A multi-source sensor data spatio-temporal feature fusion elevator guide system anomaly detection method comprises the following steps: 1) preprocessing and normalizing collected sensor data of an elevator guide system, a dynamic time regularity distance between sensor data is calculated, a Gaussian kernel weighting strategy is combined to endow adjacent sensors with higher weights, the similarity between the sensor data is obtained, and a KNN composition algorithm is adopted to construct an elevator guide system sensor time sequence diagram; 2) extracting spatial features of the guiding system sensor time sequence diagram, extracting time features of the guiding system sensor time sequence diagram, and fusing vector representation of the time features and vector representation of the spatial features to obtain unified spatial-temporal feature vector representation; and 3) predicting through a multi-layer perceptron, comparing a predicted value with an actual observation value of the sensor, and if the difference between the predicted value and the actual observation value is greater than a threshold value, judging that abnormity occurs. According to the invention, the accuracy of anomaly detection is effectively improved.
一种多源传感器数据时空特征融合的电梯导向系统异常检测方法,包括以下步骤:1)对收集到的电梯导向系统的传感器数据进行预处理和归一化,通过计算传感器数据间的动态时间规整度距离并结合高斯核加权策略为邻近的传感器赋予更高权重,得到传感器数据间的相似度,采用KNN构图算法构建电梯导向系统传感器时序图;2)提取导向系统传感器时序图的空间特征,提取导向系统传感器时序图的时间特征,将时间特征的向量表示和空间特征的向量表示进行融合,得到统一的时空特征向量表示;3)经过多层感知机进行预测,将预测值与传感器实际观测值进行比较,若二者差异大于阈值,则判定为异常。本发明有效提升了异常检测的准确性。
Multi-source sensor data spatio-temporal feature fusion elevator guide system anomaly detection method
一种多源传感器数据时空特征融合的电梯导向系统异常检测方法
2024-08-16
Patent
Electronic Resource
Chinese
IPC: | B66B Aufzüge , ELEVATORS |
ELEVATOR DOOR ANOMALY DETECTION SYSTEM AND ELEVATOR DOOR ANOMALY DETECTION METHOD
European Patent Office | 2024
|Elevator detection method and device based on multi-source big data fusion
European Patent Office | 2022
|Multi-source data fusion elevator safe operation monitoring method
European Patent Office | 2023
|Flight target anomaly detection method based on multi-source data fusion
European Patent Office | 2021
|