基于神经网络的二元混合液体自燃温度预测

自燃温度(Auto-Ignition Temperature, AIT)是防火防爆安全设计的关键临界参数之一。为解决目前多数采用试验方法测量混合物AIT费时费力且有一定危险性的问题,运用定量结构-性质关系方法,使用反向传播神经网络(Back Propagation Neural Network, BPNN)和一维卷积神经网络(one-Dimensional Convolutional Neural Network, 1DCNN)技术建立二元混合液体AIT预测模型。以二元混合液体的分子描述符为输入、试验测得的AIT为输出,经多种方法对模型的拟合性、稳定性和预测能力评价验证。结果表明,BPNN模型和1DCNN模型均有良好的预测能力,其均方根误差分别为4.780℃和9.603℃,拟合度与5折交叉验证拟合度差值分别为0.058和0.040,表明BPNN模型有更好的拟合能力,1DCNN模型有良好的稳定性。     

基于改进随机森林模型的海底管道腐蚀预测

为提高海底管道腐蚀速率预测精度,建立一种基于改进随机森林的海底管道腐蚀预测模型。首先,采用斯皮尔曼相关系数,分析实海挂片腐蚀数据的相关性,并采用因子分析降维;然后,设定K值为5的K折交叉验证,建立随机森林回归(RFR)模型,并输入降维后的数据;最后,输出模型的交叉验证得分,并对比评价该模型与其他模型的最高得分预测结果。研究结果表明:改进RFR的五折交叉验证输出平均得分值为0.912,高于BP神经网络(BPNN)模型、支持向量回归(SVR)模型2种对比模型;五折交叉验证最高得分预测结果均方根误差(RMSE)、平均绝对误差(MAE)分别为1.441和1.3,均优于对比模型相应值。     

应用电性拓扑状态指数预测烷烃自燃点

建立了一个基于人工神经网络的定量结构-性质相关性模型，用于52种烷烃化合物自燃点的预测研究。应用原子类型电性拓扑状态指数作为表征分子结构特征的描述符。该指数既能表征分子的电子特性，又反映其拓扑特征，同时易于计算，并有较强的同分异构体区分能力。采用误差反向传播（BP）神经网络方法对烷烃自燃点与电性拓扑状态指数间可能存在的非线性关系进行拟合。将52种烷烃样本随机划分为训练集（30种）、验证集（8种）和测试集（14种），并通过“试差法”确定网络的最优参数。运用最佳网络结构[64—1]对实验样本进行模拟，结果表明，多数样本的自燃点预测值与实验值符合良好，对于测试集，平均预测绝对误差为8．4℃，均方根误差为11．8，优于多元线性回归方法和传统基团贡献法所得结果。该方法的提出为工程上提供了一种根据分子结构预测有机物白燃点的有效方法。     

二元互溶可燃混合液体闪点预测模型研究

基于定量结构-性质相关性(QSPR)原理,开展二元互溶可燃混合液体闪点与其结构信息间的内在定量关系(M-QSPR)研究。以332个不同组成和配比的二元互溶可燃混合液体闪点试验数据作为研究样本,根据体系中各纯组分的结构信息,计算相应的混合物描述符(SiRMS,Simplex Representation of Molecular Structure),应用遗传-多元线性回归(GA-MLR)算法从中优化筛选出一组与该体系闪点最密切相关的原子碎片参数作为输入参数,分别采用多元线性回归(MLR)和支持向量机(SVM)算法建立理论预测模型,并将其与文献已有模型进行比较。结果表明,MLR和SVM对测试集样本的平均绝对误差(AAE)分别为4.142 K、1.551 K,均方根误差(RMSE)分别为4.911 K、2.220 K,决定系数(R~2)分别为0.818、0.962。研究表明,影响二元混合液体闪点的主要SiRMS结构因素是■和■,并且随体系中■、■、■四原子碎片增多,闪点升高;随体系中■、■、■四原子碎片增多,闪点降低。同时,与内部交互作用和分子间非加和作用相比,分子间的可加和作用对该体系闪点的影响更为显著;与原子的局部电荷相比,原子类型对该体系闪点的影响更为显著。     

