基于Fisher判别法岩溶塌陷倾向性等级分类预测

为准确预测岩溶塌陷倾向性的等级分类,通过分析大量观测实例,选取岩性系数、岩体结构系数、地下水系数、覆盖层系数、地形地貌系数和环境条件系数作为模型判别因素。对12个实际观测样本进行训练,建立了基于Fisher判别分析法(FDA)的岩溶塌陷倾向性等级分类预测模型。借助SPSS软件工具,得到判别模型的4个判别函数。根据判别函数对训练样本进行回判,并对2个待判样本进行预测。结果显示:第一、第二判别函数的综合判别效率达到100%,大于规定的85%,满足工程实际应用需求;对训练样本进行回判时,误判率为零,同时对待判样本的分类预测准确率为100%。     

矿区采空塌陷危险性预测的RS-SVM模型

为快速、准确地预测矿区采空塌陷的危险性,针对矿区采空塌陷预测的复杂非线性特点,在统计分析实测资料的基础上,选取7项指标作为初始特征指标,30组塌陷样本作为原始样本,其中,前17组为原始训练样本,后13组为测试样本;运用粗糙集(RS)理论,对原始训练样本进行对象约简和属性约简。将属性约简后的3项指标作为支持向量机(SVM)的输入向量,建立矿区采空塌陷危险性预测的RS-SVM模型。将对象约简后的7组样本作为训练样本,用于模型训练。采用回代估计法进行回检,误判率为0。利用训练好的模型对13组待评样本进行预测,并与贝叶斯、BP神经网络(BPNN)方法进行比较。结果表明,RS理论与SVM算法相结合,能降低属性维数,去除冗余样本,简化模型,该模型所得预测结果准确率为100%。     

煤与瓦斯突出预测的支持向量机(SVM)模型

基于支持向量机(SVM)分类算法,考虑影响煤与瓦斯突出的主要因素,建立了煤与瓦斯突出预测的SVM模型。该模型选取开采深度、瓦斯压力、瓦斯放散初速度、煤的坚固性系数以及地质破坏程度5个指标作为模型输入量,同时将煤与瓦斯突出程度划分为无突出、小型突出、中型突出和大型突出4个等级,进而使其评判结果更为细化。以实测数据作为学习样本进行训练,建立相应判别函数对待判样本进行预测。通过算例分析,表明该模型的方法对煤与瓦斯突出预测的合理性与有效性,可以在实际工程中推广。     

基于KPCA-MPSO-ELM的矿井突水水源判别模型

为准确判别矿井突水水源并有效预防突水事故,提出一种基于核主成分分析-改进粒子群算法-极限学习机(KPCA-MPSO-ELM)的矿井突水水源判别模型。利用核主成分分析(KPCA)法对原始数据进行属性约减,通过改进粒子群算法(MPSO)优化极限学习机(ELM)的初始权值和阈值,建立KPCA-MPSO-ELM模型;在综合考虑矿井各含水层的水化学特征的基础上,选取Ca2+、Mg2+、K++Na+、HCO3-、SO42-、Cl-等的浓度和总硬度作为矿井突水水源的主要判别依据;以新庄孜矿的45组实测数据作为样本进行实例分析,其中33组数据作为训练数据训练模型,另外12组数据作为预测样本,用该模型进行预测,并将其判别结果与MPSO-ELM、KPCA-PSO-ELM模型的判别结果进行对比。结果表明:KPCA方法能减少指标数据间的信息重叠;通过MPSO优化ELM参数,可提高算法的整体搜索性能和收敛速度; KPCA-MPSO-ELM模型的预测精度高于MPSO-ELM、KPCA-PSOELM等2个模型。     

基于KPCA-LSSVM的矿井工作面动力环境安全评价模型

为实现对采前工作面所处动力环境的客观、准确评价,选取9个直接影响工作面 动力环境的指标因素构建安全评价指标体系,建立基于核主成分分析(KPCA)和最小二乘 支持向量机(LSSVM)的工作面动力环境多因素耦合安全评价智能模型。首先根据KPCA理 论对评价指标施行简约化处理,剔除冗余信息,得出6个简约后的评价指标并输入LSSVM 模型中训练学习,最后得到评价模型。选取从平顶山矿区和大同矿区搜集到的30组工作 面历史数据,按照20∶10的比例对模型进行训练和测试,并将测试结果与其他四种模型 结果进行了对比,结果表明:KPCA方法可有效减少数据信息冗余,利用KPCA优化的 LSSVM模型可准确评价工作面动力环境,误判率为0。     

