波长,并基于特征波长分别建立水分检测的PLSR和LS-SVM模型。特征波长模型的预测精度虽然稍逊于全波段模型,但是仅用12个变量代替了全波段的151个变量,大大简化了模型,更便于实际应用。PLSR和LS-SVM特征波长模型的预测相关系数(R_p)分别为0.92和0.93,预测均方根误差(RMSEP)分别为1.31%和1.18%,取得了满意的结果。研究表明,近红外高光谱成像与化学计量学方法结合可以准确预测三文鱼肉的水分含量,为鱼肉品质的快速监测提供重要的参考。     

Extrapolation for Aeroengine Gas Path Faults with SVM Bases on Genetic Algorithm

Mining aeroengine operational data and developing fault diagnosis models for aeroengines are to avoid running aeroengines under undesired conditions. Because of the complexity of working environment and faults of aeroengines, it is unavoidable that the monitored parameters vary widely and possess larger noise levels. This paper reports the extrapolation of a diagnosis model for 20 gas path faults of a double-spool turbofan civil aeroengine. By applying support vector machine (SVM) algorithm together with genetic algorithm (GA), the fault diagnosis model is obtained from the training set that was based on the deviations of the monitored parameters superimposed with the noise level of 10%. The SVM model (C = 24.7034; γ = 179.835) was extrapolated for the samples whose noise levels were larger than 10%. The accuracies of extrapolation for samples with the noise levels of 20% and 30% are 97% and 94%, respectively. Compared with the models reported on the same faults, the extrapolation results of the GASVM model are accurate.