CN-122008322-A - Multi-sensor collaborative detection method and system for fan cabin rail-hanging inspection robot

CN122008322ACN 122008322 ACN122008322 ACN 122008322ACN-122008322-A

Abstract

The invention belongs to the technical field of fan operation and maintenance, and particularly discloses a multi-sensing collaborative detection method and a multi-mode sensor array for a fan cabin rail-hanging inspection robot, which can synchronously acquire multi-physical-field data, and can improve the data quality and provide reliable data support for fault diagnosis by carrying out layered pretreatment and collaborative calibration on the multi-physical-field data; the diagnosis model integrating the graph neural network and the attention mechanism can accurately identify the fault type under multi-factor coupling, improves the intelligent level of the operation and maintenance of the fan cabin, can realize the automatic identification of the fault type, the fault degree and the fault association factor, provides visual fault diagnosis results for operation and maintenance personnel, facilitates the operation and maintenance personnel to develop operation and maintenance work in a targeted manner, reduces operation and maintenance difficulty and cost, improves the operation and maintenance efficiency of the fan cabin, has strong suitability and expansibility, can adapt to the operation and maintenance requirements of the fan cabin under different working conditions, and has wide engineering application prospects.

Inventors

WANG YAN
TANG HUI
Dong Wuxing

Assignees

成都鼎峰汇智科技有限公司

Dates

Publication Date: 20260512
Application Date: 20260408

Claims (10)

