CN-121980318-A - Greasy dirt detection method and system based on carrying multi-sensor machine dog

CN121980318ACN 121980318 ACN121980318 ACN 121980318ACN-121980318-A

Abstract

The invention relates to an oil stain detection method and system based on a multi-sensor-carrying machine dog, wherein the method comprises the steps of controlling the multi-sensor-carrying machine dog to move in equipment to be subjected to oil stain detection according to a preset track, collecting multi-sensor data and pose data of the machine dog, preprocessing the multi-sensor data and the pose data, extracting features of the preprocessed multi-sensor data, constructing a feature matrix based on the extracted features, obtaining the overlapping rate of the feature matrix based on the duty ratio of suspected oil stain units in the feature matrix, carrying out oil stain identification on the overlapping rate based on a preset threshold value, recording the oil stain position, judging whether micro leakage occurs to the oil stain position based on the preprocessed multi-sensor data and the preprocessed pose data, and calculating the coordinates of leakage points if micro leakage occurs to the oil stain position.

Inventors

Ning Jiecheng
ZHANG XI
HUANG ZHONG
Yao Hangzhi
XU JUN
LI XUERONG
YANG BO

Assignees

四川华能太平驿水电有限责任公司
四川九洲北斗导航与位置服务有限公司

Dates

Publication Date: 20260505
Application Date: 20260127

Claims (10)

1. The method for detecting the greasy dirt based on the machine dog carrying the multiple sensors is characterized by comprising the following steps of: The method comprises the steps of controlling a machine dog carrying a plurality of sensors to move in equipment to be subjected to oil stain detection in a preset track, collecting multi-sensor data and pose data of the machine dog, and preprocessing the multi-sensor data and the pose data; Extracting features of the preprocessed multi-sensor data, and constructing a feature matrix based on the extracted features, obtaining the overlapping rate of the feature matrix based on the duty ratio of suspected greasy dirt units in the feature matrix, identifying greasy dirt based on a preset threshold value, and recording the greasy dirt position; and judging whether the oil stain position is subjected to micro leakage or not based on the preprocessed multi-sensor data and the preprocessed pose data, and if the oil stain position is subjected to micro leakage, calculating the coordinates of leakage points.
2. The method for detecting greasy dirt based on carrying multi-sensor machine dog according to claim 1, wherein the multi-sensor comprises an optical sensor, a laser sensor and a temperature sensor, wherein: The optical sensor comprises a visible light camera for acquiring visible light images and a short-wave infrared sensor for acquiring infrared temperature information; The laser sensor is used for measuring laser spectrum data of the preset gas; The temperature sensor comprises a platinum resistance temperature sensor for collecting the ambient temperature, and the collected ambient temperature is compensated by utilizing a PID algorithm.
3. The method for detecting the greasy dirt based on the multi-sensor-carried machine dog according to claim 2, wherein the preprocessing of the multi-sensor data is specifically as follows: gaussian noise reduction and edge enhancement are carried out on the visible light image, temperature normalization is carried out on infrared temperature information, and baseline correction is carried out on laser spectrum data; The preprocessing of pose data is specifically to correct the pose of the machine dog by adopting a Kalman filtering algorithm.
4. The method for detecting the greasy dirt based on the multi-sensor-carried machine dog according to claim 1, wherein the method is characterized in that the characteristics of the preprocessed multi-sensor data are extracted, and a characteristic matrix is constructed based on the extracted characteristics, and specifically comprises the following steps: Carrying out space coordinate alignment on the preprocessed multi-sensor data, and unifying the space coordinate alignment to a two-dimensional inspection plane coordinate system taking a machine dog as a center; dividing the two-dimensional inspection plane into a plurality of grid cells, wherein each grid cell is a detection unit; extracting an optical characteristic vector and a laser characteristic vector of each detection unit; Performing binarization judgment on the optical feature vector and the laser feature vector based on a preset judgment threshold value to generate a corresponding binarization feature matrix; performing logic AND operation on the optical binarization feature matrix and the laser binarization feature matrix to obtain a feature matrix, wherein if the element value in the feature matrix is 1, the optical sensor and the laser sensor both judge that the current detection unit is a suspected greasy dirt unit; and calculating the proportion of the suspected greasy dirt units to the total number of the suspected greasy dirt units judged by at least one sensor, and obtaining the overlapping rate of the feature matrix.
5. The method for detecting the greasy dirt based on the multi-sensor-carried machine dog according to claim 1, wherein the greasy dirt identification is performed on the overlapping rate based on a preset threshold, specifically: if the overlapping rate exceeds a preset first threshold value, judging that the oil stain exists; if the overlapping rate is lower than a preset second threshold value, judging that the oil stain is not generated; and triggering secondary detection if the overlapping rate is between the first threshold value and the second threshold value.
6. The method for detecting the greasy dirt based on the carried multi-sensor machine dog according to claim 2, wherein the method is characterized in that whether the greasy dirt position is micro-leaked or not is judged based on the preprocessed multi-sensor data and the preprocessed pose data, and specifically comprises the following steps: calculating the change rate of the absorption peak intensity of the laser spectrum data of the oil stain position; When the continuous multi-frame change rate of the absorption peak intensity exceeds a set leakage threshold value, determining that micro leakage occurs, and recording a current detection point; and calculating the coordinates of the leakage points based on the preprocessed pose data of the plurality of adjacent detection points of the current detection point.
7. An oil stain detection system based on a machine dog carrying multiple sensors, the system comprising: The data acquisition module is used for controlling the machine dog carrying the multi-sensor to move in the equipment to be subjected to oil stain detection according to a preset track, acquiring multi-sensor data and pose data of the machine dog, and preprocessing the multi-sensor data and the pose data; The oil stain recognition module is used for carrying out feature extraction on the preprocessed multi-sensor data and constructing a feature matrix based on the extracted features, obtaining the overlapping rate of the feature matrix based on the duty ratio of suspected oil stain units in the feature matrix, carrying out oil stain recognition on the overlapping rate based on a preset threshold, and recording the oil stain position; and the leakage point positioning module is used for judging whether the oil stain position is subjected to micro leakage or not based on the preprocessed multi-sensor data and the preprocessed pose data, and calculating the leakage point coordinates if the oil stain position is subjected to micro leakage.
8. The multi-sensor robot dog-based oil stain detection system of claim 7, further comprising the temperature calibration module performing temperature compensation on data collected by the optical sensors and the laser sensors in the multi-sensors through a PID algorithm.
9. The greasy dirt detection system based on the multi-sensor robot dog according to claim 7, further comprising a man-machine interaction terminal for receiving and displaying the greasy dirt identification and the leakage point coordinates and sending out early warning information.
10. The multi-sensor machine dog-based oil stain detection system according to claim 7, further comprising a mechanical connection module for fixedly connecting the multi-sensors and adapting to different models of machine dogs.

