基于复合特性Elman神经网络的风速预测

发布时间：2018-04-06 07:30

本文选题：Elman神经网络　切入点：迟滞　出处：《天津工业大学》2016年硕士论文

【摘要】：风速时间序列预测在风电场运行过程中起着非常重要的作用,它不仅关系到风力发电机的投入运营量,也关系到电力系统的安全、可靠、经济的运行。风速时间序列的预测精度直接影响着风电领域电力系统的经济性与稳定性,因此风速时间序列预测是国内外重点关注的研究领域之一。近些年来,随着人工神经网络建模的成熟化,人工神经网络模型作为一种重要的预测手段在风速时间序列的预测中得到广泛应用。在实际应用过程中,由于BP神经网络出现较早,技术较成熟,通常采用BP神经网络模型建模,与BP神经网络相比,Elman神经网络模型由于具备反馈性与递归性,在复杂的风速时间序列预测中具有一定的优势,然而Elman神经网络还存在易陷入局部极小值,预测精度较差等缺点。针对传统Elman神经网络存在的缺点与不足,本文提出将迟滞现象与Elman神经网络结合的方法,利用平稳化处理后的数据构建出具有复合特性Elman神经网络模型。具体做法是在传统Elman神经网络的基础上,将迟滞因子引入隐藏层和关联层,月本文中构建的迟滞函数所替代传统的Sigmoid函数。由于迟滞函数是由两条闭合的曲线构成,在进行数据处理时,依据指定的规则进行上下曲线的选择,使得输出数据具有一定的随机跳变性,可以在一定程度上跳出局部极小值,使用平稳化处理后的数据,进一步提高了预测的精度。本文采用华北某风电场实际数据对提出的模型和方法进行仿真预测,并与传统Elman神经网络进行对比,结果显示该模型可以明显提高预测精度。在提出算法的基础上,本文还提出将算法与实际结合,达到硬件上的实现,成功的做到采集—处理—展现的整个过程。在普通的PC机上即可获取远程风力发电机坐标位置的下一时刻预测风速。
[Abstract]:The prediction of wind speed time series plays a very important role in the operation of wind farm. It not only relates to the operational quantity of wind turbine, but also to the safe, reliable and economical operation of power system.The prediction accuracy of wind speed time series has a direct impact on the economy and stability of power system in wind power field, so wind speed time series prediction is one of the most important research fields at home and abroad.In recent years, with the maturity of artificial neural network modeling, artificial neural network model as an important means of forecasting wind speed time series has been widely used.In the process of practical application, because BP neural network appeared earlier and the technology was mature, the model of BP neural network is usually modeled. Compared with BP neural network, Elman neural network model has feedback and recursion.There are some advantages in the prediction of complex time series of wind speed, but the Elman neural network is prone to fall into local minima and the prediction accuracy is poor.In view of the shortcomings and shortcomings of traditional Elman neural networks, this paper proposes a method of combining hysteresis with Elman neural networks, and constructs a Elman neural network model with compound characteristics by using the data processed by stationary processing.On the basis of the traditional Elman neural network, the hysteresis factor is introduced into the hidden layer and the correlation layer. The hysteresis function constructed in this paper replaces the traditional Sigmoid function.Because the hysteresis function is composed of two closed curves, when the data is processed, the upper and lower curves are selected according to the specified rules, which makes the output data have a certain random jump variation, which can jump out of the local minimum to a certain extent.The precision of prediction is further improved by using the data after stabilization.This paper uses the actual data of a wind farm in North China to simulate and predict the proposed model and method, and compares it with the traditional Elman neural network. The results show that the model can obviously improve the prediction accuracy.On the basis of the proposed algorithm, this paper also proposes to combine the algorithm with practice to achieve the hardware implementation, and successfully achieve the whole process of collecting, processing and displaying.The wind speed can be predicted at the next moment when the coordinate position of the remote wind turbine can be obtained on the ordinary PC computer.
【学位授予单位】：天津工业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP183;TM614

【相似文献】