基于模糊时序和支持向量机的高速公路SO2浓度预测算法

针对现有SO₂浓度预测方法中存在的污染物来源和影响因素认识不统一、小样本数据敏感、易于陷入局部最优等问题,文中提出了基于模糊时序和支持向量机的高速公路SO₂浓度预测算法,为搭建高速公路环境健康监测系统提供了可靠的理论支持.该方法依据SO₂浓度的季节变动规律,以季节作为时间序列,以24h为粒化窗宽,通过高斯核函数提取原始样本数据的特征值,输入支持向量机训练模型,并利用k重交叉验证法结合网格划分优化模型参数.文中应用该方法建立了SO₂浓度预测模型,并以2014年4月至2015年3月山西省太旧高速公路某监测点SO₂小时浓度监测值为样本数据,在MATLAB平台下应用LIBSVM工具实现了计算过程.结果表明,基于模糊时序和支持向量机的高速公路SO₂浓度预测算法不受机理性理论研究的限制,支持小样本学习,非线性拟合效果好,泛化能力强.     

基于最优定权组合法的大气污染物SO2预测

为了提高对大气污染物SO₂的预测准确率,基于多个空气质量预测模式（WRF-CHEM、CMAQ、CAMx）,以过去一段时间内各单项空气质量预测模式的组合预测误差平方和最小为原则,构建出针对大气污染物SO₂的最优定权组合预测模型.选取2018年云南省楚雄、昭通、蒙自三个站点1至5月份的实际观测数据和前述三个空气质量模式的预测数据作为实验样本,然后分别采用多元线性回归法和动态权重更新法在相同的实验条件下与所提的最优定权组合预测法进行预测对比实验.实验结果表明,所提方法的预测值相较其他两种方法更加贴近实际观测值,其两项误差评估指标值均最小.总体而言,最优定权组合预测模型很好地综合了各单项空气质量预测模式的优势,提高了SO₂的预测精度.     

基于SAW气敏阵列的火电厂排放物SO2与NO2监测装置设计

针对大气污染监测的迫切需求,设计了基于SAW气敏阵列的SO₂与NO₂监测装置,并采用人工神经网络算法进行气体浓度测量。对所构建的监测装置进行了实验验证和实际测试,结果表明,装置具有良好的检测性能和自动监测优势,可推广应用到各种生产行业的大气污染监测体系中。     

基于支持向量机的烟气二氧化硫排放量预测模型

针对循环流化床锅炉控制系统的烟气SO₂对象的非线性特点,本文建立了一种基于支持向量机的烟气SO₂排放量预测模型. 由于直接网格搜索确定支持向量机回归模型参数的方法计算量大、搜索时间长,本文采用单变量参数搜索结合网格寻优的方法来确定模型参数. 仿真结果表明,基于支持向量机方法建立的循环流化床锅炉烟气SO₂排放量预测模型具有良好的预测效果.     

混合H2/H∞鲁棒控制器设计

在状态空间描述下,定义了混合H₂/H_∞控制的完整信息、完整控制、干扰顺馈、输出估计这4种典型情况.在二次稳定意义上,讨论了混合H₂/H_∞的性能指标,及这4种典型情况的混合H₂/H_∞线性反馈控制器设计,给出了充分必要条件.在典型情况分析的基础上,研究一般意义上的混合H₂/H_∞反馈控制器设计.H₂和H_∞的干扰输入阵及性能评价函数各不相同时的混合H₂/H_∞反馈控制器,与H₂和H_∞控制器设计相似,归结为解两个Riccati方程.但这两个Riccati方程含有参数,最优解要通过搜索这两个参数得到.结果包含了单纯的H₂和H_∞设计,可看作是H₂,H_∞和混合H₂/H_∞的统一设计方法.最后通过一个简单的例子,说明了控制器设计方法的可行性.     

基于注意力机制和高斯概率估计的PM2.5浓度预测

作为衡量空气污染物浓度的重要指标, 对PM_2.5浓度进行监控预测, 能够有效地保护大气环境, 进一步地减少空气污染带来的危害. 随着空气质量自动监测站的大范围建立, 由传统的机器学习搭建的空气质量预测模型已经不能满足当今的需求. 本文提出了一种基于多头注意力机制和高斯概率估计的高斯-注意力预测模型, 并对沈阳市某监测站点的数据进行了训练和测试. 该模型考虑了PM_2.5浓度受到其他空气质量数据的影响, 将空气质量数据的分层时间戳(周、日、小时)的信息对齐作为输入, 使用多头注意力机制对于不同子空间的时间序列关联特征进行提取, 能够获得更加完善有效的特征信息, 再经过高斯似然估计得到预测结果. 通过与多种基准模型进行对比, 相较于性能较优的DeepAR, 高斯-注意力预测模型的MSE、MAE分别下降了21%、15%, 有效地提高了预测准确率, 能够较准确地预测出PM_2.5浓度.     

