基于神经网络内模的岸电电源波形控制

Output Waveform Control of Shore Power Supply Based on Neural Network Internal Model

岸电电源是在船舶制造、远洋钻井平台、码头等方面应用非常广泛的稳频稳压电源,但它对输出电压波形的稳态精度、动态性能等要求较高。通过对现有各种逆变控制方法进行研究和比较,在建立和分析岸电电源三相逆变器数学模型的基础上,针对它在旋转αβ坐标系下的特点,提出了一种神经网络内模控制和PID控制相结合的复合波形控制策略。建立了3层BP神经网络内部模型,并将其作为三相逆变器的内部模型,预测实际输出,采用PID控制器在线修正、补偿内部模型与三相逆变器之间的模型失配。对比仿真和实验数据,可以看出:当系统接阻性负载或整流性负载时,该控制方法所得的稳态电压误差和输出电压的谐波畸变率均小于其他控制方法,表明该策略控制下的岸电电源稳态精度高、输出波形质量好,具有一定的工程实用价值,能很好地应用于岸电电源逆变波形控制。

Shore power is regulated power supply widely used in fields of shipbuilding,offshore drilling platforms and terminals,and the requirments for its output voltage waveform,the steady-state accuracy and dynamic performance are higher.Through researching and comparing the existing various output waveform control algorithms,according to characteristics of three-phase inverter of the shore power supply based on αβ rotating frame,we put forward a mutiple control strategy of the output waveform based on the PID theory and neural network internal model theory.The mutiple control strategy took a neural network estimation model as the internal model of three-phase inverter to estimate the actual output waveform.The PID control model was established so that the internal model could fit the actual three-phase inverter according to the estimate error.Comparing simulation and experimental data,it can be seen that the voltage steady state error and the output voltage harmonic distortion ratio from the control algorithms are less than those by other control algorithms when the system is connected to a resistive load or rectifier load.The simulation and experiment verify that the mutiple control strategy can work well,and highly accurate steady output waveform is gained.The proposed method can be well applied to shore power inverter waveform control.