城市污水再生处理工艺中发光细菌毒性变化的初步研究

保障再生水的生态安全是污水再生利用过程中非常关键的问题.本文采用发光细菌毒性测试方法测定了2个城市污水再生利用示范工程工艺流程中生物毒性的变化,尤其是可能生成有毒副产物的氯消毒过程中的毒性变化.结果表明,二级生物处理能显著降低污水的发光细菌毒性,但氯消毒显著提高了污水的毒性,并且脱氯后污水的毒性仍保持较高的水平,对生态安全造成了潜在威胁.同时,通过比较2种污水水质发现,DOC(溶解性有机碳含量)和UV254(254 nm处的紫外吸收值)较高的水样,消毒后的毒性较大,认为DOC和UV254可望作为预测污水消毒生态风险的水质指标,为污水消毒实践中确定安全合理的污水水质条件提供一定的依据.     

基于SPA模型的煤矿瓦斯危险源风险评价

根据危险源理论和三类危险源的观点,采用集对分析(SPA)方法,进行了安全一事故集对的同异反分析,指出应采取措施控制不确定项,尤其要控制其中的第二类、第三类危险源才能使系统趋于安全状态;同时建立了煤矿瓦斯危险源风险评价模型,通过联系度的计算得出煤矿瓦斯的安全等级.研究表明,将集对分析方法应用于某煤矿瓦斯危险源风险评价,计算简单,结果较为精确、可靠,为不确定系统的评价提供了一条新思路.     

应用相平衡分配法建立湘江衡阳段沉积物重金属质量基准

采集湘江衡阳段29个站点的表层沉积物样品,测定沉积物中4种重金属(Cu、Pb、Zn和Cd)含量及赋存形态、以及孔隙水中重金属含量,根据相平衡分配法的基本理论,考虑参与沉积物-水相平衡分配的重金属组成,实测法计算重金属的沉积物-水相平衡分配系数(Kp),分别引用美国EPA制定保护水生生物不受重金属慢性毒性影响的基本连续浓度(CCC)和我国地表水环境质量标准(GB 3838-2002)Ⅰ类水质标准,建立两种湘江衡阳段沉积物重金属质量基准(SQC)进行对比分析,其中基于美国CCC建立的湘江衡阳段沉积物重金属质量基准与国内外研究成果相比可比性较好,4种重金属(Cu、Pb、Zn和Cd)的SQC值分别为64.62、55.57、1 360.40和2.34μg.g-1,此SQC具有保护长期生活于沉积物中的底栖生物不受重金属慢性毒性影响的意义.通过单因子评价法将湘江衡阳段沉积物中重金属总量与沉积物质量基准值(SQC)进行比较,结果表明,湘江衡阳段沉积物中Cd和Pb含量水平对底栖生物具有较大的慢性毒性影响,Cd污染不容忽视.     

民用燃煤源中多环芳烃排放因子实测及其影响因素研究

民用燃煤是我国目前关注较多的多环芳烃(polycyclic aromatic hydrocarbons,PAHs)排放源.根据我国民用燃煤所涉及的煤种、燃烧方式和炉灶类型,利用全流稀释采样系统对5种不同成熟度的煤种,分别以散块煤和蜂窝煤形式在3种炉灶类型中燃烧产生的烟气进行了采样和分析,初步获取了我国民用燃煤烟气中PAHs排放因子(EFPAHs)的实测数据.结果表明,烟煤的EFPAHs在块煤燃烧方式下的变化范围为1.1～3.9 mg.kg-1,在蜂窝煤燃烧方式下的变化范围为2.5～21.1 mg.kg-1,无烟煤的EFPAHs在块煤与蜂窝煤燃烧方式下分别为0.2 mg.kg-1与0.6 mg.kg-1.在民用燃煤的EFPAHs的各种影响因素中,煤的地质成熟度起主导作用,不同成熟度煤种的EFPAHs差异幅度高达1～2个数量级.其次是燃烧形态(块煤/蜂窝煤),主要表现为同一种地质成熟度的煤在蜂窝煤燃烧方式下排放的PAHs比块煤高2～6倍.炉灶类型的影响最小.     

氨对垃圾焚烧飞灰浸出特性的影响及地球化学模拟

垃圾焚烧飞灰可能因为选择性非催化还原法(SNCR)脱硝过程中氨泄漏、垃圾携带的渗滤液受热挥发等原因而吸附氨.本研究中采用人为添加氨水,在pH为3.66～12.44范围内,研究氨对飞灰中溶解性有机碳(DOC)和重金属浸出的影响,并利用地球化学模拟软件Visual MINTEQ从金属化学形态分布上分析氨对飞灰浸出的影响机制.结果表明,DOC在pH>9和有高浓度氨(≥1 357 mg·L-1)存在时,其浸出量大幅增加,而在浸出液中氨的水平不高于537 mg·L-1时则受氨的影响很小;在pH<6时,飞灰中各金属主要以自由态的金属离子和金属-氯络离子形态大量溶出,且受氨的影响较小;而在pH为8～12的碱性环境中和氨浓度较高时(≥3 253 mg·L-1),氨与金属生成了可溶性的金属-氨配合物,能显著增加Cd、Cu、Ni、Zn的浸出,且在pH=9附近时浸出量达到最大值,但氨对Al和Pb的浸出影响甚微;在pH>12时,Cd、Cu、Ni、Zn主要以羟基金属离子形式存在.在氨浓度为3 253 mg·L-1时,通过利用Visual MINTEQ模拟浸出值与试验数据的对比,发现Al、Pb、Zn的浸出主要由溶解/沉淀模型控制,而Cd、Cu、Ni由溶解/沉淀模型和表面吸附反应模型同时控制,且Visual MINTEQ模型能较好地预测飞灰中Al、Cu、Pb、Zn的浸出特性.     

