基于LSTM算法的火电厂SCR烟气脱硝控制方法  

Control method of SCR flue gas denitration in thermal power plants based on LSTM algorithm

作　　者：王迎春 WANG Ying-chun

机构地区：[1]山西能源学院,山西晋中030600

出　　处：《节能》2024年第8期97-99,共3页Energy Conservation

摘　　要：由于火电厂SCR烟气脱硝过程中烟气的初始状态参数具有不稳定性,导致排除烟气NO_(x)浓度控制难度较大。为了准确控制烟气脱硝,有效降低烟气NO_(x)浓度,提出基于长短期记忆网络(LSTM)算法的火电厂选择性催化还原(SCR)烟气脱硝控制方法。以蜂窝状催化剂为基础,结合在SCR烟气脱硝过程中的作用模式,分别从孔道和内壁两个角度构建了火电厂SCR烟气脱硝模型;引入LSTM算法,以烟气中NO_(x)浓度控制要求为目标,将催化剂的投入速率作为控制参量,结合烟气的初始状态参数,实现SCR烟气脱硝过程的控制。在测试结果中,设计控制方法下脱硝反应器出口烟气NO_(x)浓度并未受到烟气流量的显著影响,始终稳定在125.0 mg/m^(3)以内,远低于临界值(150.0 mg/m^(3)),与对照组相比,具有更加可靠的脱硝效果。Due to the instability of the initial state parameters of flue gas in the SCR denitration process of thermal power plants,the control of flue gas NO_(x) concentration is challenging.Therefore,in order to accurately control the denitration of flue gas and effectively reduce the concentration of NO_(x) in flue gas,a control method for SCR flue gas denitration in thermal power plants based on the LSTM algorithm is proposed.Based on the honeycomb catalyst,combined with its action mode in the SCR flue gas denitration process,the model for SCR flue gas denitration in thermal power plants is constructed from two perspectives:Pore channels and wall interiors.The LSTM algorithm is introduced,with the control requirement of NO_(x) concentration in flue gas as the target,taking the input rate of the catalyst as the control parameter,and combining the initial state parameters of the flue gas to realize the control of the SCR flue gas denitration process.In the test results,the NO_(x) concentration of flue gas at the outlet of the denitration reactor under the designed control method is not significantly affected by the flue gas flow,and is always stable within 125.0 mg/m^(3),far below the critical value(150.0 mg/m^(3)).Compared with the control group,it has a more reliable denitration effect.

关 键 词：LSTM算法 火电厂 SCR烟气 脱硝控制 蜂窝状催化剂 SCR烟气脱硝模型 NOx浓度 投入速率 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]