基于IPO-XGBoost混合模型的汽轮机热耗率软测量方法OA

Aiming at the energy efficiency optimization dilemma caused by the insufficient accuracy of the thermal unit heat consumption rate online monitoring,this study proposes a hybrid modeling framework integrating the improved parrot optimization algorithm(IPO)and extreme gradient boosting tree(XGBoost).The framework enhances the basic parrot optimization algorithm through a triple strategy:introducing chaotic mapping and inverse learning mechanisms to enhance population diversity,designing adaptive inertia weight switching factors to balance global exploration and local exploitation capabilities,and innovating hybrid Gauss-Cauchy variation operators to enhance the probability of jumping out of the local optimal solution.In order to verify the effectiveness of the algorithm,four typical functions,single-peak(sphere),single-peak pathological(rosenbrock),multi-peak(rastrigin)and multi-peak(ackley),are selected from the IEEE CEC standard test function library for benchmarking.The experimental results show that,compared with the original PO algorithm,the IPO significantly reduces the number of convergent generations,and the global optimal solution improves the localization accuracy significantly,showing excellent optimization performance.Following this,this study innovatively applies it to the hyper-parameter optimization process of extreme gradient boosted tree model,and constructs the IPO-XGBoost prediction model,which realizes the global optimization search for the complex parameter space of the model.The experimental results show that the optimized model can capture the nonlinear characteristics of the thermal system more accurately,and its prediction results provide reliable data support for the optimization of the unit's energy efficiency,with significant emission reduction benefits.This study innovatively combines chaos optimization theory with integrated learning models to provide a new methodological framework for intelligent modeling of complex energy systems.