煤层底板突水危险性的PNN预测模型研究及应用

为准确有效地预测煤层底板突水的危险性,在分析大量观测实例数据的基础上,选取底板含水层水压、煤层采高、隔水层厚度、断层落差、煤层倾角和断层距工作面距离等6项指标作为影响煤层底板突水的初始特征指标。针对指标之间具有相关性的问题,利用主成分分析(PCA)法提取6项特征指标的主成分,将其作为概率神经网络(PNN)的输入向量,建立基于PCA的煤层底板突水危险性的PNN预测模型。选取21组煤矿实测数据作为学习样本,用于训练模型。采用回代估计法对模型回检。利用学习好的模型,预测另外4组矿井突水数据样本。结果表明,该方法有效降低了指标数据相关性,实现了降维,使PNN模型工作复杂度减弱。将该模型应用于工程实例中,所得预测结果准确率为100%。     

深部矿山锚杆腐蚀失效风险评估研究

为解决地下矿山锚杆腐蚀失效问题,提出一种基于预测锚杆腐蚀失效速率的数学模型。收集矿山腐蚀失效锚杆相关数据,采用主成分分析法和梯度提升树方法,从原始数据中提取有用信息,并结合支持向量机(SVM)模型,预测复杂矿山环境中锚杆的腐蚀失效概率,利用具有连续特征和空间数据的数据集测试该模型,分析各环境因素对腐蚀失效影响权重,并与SVM、GTB-SVM这2种模型进行对比分析。研究结果表明：基于主成分分析的特征变换是可靠的风险预测方法,PCA-SVM模型在预测精度和结果的稳健性方面表现优越,训练集AUC值达到0.84、测试集达到0.83。该模型作为一个有用的在线工具,能够支持过程系统的安全和数字化。文中提出的主成分分析SVM模型,能够准确预测锚杆腐蚀失效的腐蚀概率。     

基于集对分析的岩溶塌陷危险性预测评价

为准确、有效地预测评价岩溶塌陷的危险性,运用集对分析原理建立岩溶塌陷危险性预测模型。选择岩性、岩溶发育程度、地形地貌、地质构造、土层厚度、土层岩性、地下水位距基岩面距离、地下水位变幅、地下水径流强度、地表水入渗、人工抽水强度和其他人类工程活动12个指标为岩溶塌陷主要影响因素,并将塌陷危险性划分为5个等级,用定量或定性的方法对指标进行赋值,构建其分级标准。运用集对分析方法,将影响因素实测值分别与5个危险性等级评价标准组成集对,计算其联系度,最后根据联系数的排序结果,判定岩溶塌陷危险性等级。利用该方法对贵州定扒地区岩溶塌陷危险性进行预测,结果表明,该模型评价结果与实际情况相符。     

不均衡数据下的采空区煤自燃PCA-AdaBoost预测模型

为提高不均衡数据下采空区自然发火预测准确率,选取O_2浓度等作为指标,利用主成分分析法(PCA)提取指标的主成分,并将主成分作为自适应增强算法(AdaBoost)输入参数,发火情况作为AdaBoost算法输出参数,建立不均衡数据下采空区自然发火PCA-AdaBoost预测模型;以张家口宣东2号矿为例,选取20组实测数据作为训练样本,用于训练模型;利用受试者工作特征曲线下的面积进行评价预测效果;利用训练好的模型预测15组测试样本,并将结果与粒子群优化支持向量机(PSO-SVM)模型进行比较。结果表明:在不均衡数据集条件下,利用PCA提取的算例的3个主成分包含原始6个指标的86.831%信息,降低了指标相关性,实现了降维;温度和CH_4浓度对发火影响更大;所建模型的预测结果与实际情况吻合,其在预测精度和收敛速度方面优于PSO-SVM模型。     

