基于多源数据融合的煤矿工作面瓦斯浓度预测*

为了解决瓦斯浓度预测使用的单一数据在预测中影响还不够深入的问题,提出基于LSTM神经网络的多源数据融合瓦斯浓度预测模型。模型将上隅角瓦斯浓度、采煤机速度、工作面吨煤瓦斯涌出量等不同数据融合作为输入层参数,使用Adam优化算法更新LSTM网络层参数,利用Attention机制突出关键影响瓦斯浓度的因素,开展多源数据融合的瓦斯浓度预测,结合某矿1008工作面的实际数据,分析不同数据在瓦斯浓度预测中的作用。研究结果表明:单变量下的Attention-aLSTM预测效果相比LSTM提升14.2%;多源数据融合下的Attention-aLSTM相比自身提升了5%。     

常温下改进型EGSB处理城市污水的中试

通过中试实验,在常温下考察了回流比、水力停留时间和冲击负荷对改进型EGSB处理城市污水效果的影响。实验结果表明,随着回流比的提高,水力停留时间的缩短,改进型EGSB的出水CODfilt是下降的,但当水力停留时间低于2 h时,出水CODCr、CODfilt和SS明显增高。改进型EGSB通过改进回流水系统及布水系统,有效地降低了沉淀区水流上升速度,提高了布水效果,同未改进的EGSB相比,出水CODCr最大下降了26.7 mg/L,CODfilt最大下降了17.8 mg/L,而出水中SS最大下降了66.3 mg/L。当容积负荷突然升高时,对2套反应器都带来冲击,但改进型EGSB比EGSB要容易从冲击中恢复稳定。     

Effects of salinity on soil bacterial and archaeal community in estuarine wetlands and its implications for carbon sequestration: verification in the Yellow River Delta

Soils from two typical tidal salt marshes with varied salinity in the Yellow River Delta wetland were analysed to determine possible effects of salinity on soil carbon sequestration through changes in soil microbiology. The mean soil respiration (SR) of the salt water–fresh water mixing zone (MZ) was 2.89 times higher than that of the coastal zone (CZ) (4.73 and 1.63?μmol?m^?2?s^?1, respectively, p?Pseudomonas sp. and Limnobacter sp. that might have led to its higher dehydrogenase activity and respiratory rates. Additionally, the CZ possessed more Halobacteria and Thaumarchaeota with the ability to fix CO₂ than the MZ. Significantly lower soil salinity in MZ (4.25?g?kg^?1) was suitable for β-Proteobacteria, but detrimental for Halobacteria compared with CZ (7.09?g?kg^?1, p?相似文献   

活性污泥数学模型中异养菌产率系数的测定研究

利用活性污泥数学模型预测污水生物处理系统,首先必须确定其参数和水质组分。采用间歇活性污泥法和呼吸计量法对活性污泥数学模型中异养菌产率系数γ_H进行测定研究。结果表明,间歇活性污泥法测定结果重现性较差,试验控制条件比较严格;呼吸计量法测定结果准确性较高,重现性良好,因而,呼吸计量法是比较可行的γ_H测定方法。     

Synthesis and characterization of agricultural controllable humic acid superabsorbent

Humic acid superabsorbent polymer (P(AA/AM-HA)) and superabsorbent polymer (P(AA/AM)) were synthesized by aqueous solution polymerization method using acrylic acid (AA), acrylamide (AM) and humic acid (HA) as raw material. The effects of N,N'-methylenebisacrylamide (MBA) crosslinking agent, potassium peroxydisulfate (KPS) initiator, reaction temperature, HA content, ratio of AA to AM, concentration of monomer and neutralization of AA on water absorption were investigated. Absorption and desorption ratios of nitrogen fertilizer and phosphate fertilizer were also investigated by determination of absorption and desorption ratio of NH₄⁺, PO₄^3- on P(AA/AM-HA) and P(AA/AM). The P(AA/AM-HA) and P(AA/AM) were characterized by Fourier translation infrared spectroscopy, biological photomicroscope and scanning electron microscopy (SEM). The optimal conditions obtained were as follows: the weight ratio of MBA to AA and AM was 0.003; the weight ratio of KPS to AA and AM was 0.008; the weight ratio of HA to AA was 0.1; the mole ratio of AM to AA is 0.1; the mole ratio of NaOH to AA is 0.9; the reaction temperature was 60℃. P(AA/AMHA) synthesized under optimal conditions, has a good saline tolerance, its water absorbency in distilled water and 0.9 wt.% saline solution is 1180 g/g and 110 g/g, respectively. P(AA/AM-HA) achieves half saturation in 6.5 min. P(AA/AM-HA) is superior to P(AA/AM) on absorption of NH₄⁺, PO₄^3-. The SEM micrograph of P(AA/AM-HA) shows a fine alveolate structure. The biological optical microscope micrograph of P(AA/AM-HA) shows a network structure. Graft polymerization between P(AA/AM) and HA was demonstrated by infrared spectrum. The P(AA/AM-HA) superabsorbent has better absorbing ability of water and fertilizer, electrolytic tolerance and fewer cost than P(AA/AM) superabsorbent.     