发酵液作为EBPR碳源的动力学模拟

发酵液是一种优质的碳源,能够提高生物除磷系统(EBPR)的除磷效果.采用基于碳源代谢的修正ASM2模型,能够较好地模拟发酵液作为EBPR碳源的动力学变化规律.发酵液作为EBPR唯一碳源时,系统中的异养菌不仅不对聚磷菌(PAO)的生长构成竞争关系,反而促进PAO的生长.发酵液作为实际污水的补充碳源时,优化了污水中的碳源组成,创造了有利于聚磷菌生长的环境,使EBPR中聚磷菌达到微生物总量的40%以上,比实际污水作为碳源的EBPR中的PAO含量提高了3.3倍.     

Structural equation model of the influence of noise annoyance on miners’ unsafe behavior in coal mines

Miners' unsafe behavior (UB) is the main cause of coal mine accidents. Previous research has suggested that excessive noise in tunnels impacts miners' temperaments and safe production behavior. To explore the influence of noise on UB, four different dimensions of noise annoyance were identified: cognitive, emotional, communication, and physical mechanism annoyance. The coal mine noise annoyance scale, miners' unsafe behavior intention (UBI) scale, and miners’ UB scale were developed according to different dimensions. In the first survey, a total of 200 questionnaires were distributed in the two coal mines, and 193 valid questionnaires were recovered. Exploratory factor analysis was conducted, and invalid items were removed according to the analysis results to form the final scale. Then, 500 final scales were distributed in two Chinese coal mines, and 482 valid questionnaires were collected. Confirmatory factor analysis was conducted by collating the data of the second questionnaire. Based upon the structural equation model (SEM), the SEM of the influence of noise annoyance degree on UB and the path analysis with latent variable path analysis with the latent variables (PA-LVs) mediating model of the effect of UBI on UB were established, and the path coefficients with fitting data of the model were analyzed. The results indicated that the four dimensions of coal mine noise annoyance had a strong positive impact on UB, which was reduced by less noise annoyance. In addition, the Bootstrap method was used to verify the mediating method among the four dimensions of noise annoyance, cognitive annoyance, emotional annoyance, with communication annoyance indirectly affecting UB through UBI. The research results provide theoretical support for reducing the UB of miners and improving coal mine safety levels.     

Experimental and theoretical research on the inhibition performance of ethanol gasoline/air explosion by C6F12O

The safety issue of ethanol gasoline and the methods to control or weaken its explosion have attracted attention. To clarify the effect of C₆F₁₂O (perfluoro(2-methyl-3-pentanone)) on the explosion of ethanol gasoline-air mixtures and intrinsic mechanism, the explosion overpressure and flame propagation behavior under different equivalence ratios (φ = 0.6–0.8) and C₆F₁₂O concentrations (χ_inh = 0–4.0%) were experimentally obtained. The detailed inhibitor reaction process was also obtained by CHEMKIN based on a new assembly kinetic mechanism. The results show that the effects of C₆F₁₂O on the explosion characteristics of ethanol gasoline varied with χ_inh and φ. For rich flames, C₆F₁₂O is more effective than and heptafluoropropane (C₃HF₇) and nitrogen (N₂) in suppressing explosions; for lean and equivalence ratio flames, the addition of C₆F₁₂O may result in more severe explosions. The decrease in chemical reactivity is mainly because the mole fractions of OH and H radicals and the proportion of paths H radicals involved decrease after adding C₆F₁₂O, and R1500: CF₃COF + H = CF₃CO + HF, R965: CF₂:O + H = CF:O + HF, R863: CF₃ + H = CF₂ + HF are main suppressing reactions.     

基于BERT-BILSTM-CRF模型的电力行业事故文本智能分析*

为解决电力行业事故报告文本较长、语义复杂,难以进行有效文本识别问题,提出1种以BERT作为底层的预训练模型,并设计1种双重注意力机制编码器,结合BILSTM-CRF深度挖掘事故文本语义特征,从而实现文本智能分析。首先构建电力词典,通过对BERT预训练,进行BIO标注,然后引入BILSTM-CRF模型实现对文本标签智能分类,最后将该模型与现行其他4种深度学习模型进行对比。研究结果表明:该模型智能识别精确率、召回率及F1值(查准率)均达到约97%,较其他4种模型中效果最好的模型分别提高0.02,0.03,0.02。研究结果可为电力行业事故报告文本分析提供1种新思路。     

基于缓冲区重采样的LSTM林火预测模型*

为提高林火风险预测精度,挖掘地图上隐含的空间信息、时间序列上隐含的长期趋势和循环波动,提出1种基于缓冲区重采样的长短期记忆（LSTM）林火预测模型,选取15个与林火相关的影响因素,以方差膨胀因子为评价指标对其进行多重共线性检验,方差膨胀因子大于10的因素具有共线性,并采用信息增益率验证筛选结果的合理性。考虑到火灾的空间聚集特性,采用缓冲区分析与过采样相结合方法减少样本不均衡现象的影响,最终得到176 732条样本。对12个影响因素和研究时间段的火点建立LSTM预测模型,对森林火灾发生风险进行预测。研究结果表明:基于缓冲区重采样的LSTM林火预测模型有效考虑时空上隐含的信息,预测模型准确率为87.06%,特异性为97.99%,敏感度为76.12%,阳性预测率为97.43%,阴性预测率为80.41%,ROC曲线与AUC值均优于随机森林（RF）和支持向量机（SVM）这2种基准算法。维尔克松秩和检验发现,本文提出的模型与基准算法结果具有显著性差异。研究结果可为提高林火风险预测精度提供参考。