Description

Greasy dirt detection method and system based on carrying multi-sensor machine dog Technical Field The invention relates to the technical field of machine dogs and monitoring, in particular to an oil stain detection method and system based on a machine dog carrying multiple sensors. Background Oil stain leakage detection is a key link for guaranteeing safe operation of equipment and avoiding environmental pollution and economic loss. Along with the improvement of the industrial automation level, higher requirements are put forward on the precision, response speed and scene suitability of the oil stain detection, but the conventional oil stain detection technology still has a plurality of limitations, and the actual application requirements are difficult to meet. The Chinese patent application with publication number of CN120747620A discloses a tunnel structure water leakage identification method for machine dog detection. The technical scheme includes that safety of a picture shot by a patrol robot dog is guaranteed through an environment sensing algorithm based on fusion of laser radar point clouds and camera type laser point clouds, then moving shooting is carried out on a tunnel structure water leakage condition by means of patrol mobile robot dog equipment, data preprocessing is carried out on images, finally, characteristics of water leakage identification data connected to the inside of a tunnel are built, missing conditions of water leakage are identified by means of an improved Yolov8+ HorNet +DASI+ CBAM model, but the technical scheme relies on visual data fused by the laser radar point clouds and the camera type laser point clouds, multi-sensor fusion data and machine dog pose data are not introduced, multi-dimensional characteristic depiction of detection targets is lacked, detection targets possibly accompanied with physical attribute changes such as materials and humidity are easy to recognize and miss, meanwhile, recognition is carried out on the basis of two-dimensional pictures, influence of pose data of carriers on the water leakage positions is not considered, only qualitative recognition of water leakage areas can be achieved, functions such as position calculation and micro-leakage judgment are not involved, accurate positioning of the water leakage positions in the equipment cannot be met, and the technical scheme cannot be controlled in a complex and the requirements of the inside of the equipment cannot be met are met. Disclosure of Invention In order to solve the problems in the prior art, the invention provides a method and a system for detecting greasy dirt based on a machine dog carrying multiple sensors. The technical scheme of the invention is as follows: in one aspect, the invention provides an oil stain detection method based on a machine dog carrying multiple sensors, which comprises the following steps: The method comprises the steps of controlling a machine dog carrying a plurality of sensors to move in equipment to be subjected to oil stain detection in a preset track, collecting multi-sensor data and pose data of the machine dog, and preprocessing the multi-sensor data and the pose data; Extracting features of the preprocessed multi-sensor data, and constructing a feature matrix based on the extracted features, obtaining the overlapping rate of the feature matrix based on the duty ratio of suspected greasy dirt units in the feature matrix, identifying greasy dirt based on a preset threshold value, and recording the greasy dirt position; and judging whether the oil stain position is subjected to micro leakage or not based on the preprocessed multi-sensor data and the preprocessed pose data, and if the oil stain position is subjected to micro leakage, calculating the coordinates of leakage points. Preferably, the multi-sensor includes an optical sensor, a laser sensor, and a temperature sensor, wherein: The optical sensor comprises a visible light camera for acquiring visible light images and a short-wave infrared sensor for acquiring infrared temperature information; The laser sensor is used for measuring laser spectrum data of the preset gas; The temperature sensor comprises a platinum resistance temperature sensor for collecting the ambient temperature, and the collected ambient temperature is compensated by utilizing a PID algorithm. Preferably, the preprocessing is performed on the multi-sensor data, specifically: gaussian noise reduction and edge enhancement are carried out on the visible light image, temperature normalization is carried out on infrared temperature information, and baseline correction is carried out on laser spectrum data; The preprocessing of pose data is specifically to correct the pose of the machine dog by adopting a Kalman filtering algorithm. Preferably, the feature extraction is performed on the preprocessed multi-sensor data, and a feature matrix is constructed based on the extracted features, specifically: Carrying out space coordinate alignm