基于支持向量机的光伏逆变器损耗预测方法

A Power Loss Prediction Method of Photovoltaic Inverters Based on Support Vector Machine

提出一种光伏逆变器损耗整体预测建模的方法。区别于分别对开关管和磁性元件等建立复杂的物理和数学模型,进而得到电路损耗的传统预测方法,从电路整体的损耗出发,基于小样本采样,采用智能学习算法支持向量机建立电路的效率模型,利用遗传算法对模型2个重要参数g和C进行优化。采用一个10 kW的三相光伏并网逆变器为研究对象,以输入电压和输出功率作为支持向量机的x变量,建立了基于遗传优化支持向量机的效率模型。模型误差精度小于0.6%,验证了模型的有效性和准确性。遗传优化支持向量机的效率模型用简单的代数公式代替了复杂而不规律的效率计算公式,简化了模型,提高了模型的预测精度。

A method of modeling the whole loss prediction of photovoltaic inverter is presented.It is different from traditional prediction methods that the complex physical and mathematical models of switching device and magnetic components are estabhshed and then circuit losses are obtained.The proposed method proceeds from the power loss of overall circuit angle which based on the small sample used intelligent algorithm learning support vector machine to build efficiency model of circuit, and then uses the genetic algorithm to optimize the two important parameters g and C of the model.The research object adopts a three-phase photovoltaic grid-connected inverter based on 10 kW.The input voltage and output power are used as variable x of the support vector machine.The efficiency model based on genetic algorithm support vector machine is established.The model error accuracy is less than 0.6% which verifies the validity and accuracy of the model.The efficiency models of support vector machine optimized by genetic algorithm uses simple algebraic formula replace the complex and irregular efficiency calculation formula which simplify model complexity and improve the prediction accuracy of the model.