基于优化Bagging-LSSVM模型的冲击地压预测

为准确预测冲击地压危险性,提出一种优化Bagging算法动态集成的最小二乘支持向量机(LSSVM)的预测模型。在设计和优化Bagging-LSSVM模型流程的基础上,引入经典分类数据集,验证模型的可行性,并通过试验得出实现模型最优分类条件下的基分类模型数的最小值。综合考虑冲击地压的主要影响因素,确定其评判指标;以重庆砚石台煤矿的35组实测数据为试验样本,利用核主成分分析(KPCA)消除指标间的相关性,对比分析样本数据处理前后应用模型的预测效果;比较优化Bagging-LSSVM模型、优化Bagging-SVM模型和LSSVN模型预测冲击地压危险性的准确率。结果表明:经KPCA处理后的样本相较于原始样本,其应用于优化Bagging-LSSVM模型的预测准确率更高,耗时更少;且优化Bagging-LSSVM模型预测冲击地压危险性的准确率高于其他模型。     

基于遗传算法的支持向量机预测有机物自燃点的研究

根据定量构效关系(QSPR)原理,研究自燃点(AIT)与其分子结构间的内在定量关系。以265种有机化合物作为样本集,随机选择238种作为训练集,27种作为测试集,用遗传算法(GA)进行变量选择,分别建立多元线性回归(MLR)模型和支持向量机(SVM)模型研究有机物的自燃点与其分子结构间的关系。通过分析,发现造成模型预测效果不佳的原因是试验数据本身存在问题。通过对2个模型的比较,结果为GA-SVM模型明显优于GA-MLR模型,说明自燃点与其分子结构间具有很强的非线性关系。     

Influence of ATD versus PMHS reference sensor inputs on computational brain response in frontal impacts to advanced combat helmet (ACH)

ABSTRACT

Objective: This study analyzed the influence of reference sensor inputs from anthropomorphic test devices (ATDs) versus postmortem human subjects (PMHSs) on simulations of frontal blunt impacts to the advanced combat helmet (ACH).

Methods: A rigid-arm pendulum was used to generate frontal impacts to ACHs mounted on ATDs and PMHS. An appropriately sized ACH was selected according to standard fitting guidelines. The National Operating Committee on Standards for Athletic Equipment (NOCSAE) head was selected for ATD tests due to shape features that enabled a realistic helmet fit. A custom procedure was used to mount a reference sensor internally near the center of gravity (CG) of the PMHS. Reference sensor data from the head CG were used as inputs for the Simulated Injury Monitor (SIMon). Brain responses were assessed with the cumulative strain damage measure set at 10%, or CSDM(10).

Results: Compared to ATD tests, PMHS tests produced 18.7% higher peak linear accelerations and 5.2% higher peak angular velocities. Average times to peak for linear accelerations were relatively similar between ATDs (5.5?ms) and PMHSs (5.8?ms). However, times to peak for angular velocities were higher by a factor of up to 3.4 for PMHSs compared to ATDs. Values for were also higher by a factor of up to 13.1 when PMHS inputs were used for SIMon.

Conclusions: The preliminary findings of this work indicate that small differences in ATD versus PMHS head kinematics could lead to large differences in strain-derived brain injury metrics such as CSDM.     

引气剂和聚丙烯纤维对高水材料性质的影响研究

为了改变高水材料的破坏特点,采用引气剂和聚丙烯纤维双掺对其进行改性。试验结果表明:随着引气剂掺量的增加,浆体的流动性逐渐降低,混合浆液失流时间延长,试块密度和单轴抗压强度逐渐减小;聚丙烯纤维的掺入,对浆体流动性、失流时间、引气率影响均较小。聚丙烯纤维最佳掺量为2 kg/m3,引气剂和聚丙烯纤维的掺入使硬化体的弹性模量略有减小,且使试块由脆性破坏转变为延性破坏,在保持整体不散的情况下,提高其压缩量。SEM观察表明:钙矾石在气泡壁上集中生成,聚丙烯纤维与基体的界面处有利于针状钙矾石的生成,从而使聚丙烯纤维更好地发挥增强增韧的作用。     

Nonlinear molecular based modeling of the flash point for application in inherently safer design

New chemical process design strategies utilizing computer-aided molecular design (CAMD) can provide significant improvements in process safety by designing chemicals with required target properties and the substitution of safer chemicals. An important aspect of this methodology concerns the prediction of properties given the molecular structure. This study utilizes one such emerging method for prediction of a hazardous property, flash point (FP), which is in the center of attention in safety studies. Using such a reliable data set comprising 1651 organic and inorganic chemicals, from 79 diverse material classes, and robust dynamic binary particle swarm optimization for the feature selection step resulted in the most efficient molecular features of the FP investigations. Apart from the simple yet precise five-parameter multivariate model, the FP nonlinear behavior was thoroughly investigated by a novel hybrid of particle swarm optimization and support vector regression. Besides, 195 missing experimental FPs of the DIPPR data set are predicted via the presented procedure.     

