粒子群优化RBF神经网络的自适应谐波检测

Application of RBF neural networks based on adaptive particle swarm optimization algorithm in harmonic detection

有源电力滤波器补偿性能与所采用的谐波检测方式有很大的依赖关系,针对现有的检测方法存在精度不高、对电网频率变化比较敏感、自适应能力不强的缺点,本文提出基于粒子群优化算法的R B F神经网络的谐波检测方法。用自适应的方法对粒子群优化算法的参数进行了调整,使其能够更好地适应复杂的非线性环境,从而可以更灵活地调节P S O算法的全局搜索能力和局部开发能力。在算法的基础上,根据已开发的系统配置和学习算法,探讨了模拟电路的实现方法,运用P S I M软件对电路进行了模拟仿真。仿真结果表明,该方法具有很好的实时性、较高的检测精度以及自适应跟踪负载电流变化的能力。

The compensating capability of active power filter has relation with the way of harmonic detection,but the existing methods of harmonic detection have some disadvantages,such as the lower harmonic precision,sensitive to the change of the power system frequency,lower self-adaptive capability.This paper put forward a method for harmonic detection based on RBF Neural Network with parameters optimized by adaptive particle swarm optimization algorithm.It can better adapt to the complex non-linear environment,more flexibility in adjusting the global search ability and local development capability.On the basis of the learning algorithm for the developed system,the realization scheme of an analog circuit of the system is discussed,in the light of PSIM software,the computer simulation studies of the circuit are done.Simulation results show the accuracy and practicability of the approach and the ability of adaptive change with load current.