锅炉烟气出口氮氧化物浓度智慧控制数据

通过采集锅炉总给煤量、一次风总量、二次风总量、锅炉氧含量以及出口氮氧化物浓度，可以精准控制氨水泵给定频率，通过不断的智能学习以及模型优化，使脱硝系统达到最优控制，在确保生产稳定运行、排放达标的同时实现精准控制，最终达到降低生产成本的目的。算法包括数据输入模块与长短期记忆神经网络算法预测模块2个部分。数据输入模块包含：锅炉一次风总量G1、锅炉二次风总量G2、锅炉总给煤量T、锅炉含氧量C等多种锅炉燃烧相关数据。长短期记忆神经网络算法可以用表达为数学公式：Q=L(G1,G2,T,C)，其中氨水流量Q为根据脱硝出口氮氧化物浓度与输入数据通过算法预测的下一时刻所需氨水流量；L(X1,X2,X3,...,Xn)为长短期记忆神经网络的神经元模型，该算法由多个长短期记忆神经网络的神经元模型串联组成。G1,G2,T,C为神经元模型的输入数据，输入数据从输入门按历史数据时间先后顺序依次输入到第一个神经元模型，经过遗忘门遗忘部分信息后通过神经元模型细胞状态的激活条件判断后形成按时间先后顺序的输出从输出门输出，经过算法预先设定好的神经元模型的数量之后，最终输出为下一时刻锅炉所需氨水流量。

By collecting the total coal feed rate of the boiler, total primary air flow, total secondary air flow, boiler oxygen content, and outlet nitrogen oxide concentration, the given frequency of the ammonia water pump can be precisely controlled. Through continuous intelligent learning and model optimization, the denitration system can achieve optimal control, realizing precise control while ensuring stable production operation and compliance with emission standards, ultimately reducing production costs. The algorithm consists of two parts: a data input module and a long short-term memory (LSTM) neural network prediction module. The data input module includes various boiler combustion-related data such as total primary air flow of the boiler (G1), total secondary air flow of the boiler (G2), total coal feed rate of the boiler (T), and boiler oxygen content (C). The LSTM neural network algorithm can be expressed by the mathematical formula: Q = L(G1, G2, T, C), where the ammonia water flow rate Q is the ammonia flow required at the next moment predicted by the algorithm based on the denitration outlet nitrogen oxide concentration and input data; L(X1, X2, X3, ..., Xn) is the neuron model of the LSTM neural network, and this algorithm is composed of multiple LSTM neuron models connected in series. G1, G2, T, and C are the input data of the neuron model. The input data is sequentially input into the first neuron model in chronological order of historical data through the input gate. After part of the information is forgotten by the forget gate, the output in chronological order is formed through the judgment of the activation conditions of the neuron model's cell state, and then output through the output gate. After passing through the preset number of neuron models, the final output is the ammonia flow rate required by the boiler at the next moment.

搜集汇总

数据集介绍

特点

该数据集包含6067条锅炉烟气出口氮氧化物浓度智慧控制数据，每2秒更新一次。数据用于通过智能学习和模型优化实现脱硝系统的精准控制，降低生产成本，算法采用长短期记忆神经网络进行预测。