冬季高速公路逐时路表温度LSTM预测模型

为提高冬季路表温度的预测精度,提出一种基于多维长短时记忆(LSTM)神经网络的冬季路表温度逐时预测模型,以小时路表温度为模型输出,综合考虑多维气象因素的累积影响和路表温度的周期性,采用滑动窗口构造输入特征矩阵;构建路表温度LSTM逐时预测模型,通过深度学习高效逼近具有复杂非线性和不确定性的路表温度,并以江苏省宁宿徐高速公路、云南省麻昭高速公路为实例进行验证。结果表明：与随机森林(RF)模型和BP神经网络相比,LSTM路表温度逐时预测模型的准确率得到显著提高,在宁宿徐高速、麻昭高速的平均绝对误差(MAE)、均方误差(MSE)和均方根误差(RMSE)分别为0.303、0.295、0.543和0.581、0.694、0.833,预测值与观测值绝对误差位于[0, 1)℃之间的占比为93%和89%。LSTM模型能准确捕捉路表温度的周期性和不确定性,在阴雨天和晴朗天的预测值与实测值基本一致,模型鲁棒性较好。     

基于MATLAB工具箱的开采煤层自燃危险性预测

正确预测开采煤层自燃发火的趋势与危险性,对煤矿安全生产具有重要的指导意义。煤层自燃发火的趋势和危险程度与其影响因素之间存在着复杂的非线性关系,而人工神经网络具有极强的非线性逼近能力,能真实刻画出输入变量与输出变量之间的非线性关系。为准确预测开采煤层自燃发火的危险性,笔者针对反向BP神经网络收敛差的缺点,分别采用基于MATLAB神经网络工具箱中的VLBP和LMBP算法的改进BP神经网络模型对开采煤层自燃的危险性进行了预测。根据开采煤层自燃的特点,选取煤本身自燃倾向性、煤层地质及赋存条件、通风技术条件3个关键影响因素作为开采煤层自燃危险性的评判指标,建立了开采煤层自燃危险性预测的神经网络模型。实际应用效果表明,采用基于MATLAB神经网络工具箱的BP网络模型,能克服一般BP网络收敛较慢的缺点,能加快收敛速度;运用LMBP算法比VLBP算法快,但需较大计算机内存;该模型收敛速度快,准确性高,是一种十分有效的开采煤层自燃危险性预测方法。     

基于人工神经网络的二元液体闪点预测

建立了一个基于人工神经网络的理论模型,用于预测二元混合液体的闪点.根据所研究混合液体的物理性质,选择了相关黏度、表面张力等物理参数来表征闪点,以这些参数作为输入参数,二元混合液体的闪点作为输出值,应用反向传播(BP)人工神经网络方法对两者之间的内在定量关系进行模拟.结果表明,闪点预测值与实验值符合良好,优于传统的计算方...     

高速铁路列车晚点时间实时预测的神经网络模型

准确地预测高速列车晚点时间对提高高速铁路实时调度指挥水平及运输服务质量有重要意义。以武汉-广州高速铁路(HSR)列车运行实绩数据为基础,建立基于循环神经网络(RNN)的列车晚点预测模型。该模型中,按照列车实际运行顺序输入RNN以利用其反馈机制学习到相邻列车间相互作用关系。基于平均绝对误差(MAE)以及平均绝对百分误差(MAPE)评估模型的预测能力。结果表明:提出的深度学习模型预测精度明显高于人工神经网络、支持向量回归及马尔科夫等已有列车晚点时间预测模型。     

基于IPSO-LSTM的新能源汽车锂电池健康状态监测

为监测新能源汽车锂电池的健康状态(SOH),防范电池故障引发安全事故风险,提出改进粒子群算法(IPSO)和长短期记忆(LSTM)神经网络相结合的模型,监测锂电池的SOH。首先,采用Spearman相关性分析法,提取锂电池SOH监测的健康因子;其次,采用线性惯性权重和非对称学习因子改进传统粒子群算法(PSO),利用IPSO算法对LSTM模型的隐含层神经元个数、神经元失活率、批处理值进行关键参数寻优,进一步优化LSTM模型,建立IPSO-LSTM锂电池SOH监测模型;最后,以新能源汽车主流采用的18650锂电池数据集验证IPSO-LSTM模型,并对比分析BP、LSTM和PSO-LSTM这3种模型。结果表明：IPSO-LSTM模型的平均绝对误差(MAE)在0.02以内、均方根误差(RMSE)在0.03以内,监测误差在15%以内,相较于BP、LSTM、PSO-LSTM模型,IPSO-LSTM模型的误差指标值均最小,模型具有更高的精度和稳定性。     

基于循环神经网络的缓坡场地液化侧移预测*

为预测缓坡场地地震液化侧向位移,基于改进自适应算法（Rectified Adam）和循环神经网络模型（RNN）,提出液化侧移预测模型RA-RNN,通过对侧移数据进行样本学习,并利用改进自适应算法优化循环神经网络结构,验证RA-RNN模型可靠性,并与多元线性回归法（MLR）计算结果进行对比。结果表明:RA-RNN模型计算得到侧移一般为实测位移的0.7~1.3倍,训练结果R2,RMSE,MAE分别为0.977,0.375,0.141;土耳其科喀艾里RA-RNN模型预测结果RMSE和MAE为MLR模型的1/26,1/830;中国台湾集集镇RA-RNN模型预测结果RMSE和MAE为MLR模型的1/18,1/350,RA-RNN模型预测结果较优,预测精度及泛化能力得到很大提升。     

