LSTM预估补偿下的超临界大容量火电机组一次调频控制算法

A Primary Frequency Modulation Control Algorithm for Supercritical Large Capacity Thermal Power Units under LSTM Predictive Compensation

超临界大容量火电机组在负荷突变时,因存在多个频率信号一次调频过程会出现随机波动和非线性,导致控制效果不佳。因而,对超临界大容量火电机组提出一种基于LSTM的一次调频预估补偿控制算法。分析了超临界大容量火电机组多个频率信号情况进行信息融合;采用LSTM网络建立预估补偿控制器,根据一次调频控制示意图,获取控制过程中参数的学习率,不断调整AGC指令与负荷指令之间的非线性关系使其始终保持最佳状态,实现超临界大容量火电机组一次调频控制。为验证设计控制算法的有效性,设计算例进行性能测试,测试结果表明,与传统的一次调频算法相比,设计的LSTM预估补偿下的超临界大容量火电机组一次调频控制算法在负荷突增和突减两种场景下,能够有效减少频率变化偏差,提升调频能力。

When the load of supercritical large capacity thermal power units suddenly changes,the control effect of the primary frequency modulation process is poor due to the random fluctuations and nonlinearity of multiple frequency signals.An LSTM based primary frequency modulation predictive compensation control algorithm is proposed for supercritical large capacity thermal power units.This paper analyzes the situation of multiple frequency signals in supercritical large capacity thermal power units for information fusion,establishes a predictive compensation controller by using LSTM network,obtains the learning rate of parameters during the control process based on the primary frequency regulation control diagram,continuously adjusts the non-linear relationship between AGC instructions and load instructions to maintain the optimal state,and achieves the primary frequency regulation control of supercritical large capacity thermal power units.To verify the effectiveness of the designed control algorithm,a performance test is conducted.The test results show that compared with traditional primary frequency regulation algorithms,the designed LSTM predictive compensation based primary frequency regulation control algorithm for supercritical large capacity thermal power units can effectively reduce frequency variation deviation and improve frequency regulation ability under two scenarios of sudden load increase and decrease.