生活垃圾焚烧智能控制方法研究

Research on intelligent control method of domestic waste incineration

中文关键词: 生活垃圾焚烧  神经网络  模糊推理系统  自动控制

英文关键词: domestic waste incineration  neural network  fuzzy inference system  automatic control

中文摘要:

      焚烧因其减容、减量及能源回收利用等优势而成为生活垃圾主要处理技术．为了解决生活垃圾焚烧过程控制中多个运行操作参数调节困难的问题，利用机器学习(ML)对垃圾焚烧过程中的运行操作参数和控制变量进行高精度非线性映射，达到根据控制变量要求来自动定量调节运行操作参数的目的．采用神经网络和模糊推理系统算法来构建拟合控制模型，基于Aspen Plus垃圾焚烧模拟数据的训练和验证，得出的最优模型为SC-ANFIS，此模型对垃圾进料量、空气供给量、氨水投加量和氢氧化钙溶液投加量预测结果的决定系数(R2)分别为0.832 2、0.996 5、0.995 7和0.999 4，平均绝对百分比误差(MAPE)分别为2.133 0%、 0.683 5% 、1.878 2%和0.640 0%．因此，该模型可应用于生活垃圾焚烧过程控制，提高垃圾焚烧控制精度及自动化程度．

英文摘要:

      Incineration has become the main treatment technology of domestic waste because of its advantages in capacity reduction, quantity reduction, energy recycling and utilization. In order to solve the problem of difficult adjustment of multiple operating parameters in the process control of domestic waste incineration, machine learning (ML) is used to perform high-precision nonlinear mapping of operating parameters and control variables in the process of waste incineration, so as to achieve the purpose of automatically and quantitatively adjusting operating parameters according to the requirements of control variables. Neural network and fuzzy inference system algorithm are used to construct the fitting control model. Based on training and validation with the simulation data of waste incineration in Aspen Plus, the obtained optimal model is SC-ANFIS. The determination coefficients (R2) of the prediction results of waste feeding amount, air supply, ammonia water dosage and calcium hydroxide solution dosage are 0.832 2, 0.996 5, 0.995 7 and 0.999 4 respectively, and the mean absolute percentage errors (MAPE) are 2.133 0%, 0.683 5%, 1.878 2% and 0.640 0% respectively. Therefore, the model can be applied to the control of domestic waste incineration process to improve the accuracy and automation of waste incineration control.