Flash points measurements and prediction for binary miscible mixtures

Flash point is one of the most important parameters used to characterize the potential fire and explosion hazards for flammable liquids. In this study, flash points of twenty eight binary miscible mixtures comprised eighteen flammable pure components with different compositions were measured by using the closed cup apparatus. The obtained experimental data are further employed to develop simple and accurate models for predicting the flash points of binary miscible mixtures. Based on the vapor–liquid equilibrium theory, the normal boiling point, the standard enthalpy of vaporization, the average number of carbon atoms, and the stoichiometric concentration of the gas phase were selected as the dominant physicochemical parameters that were relevant to the overall flash point property of liquids. With these parameters for pure components as well as the compositions of mixtures, the new form of characteristic physicochemical parameters for mixtures were developed and used as the input parameters for the flash point prediction of mixtures. Both the modeling methods of multiple linear regression (MLR) and multiple nonlinear regression (MNR) were employed to model the possible quantitative relationships between the parameters for mixtures and the flash points of binary miscible mixtures. The resulted models showed satisfactory prediction ability, with the average absolute error for the external test set being 2.506 K for the MLR model and 2.537 K for the MNR model, respectively, both of which were within the range of the experimental error of FP measurements. Model validation was also performed to check the stability and predictivity of the presented models, and the results showed that both models were valid and predictive. The models were further compared to other previously published models. The results indicated the superiority of the presented models and revealed which can be effectively used to predict the FP of binary miscible mixtures, requiring only some common physicochemical parameters for the pure components other than any experimental flash point or flammability limit data as well as the use of the Le Chatelier law. This study can provide a simple, yet accurate way for engineering to predict the flash points of binary miscible mixtures as applied in the assessment of fire and explosion hazards and the development of inherently safer designs for chemical processes.     

A prediction model for the flash point of binary liquid mixtures

A prediction model based on the partial least squares of the multivariate statistical analysis methods was developed for the flash point (FP) of binary liquid mixtures. Estimation of the FP of flammable substances is important for safety measures in industrial processes. Since experimental FP data of liquid mixtures are scarce in the literature, there have been many researches to estimate the FP of liquid mixtures using physicochemical laws. In this study, the partial least squares (PLS) method using experimental data was used as a prediction model of the FP of binary liquid mixtures. The FPs predicted from the PLS method were also compared to results from the existing calculating methods using physicochemical laws such as Raoult's law and the Van Laar equation.     

二元混合液体闪点的计算

以混合溶液纯组分易燃液体闪点的饱和蒸气压为基础,应用乌拉尔定律、双液系的气-液相平衡理论,运用Le Chatelier方程和安托因方程导出二元混合液的闪点计算方法。并例举易燃液体与易燃液体组成的理想混合液、易燃液体与易燃液体组成的非理想混合液、易燃液体与不燃液体组成的非理想混合液的计算过程。乙醇溶液闪点的计算结果与现有的文献资料比较,误差在允许范围内。计算数据用Excel处理,快捷准确,用于确定二元混合液体的火灾危险性。     

Flammability hazard analysis of imidazolium-based ionic liquid binary mixtures under high temperatures

Ionic liquid (IL) mixtures are promising because they can optimize the involved properties according to industrial needs. It has already been demonstrated that IL flammability is due mainly to IL decomposition generating flammable substances. Four different ILs, 1-Butylimidazolium tetrafluoroborate ([BIM][BF₄]), 1-butylimidazolium nitrate ([BIM][NO₃]), 1-butyl-3-methylimidazolium tetrafluoroborate([BMIM][BF₄]), and 1-butyl-3-methylimidazolium nitrate ([BMIM][NO₃]), were selected as the parent salts to form the different imidazolium-based IL binary mixtures. These mixtures were tested via isothermal thermogravimetric analyzer (TGA) at different temperatures (120, 150, 180, 210, and 240 °C), then tested by the flash point analyzer after isothermal heating pretreatment at the above temperatures. Results show that the mixtures' flash point values decrease with the heating temperature increase. Vaporization of the IL mixtures’ decomposition products results in a higher concentration of flammable gases and a flash point decrease, which lead to the flammability hazard increasing. Moreover, results show that the flash points of the studied binary imidazolium IL mixtures are more similar to those of the more unstable IL in their parent ILs. Also, the flammability hazard of IL binary mixtures may obviously increase under the high temperature environment for a long time.     

