基于VMD-Informer-BiLSTM模型的超短期光伏功率预测

由于光伏出力的波动性和随机性给电网的规划和运行带来了挑战,因此提高光伏功率预测的精度对提升新能源系统的稳定性具有重要意义。为此提出了一种结合模态分解、多维特征建模、Informer、双向长短期神经网络(bidirectional long short-term memory network,BiLSTM)的超短期光伏组合预测模型。首先通过变分模态分解将光伏功率序列分解成不同频率的本征模态函数(intrinsic mode function,IMF),降低光伏功率信号的非平稳性与复杂度;随后使用离散小波变换提取天气因素中的细节分量,实现不同分解算法的优势互补,并用随机森林算法为每个IMF筛选冗余特征,然后将特征矩阵送入Informer进行建模,提取不同时间步中关键时刻的信息,提高对长时间序列的预测效率;最后为进一步提高模型预测精度,分析误差序列特性,利用BiLSTM进行误差校正。采用实际光伏数据进行算例分析,结果表明所提方法提高了超短期光伏功率预测精度。     

晶型对等规聚丙烯电导电流和空间电荷特性的影响

聚丙烯材料性能改善过程中,材料内部会出现以α晶或β晶为主的2种晶型。为研究晶型对等规聚丙烯(iPP)电导电流和空间电荷特性的影响,选用α和β成核剂制备了具有不同晶型的iPP–pure(对照组)、iPP–α和iPP–β试样,开展了25℃下电导电流、空间电荷和直流击穿特性的研究。结果表明:iPP–β正负直流击穿强度最高,而iPP–α最低;3组iPP试样的电导电流大小为iPP–αiPP–pureiPP–β,空间电荷限制电流的转折电压大小为iPP–αiPP–βiPP–pure;-20 MV/m电场强度极化30 min后,3组iPP试样在极化过程中均出现正负空间电荷积聚;去极化过程中iPP–pure中空间电荷积聚量最大,而iPP–α中最少。分析认为:α晶和β晶的晶胞形态使得iPP–α中存在大量浅陷阱、iPP–β中平均陷阱深度较深;载流子在iPP–α中容易迁移并产生俄歇效应,而在iPP–β中容易在试样表面附近被捕获。     

Characterization and comprehension of corona partial discharge in air under power frequency to very low frequency voltage

For the partial discharge test of electrical equipment with large capacitance, the use of lowfrequency voltage instead of power frequency voltage can effectively reduce the capacity requirements of test power supply. However, the validity of PD test under low frequency voltage needs to be evaluated. In order to investigate the influence of voltage frequency on corona discharge in the air, the discharge test of the tip-plate electrode under the frequency from 50 to 0.1 Hz is carried out based on the impulse current method. The results show that some of the main features of corona under low frequency do not change. The magnitude of discharge in a positive half cycle is obviously larger than that in a negative cycle. The magnitude of discharge and interval in positive cycle are random, while that in negative cycle are regular. With the decrease of frequency, the inception voltage increases. The variation trend of maximum and average magnitude and repetition rate of the discharge in positive and negative half cycle with the variation of voltage frequency and magnitude is demonstrated, with discussion and interpretation from the aspects of space charge transportation, effective discharge time and transition of discharge modes. There is an obvious difference in the phase resolved pattern of partial discharge and characteristic parameters of discharge patterns between power and low frequency. The experimental results can be the reference for mode identification of partial discharge under low frequency tests. The trend of the measured parameters with the variation of frequency provides more information about the insulation defect than traditional measurements under a single frequency (usually 50 Hz). Also it helps to understand the mechanism of corona discharge with an explanation of the characteristics under different frequencies.