【作者】 刘建军；

【导师】 方庆艳； 周怀春；

【作者基本信息】 华中科技大学 ， 热能工程， 2011， 硕士

【摘要】 在可预见未来的几十年内,煤炭仍将是我国主要的一次能源。以煤炭为主的一次能源格局,决定了在我国的电力工业中,燃煤火力发电占有主导地位。常规性的锅炉燃烧优化调整往往针对重要控制参数,给出特定工况点的优化策略,缺乏普遍适用性。基于模型预测和多目标寻优技术的燃烧优化系统可实现系统的闭环控制,达到经济与环保的多重优化目标,得到了广泛的应用。本文利用一套炉膛燃烧温度场检测系统开展了锅炉多工况燃烧试验检测,获得了沿炉膛高度断面温度分布、燃烧效率和NOx排放量等数据。基于多工况燃烧试验检测结果,加入炉膛断面温度信息,建立了燃烧效率与NOx排放量的BP神经网络预测模型。模型预测的燃烧效率和NOx排放量与试验值的相对误差分别小于1%和5%。基于已建立的预测模型,分别采用遗传算法和颗粒群算法进行了锅炉高燃烧效率和低NOx排放量的单目标以及多目标燃烧优化。多目标优化结果表明,采用加入炉膛温度信息的预测模型的优化结果更加符合实际情况;在134MW、154MW、172MW时,遗传算法优化后的锅炉燃烧效率分别上升了0.84%、1.37%和2.61%,NOx排放量分别下降了18%、6%、18%。最后采用数值模拟方法对优化过程和结果进行了验证。断面平均温度数值模拟结果与试验值误差小于8%,表明数值模拟能够比较准确地模拟锅炉燃烧过程。锅炉NOx排放和燃烧效率数值模拟值能与优化结果的变化趋势较好吻合,也说明优化结果是合理的。本文的研究表明:基于炉膛燃烧温度场可视化系统检测炉膛燃烧温度信息建立燃烧神经网络预测模型,能够更加准确地预测锅炉燃烧污染物排放过程;遗传算法的优化结果更接近实际情况,其结果能指导锅炉优化燃烧,也可为建立锅炉在线燃烧优化指导系统奠定基础。更多还原

【Abstract】 Coal will remain China’s main primary energy in the foreseeable few decades. Coal-fired thermal power dominates the country’s power industry under the coal-based energy structure. Normally, boiler combustion system optimization focuses on important control parameters, spotting the optimization strategy on a specific operating point, but lacks general adaptability. Model prediction and multi-objective optimization based combustion optimization system can achieve system closed-loop control, and the multiple optimization objectives of economic and environmental, thus it has been widely applied.In this paper, the furnace combustion temperature field detection system was used to detect combustion experiment under boiler’s multiple working conditions, and temperature distribution along the high degree cross-section of the furnace, combustion efficiency and NOx emission data were obtained. Adding the furnace cross-section temperature information to the multiple combustion detects results, the combustion efficiency and NOx emission prediction BP neural network model was established. The relative error between the combustion efficiency and NOx emissions predicted by the model and detect value was less than 1% and 5%. Based on the established prediction model, the genetic algorithm and particle swarm optimization algorithm were respectively used for the higher boiler combustion efficiency and lower NOx emission combustion optimization with single and multi-objective. Multi-objective optimization results show that adding information of the furnace temperature prediction model is more realistic; under 134MW, 154MW, 172MW, the optimized combustion efficiency rose by 0.84%, 1.37% and 2.62%, NOx emission decreased by 18%, 6%, 18% respectively. Finally, the optimization process and results were verified by numerical simulation method. The error of cross-section average temperature between simulation results and the detect were less than 8%, indicating that the numerical simulation can accurately simulate the combustion process. Boiler NOx emissions and combustion efficiency of the numerical simulation going in good agreement with the trend of the results of the optimization value also shows that the optimization results are reasonable. This study showed that the neural network predict model of combustion based on furnace combustion temperature field visual system, detecting the furnace combustion temperature information, can predict the combustion process more accurately, and guide the boiler combustion optimization, and also lay the foundation for establishing on line combustion optimization system.更多还原