考虑多信息融合的风电机组状态评估与预测系统开发
Development of the Wind Turbine State Evaluation and Prediction System Considering Multi-Information Fusion
马铁强 1向凯1
作者信息
- 1. 沈阳工业大学机械工程学院,辽宁沈阳 110870
- 折叠
摘要
为降低风电机组故障发生概率,提高其可靠性,该文利用数据采集与监控系统(supervisory control and data acquisition,SCADA)检测获得的风电机组运行状态数据,通过研究多指标融合状态评价模型及其预测算法,解决风电机组状态参数评估与预测难题.结合SCADA系统结构,设计并规划风电机组状态参数评估与预测系统架构与功能;利用输出功率波动、风能利用率以及开机运行比率 3 项参数,基于阈值法,建立风电机组状态退化评价指标模型,通过主成分分析法对 3个评估标准进行权重计算,并将各指标进行信息融合,综合反应风电机组运行状态;设计Convolutional Neural Network-Long Short-Term Memory风电机组状态预测模型,实施风电场运行状态参数预测;开发风电机组状态评估与预测系统软件,验证所提方法的有效性.
Abstract
To reduce the probability of failures of the wind turbine and improve its reliability,this paper uses the wind turbine running state data obtained by SCADA(supervisory control and data acquisition)detection and studies the multi-index fusion state evaluation model and its prediction algorithm to overcome difficulties in the evaluation and prediction of wind turbine state parameters.Firstly,the architecture and functions of the wind turbine state parameter evaluation and prediction system are designed and planned based on SCADA system structure.Secondly,three parameters of output power fluctuation,wind energy utilization rate and start-up operation ratio are used to establish the wind turbine state degradation evaluation index model based on the threshold method.The principal component analysis method is used to calculate the weight of the three evaluation criteria,and the information of each index is fused to reflect the wind turbine state comprehensively.In addition,the CNN-LSTM wind turbine state prediction model is designed to implement the wind farm operating state parameter prediction.Finally,a wind turbine state evaluation and prediction system software is developed based on the above theory,and the effectiveness of the above method is verified.
关键词
风电机组/状态评估/CNN-LSTM/信息融合Key words
wind turbine/state evaluation/CNN-LSTM/information fusion引用本文复制引用
基金项目
国家自然科学基金重点项目(51537007)
出版年
2024
NETL
NSTL
维普
万方数据