基于博弈论的H2/H∞混合控制及其在汽车主动悬架中的应用

文章提出把H₂/H_∞混合控制问题抽象为两个对局者信息不完全情况下的非零和博弈模型.在构造2×2非零和博弈模型中把反映系统鲁棒性能通道和动态性能通道作为参加博弈的两方,以H_∞和H₂控制方案作为两种纯策略,基于纳什谈判解原理设计出求解H₂/H_∞混合控制问题纳什均衡点的一般算法.把该算法应用于汽车主动悬架设计出基于纳什均衡点的H₂/H_∞输出反馈控制器.使用MATLAB进行仿真,仿真结果表明主动悬架系统在保持鲁棒稳定性与获得优化的动态性能指标之间取得平衡.     

MIMO离散时间系统混合l1/H2控制

研究了MIMO(多输入多输出)离散时间系统的混合l₁/H₂优化问题,该问题可描述为最优化一个传递函数矩阵的l₁范数同时保证另一个传递函数矩阵的H₂范数满足预定的指标.研究了最优目标函数值关于H₂范数指标的连续性.证明了MIMO系统混合l₁/H₂控制问题最优解的存在性.由于基于标定-Q(scaled-Q)方法求解MIMO混合l₁/H₂问题,避免了进行零点插值运算的困难.通过求解有限维非线性规划问题可得到最优目标值的收敛的上下界.     

基于多元线性回归的雾霾预测方法研究

提出了一种在线样本更新的多元线性回归分析的雾霾预测方法。首先搜集了北京市天气状况,包括平均气温、湿度、风级等气象数据以及PM2.5、CO、NO₂、SO₂等大气成分浓度数据,然后通过散点图对这些因素进行主要影响因素分析,筛选出对雾霾影响比较明显的因素作为雾霾预测的依据。通过在线样本更新的多元线性回归建立了PM2.5含量预测模型,并将气象要素作为雾霾的判断标准。最后给出实际例子,利用多元线性回归对北京未来一天、三天及一周的PM2.5含量进行较为精确的预测。     

混合H2/H∞参数辨识

研究了混合H₂/H_∞参数辨识问题.将混合H₂/H_∞估计方法应用到系统参数辨识中,给出了混合H₂/H_∞参数辨识算法.所得的算法不仅能满足规定的鲁棒性能,且为最小二乘(LS)参数估计误差判据提供了一个最优上界.结果表明:提高辨识的鲁棒性,需要牺牲辨识的精度作为代价.最后,仿真结果也验证了该方法的有效性.     

Vanadia doped tungsten-titania SCR catalysts as functional materials for exhaust gas sensor applications

Urea-SCR systems (selective catalytic reduction) are required to meet future NO_x emission standards of heavy-duty and light-duty vehicles. It is a key factor to control the SCR systems and to monitor the catalysts’ functionalities to achieve low emissions. The novel idea of this study is to apply commercially available SCR catalyst materials based on vanadia-doped tungsten-titania as gas sensing films for impedimetric thick-film exhaust gas sensor devices. The dependence of the impedance on the surrounding gas atmosphere, especially on the concentrations of NH₃ and NO₂, is investigated, as well as cross interferences from other components of the exhaust. The sensors provide a good NH₃ sensitivity at 500 °C. The sensor behavior is explained in light of the literature combining the fields of catalysts and semiconducting gas sensors.     

基于TDLAS的二氧化碳检测技术综述

精确检测CO_(2)气体浓度、控制CO_(2)气体排放是治理大气温室效应过程中最重要的部分。可调谐半导体激光吸收光谱(TDLAS)因具有高灵敏度和高可靠性的特点,广泛应用于在线监测、微量气体检测等方面。分析了TDLAS测量气体浓度的基本原理,重点介绍了直接吸收法和波长调制法并比较了两种方法的优缺点,随后介绍了近几十年来国内外应用TDLAS技术在气体检测方面取得的研究进展。最后总结了基于TDLAS的二氧化碳气体检测技术,并对其未来应用进行了展望。     

Enhanced long-term stability of SnO2-based CO gas sensors modified by sulfuric acid treatment

The gas-sensing properties and long-term transients of SnO₂-based CO gas sensors modified by sulfuric acid treatments have been investigated. The sulfuric acid treatments are carried out in two ways: kneading of SnO₂ powders with a sulfuric acid solution or dipping of the sintered sensor element in the same solution. Both sulfuric acid treatments enhance the sensitivity to CO and H₂, although the enhancement is more remarkable with the dipping treatment. The effects of these treatments on the selectivity to CO and some parameters relating to the reversibility and accuracy for CO detection are also evaluated. Furthermore, the long-term stabilities of the modified and unmodified sensor elements for over 550 days are investigated. The results indicate that the dipping treatment significantly improved the stability and reliability of the SnO₂-based CO gas sensor.     

