武钢7.63 m焦炉上升管荒煤气温度变化规律研究

为研究武钢7.63 m焦炉上升管余热利用工程实施的可行性,对2#炭化室上升管根部和桥管处荒煤气温度进行连续在线检测。分析发现:上升管根部荒煤气温度在整个结焦过程呈现类似Z字型周期变化规律,最高温度发生在推焦前烧石墨阶段,达1300℃~1400℃,整个结焦周期的平均温度约为771℃;桥管处荒煤气温度仍保持较完整的周期性规律;7.63 m焦炉单孔炭化室荒煤气产量为1156.42 m^(3)/h,一个结焦周期内单孔炭化室荒煤气可利用理论热量约13.76 GJ,工程理论吨焦产汽量(1.6 MPa低压饱和蒸汽)为108 kg。

煤粉处理方式对干燥煤成焦的焦炭指标的影响

试验比较了配合煤水分在2%的条件下,煤粉处理方式对配合煤堆密度、干馏焦炭质量的影响。结果发现,分离煤粉成型处理后,配合煤的装煤堆密度最高,结焦焦饼的宽向收缩最小,焦炭裂纹增加,40mm以上焦炭质量百分比大幅减少,所得干馏焦炭的耐磨强度改善明显;配合煤分离去除煤粉后,焦饼收缩最大,40mm以上焦炭质量百分比增加;配合煤干燥后直接装煤,焦炭反应后强度略高,其他各项指标处于居中水平。

采用跳层卷积神经网络的RGB-D图像显著性检测

RGB-D图像显著性检测旨在提取三维图像中的显著目标。为解决当前显著性检测算法难以检测出光线干扰场景内的目标和低对比度的目标等问题,提出了基于跳层卷积神经网络的RGB-D图像显著性检测方法。利用VGG网络分离出RGB图像和深度图像的浅层与深层特征,而后进行特征提取;以跳层结构为基础连接提取到的特征,实现融合深度、颜色、高级语义和细节信息的目标,同时生成侧输出;将侧输出进行融合,得到最佳的显著性检测图。实验结果表明,相比于深度监督显著性检测和渐进式互补感知融合显著性检测方法,在F值指标上分别提高了0.095 3和0.060 6,在平均绝对误差指标上降低了0.026 7和0.058 1。

Purification and reclamation of wastewater by an integrated biological pond system

The feasibility of an inexpensive wastewater treatment system is evaluated in this study.The experiment was conducted in 3 phases with different treatment combinations for testing theirpurification efficiencies. The pond system was divided into 3 functional regions: influentpurification, efficient upgrading and multi-utilization. Various kinds of aquatic organisms, wereeffectively cooperated in this system. The system attained high reduction of BOD_5, COD, TSS, TN,TP and other pollutants. The mutagenic effect and number of bacteria and virus significantly de-clined during the process of purifieation. After the wastewater flowed through the upgrading zone,the concentrations of pollutants and algae evidently decreased. Plant harvesting did not yield drama-tic effects on reductions of the main pollutants, through it did affect remarkably the biomassproductivity of the macrophytes. The wastewater was reclaimed for various purposes.