基于子空间混合相似度的气象站dy-01电源故障智能监测方法

摘要:

气象站设备的差异化和多样化导致电源在出现故障时,通过设备性能下降、数据异常、启动困难等间接方式反映。这些故障特征微弱且多变,且可能与其他设备故障或环境因素混淆,增加了故障特征准确识别的难度。为提高气象站DY-01电源故障的监测效果,提出基于子空间混合相似度的气象站DY-01电源故障智能监测方法。以气象站DY-01电源为监测对象,通过高精度传感器实时采集气象站DY-01电源的电压、电流、功率等运行数据,利用子空间技术将预处理后的高维数据映射到低维子空间中,从时域和频域两个角度细化电源故障特征分量,识别电源故障状态与类型。采用混合相似度算法计算实时数据与历史故障数据之间的相似度,并依照相似度预设的阈值,执行可视化输出与预警程序,实现气象站DY-01电源故障智能监测任务。实验结果表明,所提方法的电压和电流监测误差分别减小2.7V和0.245A,误警率在2%以下,同时监测覆盖系数无限接近1.0,由此证明所提方法具有更优的监测精度和范围。

Abstract:

The differentiation and diversification of meteorological station equipment lead to indirect reflection of power supply failures through equipment performance degradation, data anomalies, and difficulty starting. These fault features are weak and varied, and may be confused with other equipment faults or environmental factors, increasing the difficulty of accurately identifying fault features. To improve the monitoring effectiveness of DY-01 power failure in meteorological station, an intelligent monitoring method for DY-01 power failure based on subspace mixed similarity is proposed. Taking the DY-01 power supply of the meteorological station as the monitoring object, high-precision sensors are used to collect real-time operating data such as voltage, current, and power of the DY-01 power supply. The preprocessed high-dimensional data is mapped to a low dimensional subspace using subspace technology, and the power fault characteristic components are refined from both time domain and frequency domain perspectives to identify the power fault status and type. Using a hybrid similarity algorithm to calculate the similarity between real-time data and historical fault data, and executing visual output and warning programs according to the preset threshold of similarity, the DY-01 meteorological station achieves intelligent monitoring of power faults. The experimental results show that the voltage and current monitoring errors of the proposed method are reduced by 2.7V and 0.245A, respectively, with a false alarm rate of less than 2%. At the same time, the monitoring coverage coefficient is infinitely close to 1.0, which proves that the proposed method has better monitoring accuracy and range.