Solid-state potentiometric SO2 sensor combining NASICON with V2O5-doped TiO2 electrode

A compact tubular sensor based on NASICON (sodium super ionic conductor) and V₂O₅-doped TiO₂ sensing electrode was designed for the detection of SO₂. In order to reduce the size of the sensor, a thick-film of NASICON was formed on the outer surface of a small Al₂O₃ tube; furthermore, a thin layer of V₂O₅-doped TiO₂ with nanometer size was attached on the NASICON as a sensing electrode. This paper investigated the influence of V₂O₅ doping and sintering temperature on the characteristics of the sensor. The sensor attached with 5 wt% V₂O₅-doped TiO₂ sintered at 600 °C exhibited excellent sensing properties to 1–50 ppm SO₂ in air at 200–400 °C. The EMF value of the sensor was almost proportional to the logarithm of SO₂ concentration and the sensitivity (slope) was −78 mV/decade at 300 °C. It was also seen that the sensor showed a good selectivity to SO₂ against NO, NO₂, CH₄, CO, NH₃ and CO₂. Moreover, the sensor had speedy response kinetics to SO₂ too, the 90% response time to 50 ppm SO₂ was 10 s, and the recovery time was 35 s. On the basis of XPS analysis for the SO₂-adsorbed sensing electrode, a sensing mechanism involving the mixed potential at the sensing electrode was proposed.     

Gas sensitivity of Zn doped α-Fe2O3(SO4, Sn, Zn) to carbon monoxide

It is shown that the doping of Zn and Sn can improve the gas sensitivity of α-Fe₂O₃-based sensing material to CO. X-ray photo-electron spectroscopy analysis suggests that this is mainly due to the fact that the simultaneous doping of Zn and Sn can increase the S and hence SO₄²⁻ contents in the α-Fe₂O₃(SO₄²⁻, Sn, Zn) sensing material. The results also suggest that under a given condition, the gas sensitivity of α-Fe₂O₃(SO₄²⁻, Sn, Zn) to CO can be optimised by properly adjusting the doped Zn content.     

Diesel engine dynamometer testing of impedancemetric NOx sensors

Prototype solid-state electrochemical sensors using a dense gold sensing electrode, porous yttria-stabilized zirconia (YSZ) electrolyte, and a platinum counter electrode (Au/YSZ/Pt) were evaluated for measuring NO_x (NO and NO₂) in diesel exhaust. Both electrodes were exposed to the test gas (i.e., there was no reference gas for the counter electrode). An impedancemetric method was used for NO_x measurements, where the phase angle was used as the response signal. A portion of the tailpipe exhaust from the dynamometer test stand was extracted and fed into a furnace containing the experimental sensor. The prototype sensor was tested along with a commercially available NO_x sensor. Simultaneous measurements for NO_x, O₂, CO₂, H₂O, CO, and CH₄ in a separate feed stream were made using Fourier transform infrared (FTIR) spectroscopy and an oxygen paramagnetic analyzer. The experimental sensor showed very good measurement capability for NO in the range of 25-250 ppm, with a response paralleling that of the FTIR and commercial sensor. The prototype sensor showed better sensitivity to NO_x at the lower concentration ranges. O₂ is an interferent for the experimental sensor, resulting in decreased sensitivity for measurement of NO_x. Methods to overcome this interference are discussed.     

Analysis of data for the carbon dioxide capture domain

To tackle the global concern for adverse impact of greenhouse gas (GHG) emissions, the post combustion carbon dioxide (CO₂) capture technology is commonly adopted for reducing industrial CO₂ emissions, for example, from power generation plants. The research on post combustion CO₂ capture has been ongoing in the last two decade, and its primary objective is to improve efficiency of the CO₂ capture process while reducing specific operating problems such as solvent degradation and corrosion. This objective requires a good understanding of the intricate relationships among parameters involved in the CO₂ capture process. From a review of the relevant literature, we observed that the most significant parameters influencing the CO₂ production rate include: heat duty, circulation rate of the solvent, CO₂ lean loading, and solvent concentration. To study the nature of relationships among the key parameters, we conducted data modeling and analysis based on the amine-based post combustion CO₂ capture process at the International Test Centre for Carbon Dioxide Capture (ITC) located in Regina, Saskatchewan of Canada. In our study, the experimental data collected from ITC from year 2003 to 2006 were analyzed using the combined approach of neural network modeling and sensitivity analysis. The neural network was trained for modeling the relationships among parameters, and the sensitivity analysis method illustrated the order of significance among the parameters. The modeling results were validated by the process experts. This paper describes the procedure of our work, and discusses the results of the analysis.