CN-122014522-A - Fan blade vibration real-time monitoring system and method based on edge calculation

CN122014522ACN 122014522 ACN122014522 ACN 122014522ACN-122014522-A

Abstract

The invention belongs to the technical field of monitoring of thermal power plants, and discloses a fan blade vibration real-time monitoring system and method based on edge calculation, which are deployed on a fan site through edge calculation nodes and integrated into a plurality of modules to realize data localization real-time processing, the method has the advantages of greatly reducing data transmission delay, ensuring that hidden dangers such as abnormal vibration and the like can be diagnosed in time, avoiding abnormal vibration and even fracture accidents caused by fatigue load, corrosion and the like of the blade, and improving the timeliness and reliability of monitoring. The multidimensional sensor network collects vibration signals and auxiliary monitoring signals, and the multidimensional sensor network is matched with the data fusion unit to integrate multisource data, so that the signal quality is effectively improved, and data support is provided for accurate anomaly diagnosis. The cooperative architecture of the edge computing node and the cloud processing platform enables the edge side to bear preprocessing, feature extraction and abnormality diagnosis tasks with high real-time requirements, and the cloud is focused on depth model training, historical data mining and long-term trend analysis, so that the remote data transmission quantity is reduced, the bandwidth occupation and transmission cost are reduced, the real-time monitoring and long-term trend control are realized, and the running performance of the whole monitoring system is remarkably optimized.

Inventors

Zheng chuangwei
WANG PENG
ZHOU QUAN
YANG QING
CHEN HONGQING
XU JIAMENG
SHEN YINING
ZHANG BO

Assignees

华能汕头海门发电有限责任公司
西安热工研究院有限公司

Dates

Publication Date: 20260512
Application Date: 20260122

Claims (10)

1. Fan blade vibration real-time monitoring system based on edge calculation, its characterized in that includes: The edge computing node is deployed on the fan site, integrates a sensor interface, a data processing unit and a communication module and is used for realizing data localization real-time processing; the multidimensional sensor network is used for collecting vibration signals and auxiliary monitoring signals of the fan blades; The data fusion unit is used for integrating multi-source data acquired by the multi-dimensional sensor network and improving the signal quality; The cloud processing platform interacts with the edge computing node through a communication network; And the edge computing node performs data preprocessing, feature extraction and real-time anomaly diagnosis, and the cloud processing platform performs depth model training, historical data mining and long-term trend analysis.
2. The edge computing-based fan blade vibration real-time monitoring system of claim 1, wherein the edge computing node is an ARM architecture processor or an embedded computing module with a computing power not lower than 20 TOPS.
3. The fan blade vibration real-time monitoring system based on edge calculation, which is disclosed in claim 1, is characterized in that the multi-dimensional sensor network comprises acceleration sensors and temperature sensors, wherein the acceleration sensors are arranged at the root, middle and tip positions of the blade and are used for capturing space-time characteristics of vibration signals, the temperature sensors are used for monitoring the surface temperature change of the blade and assisting in vibration abnormality diagnosis, the acceleration sensors are industrial triaxial accelerometers, the measuring range is not lower than +/-500 g, and the sampling rate is not lower than 10kHz.
4. The edge-calculation-based fan blade vibration real-time monitoring system according to claim 1, wherein the data fusion unit integrates multi-source data by adopting a kalman filter or a modified algorithm thereof.
5. The edge computing-based fan blade vibration real-time monitoring system of claim 1, wherein the edge computing node is configured with a noise suppression module that eliminates environmental noise interference using a Savitzky-Golay filter or a signal smoothing algorithm.
6. The edge-calculation-based fan blade vibration real-time monitoring system according to claim 1, wherein the edge calculation nodes deploy a machine learning model to realize real-time judgment of vibration overrun, diagnosis delay is not more than 50ms and trigger primary alarm, and the cloud processing platform adopts a time sequence deep learning model to identify blade cracks and corrosion typical faults.
7. The edge computing-based fan blade vibration real-time monitoring system of claim 6, wherein the machine learning model is a random forest model and the time-series deep learning model is a Bi-LSTM model.
8. The edge-computing-based fan blade vibration real-time monitoring system according to claim 1, wherein the edge computing nodes are constructed based on an ROS2 operating system and support Python/C++ hybrid programming, and the cloud processing platform supports lateral expansion by adopting fault diagnosis micro-services deployed in a containerized manner.
9. The edge computing-based fan blade vibration real-time monitoring system according to claim 1, further comprising a report generation module supporting customized report template generation, automatic pushing and version management, and having a scene template adaptation capability, and adjusting monitoring parameters and report dimensions according to different operation environments.
10. The fan blade vibration real-time monitoring method based on edge calculation is characterized by comprising the following steps of: s1, collecting blade vibration signals and auxiliary monitoring signals through a multi-dimensional sensor network; s2, performing noise suppression and data integration on the acquired data by utilizing an edge computing node; s3, extracting key features of the preprocessed data by a time-frequency domain analysis method; s4, carrying out real-time abnormality judgment on the extracted features based on the lightweight model, and triggering a primary alarm; S5, uploading the local diagnosis result and the key feature data to a cloud, and performing accurate fault identification through a depth model to generate a cloud analysis result; And S6, generating operation and maintenance suggestions by combining cloud analysis results.

