According to the issues of low accuracy and insufficient generalization ability of NOx concentration prediction models in selective catalytic reduction(SCR)systems

it establishes a least squares support vector machine(LSSVM) prediction model based on the progressive particle swarm optimization algorithm(PrePSO).Through dual innovation enhancing predictive performance

firstly

it uses the optimal Latin hypercube sampling for particle initialization

utilizing its spatial filling properties to evenly cover the design domain

effectively avoiding local aggregation caused by traditional random initialization.Secondly

it dynamically adjusts the inertia weights and learning factors to construct a nonlinear parameter update mechanism that balances global exploration and local development capabilities.Finally

based on the operating data of a 660 MW unit in a power plant

it constructs a PrePSO-LSSVM model.Through three error indicators

it compares and analyzes the predictive performance of the PrePSO-LSSVM model with LSSVM and PSO-LSSVM models.The results show that the improved model improves the prediction accuracy by more than 30% compared to the basic model.This method provides an efficient predictive tool for precise control of NOx in SCR systems and has important engineering application value in reducing ammonia escape rate.