1. A multi-sensor collaborative detection method of a fan cabin rail-hanging inspection robot is characterized by comprising the following steps: Acquiring a multidimensional sensing data set transmitted by a main controller of a fan cabin rail-hanging inspection robot, wherein the multidimensional sensing data set is obtained by synchronously acquiring a multi-mode sensor array at an inspection end by the main controller of the fan cabin rail-hanging inspection robot, and comprises vibration sensing signals, infrared thermal imaging data and electromagnetic sensing signals; preprocessing and extracting characteristics of the vibration sensing signals, the infrared thermal imaging data and the electromagnetic sensing signals respectively to obtain vibration characteristic data, infrared temperature characteristic data and electromagnetic characteristic data; Carrying out data collaborative calibration on vibration characteristic data, infrared temperature characteristic data and electromagnetic characteristic data to obtain a multi-physical-field collaborative calibration data set, wherein the multi-physical-field collaborative calibration data set comprises the calibrated vibration characteristic data, infrared temperature characteristic data and electromagnetic characteristic data; inputting the multi-physical field collaborative calibration data set into a pre-trained multi-physical field coupling fault diagnosis model to carry out fault diagnosis, so as to obtain a fault diagnosis result, wherein the fault diagnosis result comprises a fault type, a fault severity and a fault correlation factor, the multi-physical field coupling fault diagnosis model comprises an input layer, a feature fusion layer, a coupling diagnosis layer and an output layer which are sequentially connected, the input layer is used for receiving the multi-physical field collaborative calibration data set, splicing the calibrated vibration feature data, infrared temperature feature data and electromagnetic feature data into a multi-dimensional input vector, the feature fusion layer is used for carrying out self-adaptive weighted fusion on the multi-dimensional input vector through an attention mechanism so as to obtain a coupling feature vector, the coupling diagnosis layer is used for carrying out multi-physical field coupling relation mining on the coupling feature vector through a graph neural network, extracting the fault feature vector, and the output layer is used for mapping the fault feature vector into the fault diagnosis result through a full connection layer and an activation function and outputting the fault diagnosis result; and transmitting the fault diagnosis result to a main controller of the fan cabin rail-hanging inspection robot.
2. The multi-sensing collaborative detection method for a fan cabin rail-mounted inspection robot according to claim 1, wherein the preprocessing and feature extraction are performed on vibration sensing signals, infrared thermal imaging data and electromagnetic sensing signals respectively to obtain vibration feature data, infrared temperature feature data and electromagnetic feature data, and the method comprises the following steps: Carrying out noise reduction treatment on the vibration sensing signal by adopting a wavelet threshold noise reduction algorithm to obtain a noise-reduced vibration sensing signal, carrying out time domain analysis and frequency domain analysis on the noise-reduced vibration sensing signal, extracting time domain features and frequency domain features, carrying out normalization treatment on the time domain features and the frequency domain features, and forming vibration feature data by utilizing the normalized time domain features and frequency domain features; Performing gray correction and histogram equalization on the infrared thermal imaging data to obtain enhanced infrared thermal imaging data, extracting temperature characteristics of key parts of the infrared thermal imaging data to obtain temperature characteristics, performing normalization on the temperature characteristics, and forming infrared temperature characteristic data by using the normalized temperature characteristics; And filtering the electromagnetic sensing signal by adopting a self-adaptive filtering algorithm to obtain a filtered electromagnetic sensing signal, extracting amplitude characteristics, phase characteristics and frequency characteristics of the filtered electromagnetic sensing signal to obtain the amplitude characteristics, the phase characteristics and the frequency characteristics, normalizing the amplitude characteristics, the phase characteristics and the frequency characteristics, and forming electromagnetic characteristic data by utilizing the normalized amplitude characteristics, the normalized phase characteristics and the normalized frequency characteristics.
3. The multi-sensing collaborative detection method for a fan nacelle rail inspection robot according to claim 2, wherein the time domain features include a peak value, an effective value, kurtosis and a waveform factor, the frequency domain features include a feature frequency and a harmonic amplitude, the temperature features include a maximum temperature, a minimum temperature, an average temperature and a temperature gradient, the amplitude features include a peak amplitude and an effective value amplitude, the phase features include a phase offset, and the frequency features include a main frequency and a harmonic frequency.
4. The multi-sensor collaborative detection method of a fan cabin rail-mounted inspection robot according to claim 1, wherein the performing data collaborative calibration on vibration feature data, infrared temperature feature data and electromagnetic feature data to obtain a multi-physical field collaborative calibration data set comprises: Performing linear calibration on the vibration characteristic data by using preset calibration vibration data to obtain calibrated vibration characteristic data, performing linear calibration on the infrared temperature characteristic data by using preset calibration temperature data to obtain calibrated infrared temperature characteristic data, and performing linear calibration on the electromagnetic characteristic data by using preset calibration electromagnetic data to obtain calibrated electromagnetic characteristic data; and carrying out time-associated storage on the calibrated vibration characteristic data, the infrared temperature characteristic data and the electromagnetic characteristic data to form a multi-physical-field cooperative calibration data set.
5. The multi-sensing collaborative detection method for the fan cabin rail-mounted inspection robot is characterized in that an input vector dimension of an input layer of a multi-physical field coupling fault diagnosis model is 22-dimensional, a feature fusion layer is a multi-head self-attention mechanism, the number of heads is set to be 6, the number of linear transformation dimensions is 22-dimensional, feature internal correlation weights are calculated through self-attention, coupling feature vectors of the 22-dimensional are calculated through cross attention, the coupling diagnosis layer is a graph neural network, the number of graph convolution layers of the graph neural network is set to be 2, the output dimension of a first layer graph convolution is 32-dimensional, the output dimension of a second layer graph convolution is 16-dimensional, a graph pooling layer of the graph neural network is global average pooling, the fault feature vectors of the 16-dimensional are output, the number of neurons of the first layer of the full-connection layer is 32, the activation function is a LU (re) activation function, the number of neurons of the second layer of the full-connection layer is 8, and the activation function is a Sotmax activation function.
6. The method for multi-sensor collaborative detection of a fan nacelle rail inspection robot of claim 5, wherein prior to inputting a multi-physical field collaborative calibration dataset into a pre-trained multi-physical field coupling fault diagnosis model for fault diagnosis, the method further comprises: The method comprises the steps of constructing a multi-physical field coupling fault diagnosis model, training the multi-physical field coupling fault diagnosis model by utilizing a plurality of preset groups of multi-physical field cooperative calibration data set positive samples and a plurality of groups of multi-physical field cooperative calibration data set negative samples to obtain the trained multi-physical field coupling fault diagnosis model, wherein the multi-physical field cooperative calibration data set positive samples are marked with normal labels, and the multi-physical field cooperative calibration data set negative samples are marked with fault types, fault severity and fault correlation factor labels.
7. The multi-sensor collaborative detection system of the fan cabin rail-hanging inspection robot is characterized by comprising a multi-mode sensor array, a main controller and an edge calculation module, wherein the multi-mode sensor array is arranged at an inspection end of the fan cabin rail-hanging inspection robot and comprises a vibration sensor, an infrared thermal imaging sensor and an electromagnetic sensor, the vibration sensor is used for collecting vibration sensing signals of a detected component, the infrared thermal imaging sensor is used for collecting infrared thermal imaging data of the detected component, the electromagnetic sensor is used for collecting electromagnetic sensing signals of the detected component, the main controller and the edge calculation module are arranged in the fan cabin rail-hanging inspection robot, and the main controller is used for synchronously controlling the acquisition time sequence of the vibration sensor, the infrared thermal imaging sensor and the electromagnetic sensor and synchronously acquiring the vibration sensing signals of the vibration sensor, the infrared thermal imaging data of the infrared thermal imaging sensor and the electromagnetic sensor to form a multi-dimensional sensing data set, and transmitting the multi-dimensional sensing data set to the edge calculation module for executing the method of the fan cabin rail-hanging inspection robot according to any one of the claims 6.
8. The multi-sensor collaborative detection system of a fan cabin rail-mounted inspection robot according to claim 7, wherein the main controller adopts a single chip microcomputer or a DSP chip, and the main controller synchronizes clock signals to a vibration sensor, an infrared thermal imaging sensor and an electromagnetic sensor through a CAN bus, an Ethernet PTP synchronization protocol or a UART bus.
9. The multi-sensor collaborative detection system for a fan cabin rail inspection robot according to claim 7, wherein the vibration sensor is a piezoelectric vibration sensor or a capacitive vibration sensor, the infrared thermal imaging sensor is a non-contact infrared thermal imaging sensor or a refrigeration type infrared thermal imaging sensor, and the electromagnetic sensor is a hall electromagnetic sensor or an induction type electromagnetic sensor.
10. Multi-sensor collaborative detection system of fan cabin rail-hanging inspection robot, which is characterized by comprising: a memory for storing instructions; The processor is used for reading the instructions stored in the memory and executing the multi-sensing collaborative detection method of the fan cabin rail hanging inspection robot according to the instructions.