Description

Fan blade vibration real-time monitoring system and method based on edge calculation Technical Field The invention belongs to the technical field of monitoring of thermal power plants, and particularly relates to a fan blade vibration real-time monitoring system and method based on edge calculation. Background The electric energy is produced by using combustible matters as fuel through a conversion process of chemical energy, heat energy, mechanical energy and electric energy, a prime motor covers a steam engine, a gas turbine and the like, fan blades are key components in the operation of the electric energy generating machine, and related technologies have important application positions in the electric energy production industry. The fan blade of the thermal power plant is easily influenced by factors such as fatigue load, corrosion and the like in long-term operation, so that abnormal vibration and even broken safety accidents are caused. Aiming at the problem of abnormal operation of the fan blade, the technical scheme adopted at present is to monitor the operation state of the blade through a monitoring system so as to realize early warning and investigation of hidden danger such as abnormal vibration. The core processing mode of the monitoring system is cloud centralized processing, namely, the fan blade operation related data collected by the front end are transmitted to a cloud platform, and the cloud completes core processing work such as data analysis and judgment, and then a monitoring result of the blade operation state is obtained. The current monitoring scheme relying on cloud centralized processing has obvious technical defects, and the core problem is that the real-time and high-efficiency requirements of fan blade monitoring are difficult to meet. In particular, the data is transmitted from the front-end acquisition equipment to the cloud end, higher transmission delay exists in the process of transmitting the data to the cloud end, abnormal states of the operation of the blades cannot be fed back in time, hidden trouble investigation is possibly caused to be untimely, and safety accidents are further caused. Disclosure of Invention The invention provides a fan blade vibration real-time monitoring system and method based on edge calculation, which solve the problem that the real-time and high-efficiency requirements of fan blade monitoring are difficult to meet in the prior art. In order to achieve the above purpose, the present invention provides the following technical solutions: Edge calculation-based fan blade vibration real-time monitoring system comprises: The edge computing node is deployed on the fan site, integrates a sensor interface, a data processing unit and a communication module and is used for realizing data localization real-time processing; the multidimensional sensor network is used for collecting vibration signals and auxiliary monitoring signals of the fan blades; The data fusion unit is used for integrating multi-source data acquired by the multi-dimensional sensor network and improving the signal quality; The cloud processing platform interacts with the edge computing node through a communication network; And the edge computing node performs data preprocessing, feature extraction and real-time anomaly diagnosis, and the cloud processing platform performs depth model training, historical data mining and long-term trend analysis. Preferably, the edge computing node is an ARM architecture processor or an embedded computing module with computing power not lower than 20 TOPS. Preferably, the multi-dimensional sensor network comprises acceleration sensors and temperature sensors, wherein the acceleration sensors are arranged at the root, middle and tip positions of the blade and are used for capturing space-time characteristics of vibration signals, the temperature sensors are used for monitoring temperature changes of the surface of the blade and assisting in vibration abnormality diagnosis, the acceleration sensors are industrial triaxial accelerometers, the measuring range is not lower than +/-500 g, and the sampling rate is not lower than 10kHz. Preferably, the data fusion unit integrates multi-source data by adopting Kalman filtering or a modified algorithm thereof. Preferably, the edge computing node is configured with a noise suppression module, and the noise suppression module adopts a Savitzky-Golay filter or a signal smoothing algorithm to eliminate environmental noise interference. Preferably, the edge computing node deploys a machine learning model to realize real-time judgment of vibration overrun, diagnosis delay does not exceed 50ms and triggers primary alarm, and the cloud processing platform adopts a time sequence deep learning model to identify blade cracks and typical corrosion faults. Preferably, the machine learning model is a random forest model, and the time sequence deep learning model is a Bi-LSTM model. Preferably, the edge computing node is const