论文部分内容阅读
建立通用而精确的太阳电池热模型对光伏系统的建模、输出功率与转换效率的损失分析至关重要.基于复杂的太阳电池温度机理,分别研究了太阳电池温度的稳态热模型(steady state thermal model,SSTM)和支持向量机(support vector machines,SVM)方法建立的精确预测热模型.首先,基于空气温度、太阳辐射强度、风速3个最主要因素与太阳电池温度的近似线性关系,在已有SSTM的基础上,建立并校正了太阳电池的SSTM并采用差分进化算法提取模型的未知参数.其次,为提高SVM的模型预测精度,采用粒子群优化(particle swarm optimization,PSO)算法对SVM的核参数和惩罚因子进行动态寻优,在确定输入/输出样本集并划分训练集和测试集的基础上,建立了基于粒子群优化支持向量机(PSO-SVM)的太阳电池温度精确预测热模型.最后,搭建实验平台,在实验操作过程中减弱空气湿度、太阳入射角和热迟滞效应等因素对太阳电池温度的耦合.通过实验对比表明,建立的预测热模型性能可靠、全面、简洁,其参数寻优算法优于遗传算法和交叉校验法,模型预测精度优于反向传播神经网络(back propagation neural network)和SSTM.
It is very important to build a general and accurate thermal model of solar cell for the modeling of PV system, loss of output power and conversion efficiency.Based on the complicated solar cell temperature mechanism, the steady state thermal model of solar cell temperature thermal model, SSTM and support vector machines (SVM) method.Firstly, based on the approximate linear relationship between air temperature, solar radiation intensity and wind speed, and the solar cell temperature, On the basis of SSTM, the SSTM of solar cell was established and the unknown parameters of the model were extracted by differential evolution algorithm.Secondly, in order to improve the model prediction accuracy of SVM, particle swarm optimization (PSO) Based on Particle Swarm Optimization Support Vector Machine (PSO-SVM), the accurate prediction of the solar cell temperature is established based on the dynamic optimization of the nuclear parameters and penalty factors of the input / output samples and the classification of the training set and the test set. Model.Finally, the experimental platform was set up to attenuate the air humidity, solar incident angle and thermal hysteresis during the experiment The experimental results show that the proposed thermal model is reliable, comprehensive and concise, and its parameter optimization algorithm is superior to the genetic algorithm and cross-check method, and the prediction accuracy of the model is better than that of the backpropagation neural network back propagation neural network and SSTM.