Description

Multi-sensor collaborative detection method and system for fan cabin rail-hanging inspection robot Technical Field The invention belongs to the technical field of fan operation and maintenance, and particularly relates to a multi-sensor collaborative detection method and system for a fan cabin rail-hanging inspection robot. Background The interior environment of the cabin of the wind generating set is complex, and the wind generating set comprises key operation and maintenance components such as a gear box, a generator, a bearing and the like, and the running state of the wind generating set directly determines the generating efficiency and the safety stability of a fan. The fan cabin rail-hanging inspection robot is used as core equipment for cabin operation and maintenance, can replace manual inspection and fault diagnosis work in high-altitude and high-risk environments, and becomes a research hot spot for fan operation and maintenance technology. The existing fan cabin rail-hanging inspection robot operation and maintenance technology is concentrated in two directions of motion control (such as path planning and obstacle avoidance navigation) and single fault detection (such as bolt loosening detection and oil leakage detection), although the automation level of cabin operation and maintenance is improved to a certain extent, the fault diagnosis accuracy under complex working conditions is still obviously insufficient, the actual operation and maintenance requirements are difficult to meet, and the problem of fault misjudgment and missed detection of core pain points focused under a multi-physical field coupling scene is solved. The fan cabin is internally provided with a plurality of physical fields such as temperature, vibration, electromagnetic interference and the like, and all the physical fields are mutually coupled and mutually influenced, so that faults of key components show multi-factor correlation characteristics. For example, the generation and development of bearing wear faults are combined by the high-temperature environment in the cabin, equipment running vibration and electromagnetic interference, and the monitoring data of a single physical field cannot fully reflect the fault essence. The fault diagnosis scheme of the existing fan cabin rail-hanging inspection robot mostly relies on a single type of sensor (such as a vibration sensor and a temperature sensor) to collect data, or adopts a single dimension diagnosis model to conduct fault identification, and the influence of multiple physical field coupling is not fully considered. On the one hand, the data collected by a single sensor is limited in dimension, and fault characteristics under the interaction of multiple factors cannot be captured. On the other hand, the single-dimension diagnosis model is difficult to mine the association relation between different physical field data, and the situations of fault misjudgment and missed detection are easy to occur, for example, the early characteristics of bearing abrasion are easy to be ignored under the synergistic effect of high temperature and electromagnetic interference in the traditional vibration analysis method, so that the fault is enlarged, and the operation and maintenance cost and the safety risk are increased. Therefore, a technical scheme for collaborative detection of a fan cabin rail hanging inspection robot capable of realizing accurate multi-dimensional data acquisition and multi-factor associated fault identification is needed, so that the problems of low fault diagnosis accuracy and weak anti-interference capability in the prior art are solved. ‌ ‌ A Disclosure of Invention The invention aims to provide a multi-sensing collaborative detection method and system for a fan cabin rail-hanging inspection robot, which are used for solving the problems in the prior art. In order to achieve the above purpose, the present invention adopts the following technical scheme: in a first aspect, a multi-sensor collaborative detection method for a fan cabin rail-hanging inspection robot is provided, including: Acquiring a multidimensional sensing data set transmitted by a main controller of a fan cabin rail-hanging inspection robot, wherein the multidimensional sensing data set is obtained by synchronously acquiring a multi-mode sensor array at an inspection end by the main controller of the fan cabin rail-hanging inspection robot, and comprises vibration sensing signals, infrared thermal imaging data and electromagnetic sensing signals; preprocessing and extracting characteristics of the vibration sensing signals, the infrared thermal imaging data and the electromagnetic sensing signals respectively to obtain vibration characteristic data, infrared temperature characteristic data and electromagnetic characteristic data; Carrying out data collaborative calibration on vibration characteristic data, infrared temperature characteristic data and electromagnetic characteristic data to obtai