Research On NOx Emission Optimization And Safety Risk Of Waste Incinerator Based On Deep Learning

With the growth of domestic waste removal and transportation in my country,incineration power generation has increasingly become the mainstream solution for the harmless waste treatment.However,the NOx generated by waste combustion has significant harm to the environment.Due to the large scale and complex operating conditions of waste incinerators,NOx emission is difficult to control.It is the key to solve the current problem to use a variety of parameters to build a prediction model of NOx emission concentration and optimize NOx emission As the core area of combustion reaction,incinerators are prone to failures,which endanger system safety and cause abnormal NOx emissions.In view of this,this paper used feature filtering,high-precision modeling,target optimization,risk assessment to carry out research on NOx emission optimization and safety risk of waste incinerators.The specific contents are as follows:(1)There are many operating parameters that can be collected for waste incinerators.When establishing a NOx prediction model,it is necessary to select parameters with a high degree of correlation with NOx emissions.On the basis of the mechanism analysis data,using the maximum information coefficient method,19 parameter variables,such as boiler load,primary air volume and secondary air volume,which are strongly correlated with NOx emission concentration were determined.(2)Aiming at the instability of the influence of the operating parameters of the waste incinerator on NOx emissions,which leads to the problem of decentralized feature mining of the multi-layer perceptron model,the adaptive adjustment method of the feature attention mechanism was introduced to establish the NOx of the waste incinerator based on AM-MLP.prediction model.the training of hyperparameters were adjusted to make the prediction effect to be optimal.The results show that the error between the predicted value and the measured value of the model in this paper is 4.18%,which is better than the MLP model,the SVR model and the RF model.(3)In order to reduce NOx emissions,by defining non-adjustable parameters,adjustable parameters,constraints,and objective functions,a mathematical model of NOx emission optimization was established.GA-PSO algorithm was used to solve the optimal adjustable parameters for reducing NOx emissions.According to the actual boiler operation situation of the enterprise,it is found that the control strategies such as primary air volume,secondary air volume,and ammonia injection parameters are different under different loads.Therefore,this study divided the boiler operating conditions into three operating conditions: low,medium and high for NOx emission optimization.The results show that the adjustable parameters obtained under the three loads can effectively reduce NOx emissions.In terms of air distribution parameters,under low load conditions,the secondary air distribution ratio is higher before optimization,and the opening of the OFA valve increases after optimization,and the secondary air distribution ratio increases after optimization.The air distribution ratio is reduced,and the generation of fuel-based NOx is suppressed.In the medium and high load conditions,the air distribution method has been adopted before optimization.The space for air distribution optimization to reduce NOx emissions is small.In terms of ammonia injection parameters,under low and medium loads,the opening of ammonia water valve increases after optimization.Under high load conditions,the opening degree of ammonia water valve decreases after optimization.(4)The unstable operation of the waste incinerator led to the risk of abnormal NOx emission.Fuzzy Petri nets was used to establish safety risk rules for waste incinerators.MYCIN algorithm was used for quantitative analysis.It is found that the failure of grate system,the failure of air supply system and the failure of furnace system are the three most reliable risks,for which countermeasures and suggestions are put forward.