动态属性权重智能前馈神经网络电力控制方法

Power Control Method Based on Intelligence Feed Forward Neural Network with Dynamic Attribute Weights

在工业园区光伏并网控制中,电力控制驱动过程中呈现阻尼振荡,影响系统响应速度和稳定性。提出一种改进的基于动态属性权重人工智能前馈神经网络的电力控制方法,提高工业园区光伏并网电力控制性能。控制系统设计中,采用两个PI控制器,并结合使用磁链控制器和转速控制器,在转子磁场坐标系下,利用仿射PARK变换,使得工业光伏并网控制电力系统所有运算在转子磁场坐标系下实现。构建人工智能前馈神经网络,通过功率前馈控制加快并网系统的响应速度,采用动态属性权重预测电流控制技术补偿延时,实现无差拍控制和正弦脉宽调制。系统测试表明,采用该控制方法进行工业园区光伏并网电力控制,能有效抑制控制延时和电感量偏差对并网电流造成的畸变,控制系统鲁棒性和稳定性较高。

According to the power control of grid connected photovoltaic industrial park, power control driven process has oscillation effect, it affects the system response speed and stability. A power control method for dynamic attribute weights of artificial intelligence based on the feed forward neural network is proposed, it improves industrial park photovoltaic grid connected power control performance. In the design of the control system, two PI controllers are used, combined with the use of flux controller and speed controller, a magnetic field in the rotor coordinate system is obtained. Using the affine PARK transform, make industrial photovoltaic grid connected power system control of all operations in the rotor magnetic field coordinates. Dynamic attribute weights is proposed to predictive current control technique for compensating the delay, the deadbeat control and sinusoidal pulse width modulation is realized. System test shows that, by using the control method to control the power industrial park, the distortion and control delay can be effectively restrained, it has high stability and robustness of the control system.