Real-Time Optimization of Power Plant Boiler Based on Pattern Matching and Neural Networks

Using machine models to characterize the operational experience of complex systems, instead of manual optimization decisions, has become an advanced solution for optimizing the operation of various power plants. A new modeling framework based on pattern matching and neural networks is proposed to improve the effectiveness and real-time performance of coal-fired boiler operations in power plants. A three-layer optimization mechanism of attention mechanism parameters, interval frequency method, and the regulation minimum is introduced to ensure the effectiveness of the pattern matching scheme. Furthermore, we build the neural networks-based agent model that characterizes pattern matching steps for online applications. The case results show that the proposed pattern matching optimization model can effectively find the optimal boiler operation scheme; the similarity of working conditions and operation is more than 95%; the boiler efficiency can be increased by 2.15% in practice; the mean square error of the trained agent model is within 0.54%. Compared with the traditional methods, this method avoids the influence of generalization error caused by the optimization solver algorithm and improves the boiler efficiency while having high reliability and calculation response performance.