A new method for the prediction of flash points for ternary miscible mixtures

The flash point is one of the most important physicochemical parameters used to characterize the fire and explosion hazard for flammable liquids. The flash points of ternary miscible mixtures with different components and compositions were measured in this study. Four model input parameters, being normal boiling point, the standard enthalpy of vaporization, the average number of carbon atoms and the stoichiometric concentration of the gas phase for mixtures, were employed and calculated based on the theory of vapor–liquid equilibrium. Both multiple linear regression (MLR) and multiple nonlinear regression (MNR) methods were applied to develop prediction models for the flash points of ternary miscible mixtures. The developed predictive models were validated using data measured experimentally as well as taking data on flash points of ternairy mixtures from the literature. Results showed that the obtained average absolute error of both the MLR and the MNR model for all the datasets were within the range of experimental error of flash point measurements. It is shown that the presented models can be effectively used to predict the flash points of ternary mixtures with only some common physicochemical parameters.     

基于长短时记忆网络的水质预测模型研究

在当前水质数据急剧增加的背景下,为了挖掘水质时间序列中的更多信息,提升水质预测的精度,构建了基于缺失值填补算法和长短时记忆网络(LSTM)相结合的水质预测模型。通过缺失值填补算法进行水质数据的缺失值处理,利用LSTM网络分别构建不同水质参数的预测模型,以太湖水质监测数据为样本,对模型进行检验。结果表明,基于缺失值填补算法-LSTM的水质预测模型适应性强,相较传统SVM、BP神经网络、RNN、LSTM模型预测精度更高,对水环境保护具有重要意义。     

Theoretical and experimental study of the inhibition and inert effect of HFC125, HFC227ea and HFC13I1 on the flammability of HFC32

HFC32 is a potential alternative refrigerant with excellent thermal performance, but the flammability is a main obstacle for its applications. The group contribution method is utilized to analyze the inhibition efficiency of nonflammable refrigerants in binary mixtures. Furthermore, a novel equation of predicting the minimum inerting concentration of nonflammable refrigerants has been proposed by analyzing the variation of the flame propagation velocity and the flammable refrigerant concentration. Experimental studies of the explosion limits of HFC125/HFC32, HFC227ea/HFC32 and HFC13I1/HFC32 were carried out and the ranges of explosion limits were obtained. At the same time, the relationship between the maximum charge of the flammable refrigerants and lower flammability limit (LFL) was analyzed. The result demonstrates that the proposed novel theoretical equation can effectively predict the minimum inerting concentration of nonflammable refrigerants to flammable refrigerants, and the theoretical results have significance on the security application of the binary mixtures.     

基于GASA-BP神经网络的煤层瓦斯含量预测方法研究*

为提高煤层瓦斯含量预测的精准度和效率,提出1种利用遗传算法（GA）和模拟退火算法(SA)混合初始化BP神经网络（BPNN）的瓦斯含量预测新模型（GASA-BPNN模型）。利用灰色关联分析法（GRA）筛选瓦斯含量主控因素并作为GASA-BPNN预测模型的输入。为解决BPNN收敛速度慢和易陷入局部极小陷阱的问题,将GA和具有时变概率突跳性的SA整合为GASA算法协同初始化BPNN的权值和阈值,有效地提高BPNN的参数学习能力。将该模型应用于煤炭生产现场,结果表明:BPNN模型、GA-BPNN模型和GASA-BPNN模型瓦斯含量预测总平均相对误差分别为15.79%,9.03%,5.56%。相比BPNN模型和GA-BPNN模型,GASA-BPNN模型对样本的泛化能力更强,参数训练速度最快并且预测精准度最高。     

基于绝热火焰温度混合气体爆炸下限的预测

准确地预测可燃混合气体的爆炸极限,对防止工业生产中时有发生的混合气体爆炸事故有着重大的意义。通过采用Gaseq软件计算CH4,C3H8,C2H4,C3H6,CH3OCH3和CO的绝热火焰温度(CAFT),分析初始温度对甲烷和丙烷混合气体(体积比1∶1)爆炸下限(LEL)的影响。结果表明:随着初始温度的升高,临界火焰温度基本不变,而LEL线性下降。使用计算绝热火焰温度法对不同比例的二元混合气体(体积比1∶1,3∶1,1∶3)以及三元混合气体(体积比1∶1∶1)的LEL进行预测,在选取的35组不同组份的混合气体中,LEL的预测值与文献值的平均绝对误差为0.081 8,平均相对误差为0.02。