硫酸盐还原过程中SRB的限制性生态因子的BP神经网络建模与仿真

利用BP神经网络对产酸脱硫反应器（专利设备，ZL：98240801.3）处理高浓度硫酸盐废水时硫酸盐还原菌（SRB）的限制性生态因子（COD/SO^2-4比、pH值、ALK、SO^2-4负荷率）进行建模与仿真，从而实现了多变量函数关系的映射与泛化，并从数学生态学角度定量表征了产酸脱硫微生物生态系统中SRB种群的代谢活性与实现生态位。     

汽车零配件涂装过程VOCs排放特征与案例分析

VOCs（挥发性有机物）现已被列为我国大气环境领域的核心污染物.随着汽车零配件制造行业减排要求的提出,于2018年6月选取典型汽车零配件制造企业,采用美国TO-15方法分析VOCs物种,采用FID（氢离子火焰检测器）对NMHC（非甲烷总烃）进行实测,分析汽车零配件涂装过程的VOCs排放特征.结果表明:①由于分析方式的不同,有组织排放的ρ（NMHC）比ρ（VOCs）高1.3~1.9倍,其中末端未安装VOCs处理设施的排气筒排放的ρ（NMHC）最高.②汽车零配件涂装过程排放的主要VOCs物种质量浓度占比范围分别为46.72%~98.33%（芳香烃）、1.20%~52.90%（含氧VOCs）,其中ρ（二甲苯）、ρ（苯系物）超标（DB 31/933—2015《大气污染物综合排放标准》）情况较为严重.③未进入VOCs处理装置前的VOCs物种组成与原辅料中VOCs物种组成一致,二者主要VOCs物种的质量分数大致相同,说明生产工艺的不同对VOCs的排放组成影响较小.④比较RTO（蓄热式热力燃烧装置）和活性炭吸附装置处理VOCs前、后废气组成的差异发现,活性炭吸附装置处理对VOCs排放的组成基本无影响,经RTO处理后排放物种以芳香烃和含氧VOCs为主,但是w（芳香烃）和w（含氧VOCs）变化不一致,说明RTO对芳香烃和含氧VOCs处理效率不同.研究显示,为满足国家对汽车零配件制造行业VOCs的减排要求,源头使用高固分涂料或水性涂料替代溶剂型涂料,优化过程收集系统,增强末端处理技术的净化效果、安全性和稳定性,是实现汽车零配件制造行业全过程减排的重要手段.      

活性污泥系统动力学模拟方法的综合分析

活性污泥法的应用现状和污水中氮磷排放标准的日益严格，使得传统数学模型已满足不了目前的要求，需要对活性污泥系统复杂的动力学规律进行有效模拟。文章在综合分析活性污泥动态模型国内外研究现状的基础上，介绍了3种占主流地位的模型：活性污泥数学模型、神经网络模型和混合模型。这3种模型在污水处理的设计、运行控制和工艺优化等方面各有其独到之处。     

SBR及其改良型技术分析

通过分析研究SBR及其衍生的各种改良型技术用于污水处理的运行方式及特点,指出SBR及其衍生的改良型技术可在中小型污水处理领域降低建设成本和运行费用,是一种适合中国国情的绿色小区污水回用处理技术.     

活性污泥数学模型中异养菌产率系数测定方法的研究

采用间歇活性污泥法和呼吸计量法测定活性污泥数学模型中异养菌产率系数.研究结果表明,间歇活性污泥法测定结果受试验控制条件特别是污泥有机负荷的影响非常大,且试验周期比较长;人工配水条件下,呼吸计量法测定异氧菌产率系数(YH)在0.71以上,比活性污泥数学模型推荐值高,其结果与底物性质有关,该方法准确性高,重现性良好.