Automation design in advanced control rooms of the modernized nuclear power plants

This study evaluated the effects of levels of automation (LOAs) decisions in advanced control rooms of the modernized nuclear power plants. Following advancements in design of digitalized human–system interfaces (HSIs), the roles of human operators have changed significantly. Negative performance and safety consequences may occur as a result of these changes. These problems are viewed as the out-of-the-loop (OOTL) performance problems. This study conducted an experiment to compare the effects of different LOAs under different operating procedures on operating performance. Experimental results indicated that blended decision-making (level 6 LOA) generates the lowest mental workload. Furthermore, the pattern of SA observed in this study is found better SA at intermediate LOA and poorer SA at low level of automation and full automation. Subjective rating results suggest that LOAs distribute the roles of option generation, and selection between human and/or computer servers which significantly impacts automated system performance. This study provides a direction for the HSI designers in nuclear power plants (NPPs). Additionally, based on results obtained by this study, the user interfaces of PCTRAN system and the alarm reset system should be improved to ensure safe operation of NPPs.     

A comparative study on the autoxidation of dimethyl ether (DME) comparison with diethyl ether (DEE) and diisopropyl ether (DIPE)

The Japanese government is planning to introduce DME as a substituted energy for oil and LNG. Introduction of DME could contribute greatly to both the prevention of global warming and the formation of resource-recycling societies. In these circumstances, a safety assessment of DME is very important when DME is used on a large scale. There is a possibility that prolonged exposure in air induces autoxidation to produce explosive organic peroxides during transportation and storage of DME. Therefore, the reactivity of DME with oxygen and the mechanism of the autoxidation were investigated. Accelerating Rate Calorimetry (ARC) was used to evaluate the thermal stability of DME and DIPE, a known peroxide producers, under adiabatic and various atmospheric conditions. In ARC studies of DME under oxygen, exothermic decompositions were detected although its self-heating rate was low in comparison with DIPE. Oven storage tests were carried out and iodimetry was used to measure the concentration of peroxides produced from DME in comparison with DIPE and DEE. However, no products could be found for DME either by GC/MS or by iodimetry, while some evidence of autoxidation of both DEE and DIPE were obtained from these experiments.     

单板层积梁弯曲破坏的试验研究与分析

在万能力学试验机上测试由高频热压方法制造的桦木和椴木单板层积梁的密度、静曲强度(MOR)、弹性模量(MOE)和水平剪切强度等力学性质;讨论LVL(单板层积材)梁受弯曲、水平剪切载荷时的主要破坏形式并得出相关结论。测试结果表明,在该试验条件下,桦木和椴木LVL梁具有较大的密度、MOR和较高的MOE。其主要物理力学性能受树种的影响较大,桦木LVL梁的主要性能均优于椴木LVL梁。当LVL梁受弯曲破坏时,其主要破坏形式为整体断裂破坏、开裂破坏以及混合破坏等3种;受水平剪切应力破坏时,以基材剪切破坏和胶层剪切破坏两种形式为主。     

基于案例推理的大型养路机械故障诊断系统

为实现铁路大型养路机械(简称"大机")故障的智能检索和诊断,使用基于案例推理(CBR)的思路分析和设计用以实现这些目的的系统。阐述案例的表示和构建方法。提出捣固车的故障案例模型构建方法。设计捣固车故障诊断的CBR系统。给出带权值的k-近邻法的案例相似性检索方法。提出大机故障诊断智能决策系统,采用定性和定量检索相结合的方法。故障诊断系统能够进行案例的定性、定量和混合检索。用捣固车的具体故障案例,验证所设计的系统的可行性。诊断系统检索得到的故障类型与依据现场采集的数据判定的故障类型基本一致,能够进行案例调用和修改。     

ZVI与TCE反应参数优化及动力学方程的建立

以批量研究的方法,考察了ZVI纯度、ZVI粒径、ZVI投量、p H值、温度和初始TCE浓度对TCE去除的影响,建立了ZVI去除TCE的动力学方程。最佳参数为:ZVI纯度92%,ZVI粒径30目,ZVI投量30 g,p H值为6.0,温度25℃,初始TCE质量浓度50 mg/L。最佳条件下TCE去除率可达73.6%,反应符合一级动力学方程。