CN-121979216-A - Multi-mode sensing inspection robot for power distribution room based on visual navigation

CN121979216ACN 121979216 ACN121979216 ACN 121979216ACN-121979216-A

Abstract

The invention discloses a power distribution room multi-mode sensing inspection robot based on visual navigation, which comprises a mobile platform, a visual navigation module, a multi-mode sensing module and a control module, wherein the visual navigation module is arranged on the mobile platform and used for collecting power distribution room environment visual and distance information and realizing robot positioning and autonomous path planning, the multi-mode sensing module is arranged on the mobile platform and used for synchronously collecting power distribution room equipment state and environment multi-dimensional data, and the control module is respectively and electrically connected with the mobile platform, the visual navigation module and the multi-mode sensing module and used for receiving positioning path information and sensing data and controlling movement of the mobile platform. The invention has the advantages that a visual navigation scheme of combining a binocular camera with a laser radar is adopted, a multi-mode sensing module integrates multi-dimensional sensing units such as infrared sensing units, sound sensing units, gas sensing units, temperature sensing units and humidity sensing units, the omnibearing detection of equipment states and environments is realized, the fault detection omission caused by a single sensing mode is avoided, and a crawler-type moving mechanism is matched with a damping component.

Inventors

XU KAIXI
Gong Zezhang
FANG DONG
SHUAI YONGSHENG
XIA XIANG
He Ruolan

Assignees

国网湖北省电力有限公司孝感供电公司

Dates

Publication Date: 20260505
Application Date: 20260203

Claims (10)

1. A multi-mode sensing inspection robot for a power distribution room based on visual navigation is characterized by comprising a mobile platform; The visual navigation module is arranged on the mobile platform and used for collecting the environmental visual and distance information of the power distribution room and realizing the positioning and autonomous path planning of the robot; The multi-mode sensing module is arranged on the mobile platform and used for synchronously collecting the state of the power distribution room equipment and the environmental multi-dimensional data; The control module is respectively and electrically connected with the mobile platform, the visual navigation module and the multi-mode sensing module and is used for receiving positioning path information and sensing data, controlling the motion of the mobile platform, and carrying out fusion processing and anomaly detection on the sensing data; And the communication module is electrically connected with the control module and is used for transmitting the processed inspection data and the processed alarm information to a remote monitoring center.
2. The multi-mode sensing inspection robot for the power distribution room based on visual navigation according to claim 1, wherein the visual navigation module comprises a binocular camera, a laser radar and a navigation processor, the binocular camera collects environment image information, the laser radar collects environment distance information, the navigation processor constructs a three-dimensional map of the environment of the power distribution room through a SLAM algorithm, and real-time positioning and path planning are achieved based on the three-dimensional map.
3. The multi-mode sensing inspection robot of a power distribution room based on visual navigation according to claim 1, wherein the SLAM algorithm comprises the following steps: S1, ORB feature extraction and descriptor generation; setting image pixel point The gray value is To Selecting 16 sampling points on a circle with radius of 3 pixels as the center, if continuous exists The gray value of each point satisfies , ; Wherein the method comprises the steps of The sampling point is t, the gray threshold value is t, and the corner point is p; s2, inter-frame feature matching and outlier rejection, and framing Feature point descriptor Frame of Feature point descriptor Matching cost: ; Is an exclusive OR operation, wherein Representing a hamming distance; an element indicating a kth position in the data at the time t; the element corresponding to the kth position in the data at the t-1 time; s3, estimating the gesture, solving an essential matrix E, decomposing to obtain (R, t), and correcting the scale by the laser radar; interior point matching pairs satisfy the essential matrix constraint , ; Wherein the method comprises the steps of Is pixel coordinates; matching point coordinates; Is an antisymmetric matrix of translation vectors; in order to be a matrix to be decomposed, Is a left singular vector matrix; the transposition of the right singular vector matrix; Singular value diagonal matrix, the diagonal elements being the singular values of E Wherein the method comprises the steps of The rotation matrix is adopted, and t is a translation vector; ; s4, three-dimensional map construction and beam method adjustment optimization based on adjacent frame posture Triangularizing to obtain three-dimensional coordinates of the feature points; : ; Wherein the method comprises the steps of Focal length of the camera; is the main point of the camera; Coordinates of points in three-dimensional space; A rotation matrix element which is an external parameter of the camera; The translational component of the camera's external parameters, The pixel coordinates of the two-dimensional image which are finally output are obtained; S5, closed loop detection and map correction are carried out, a key position feature library of the power distribution room is prestored, and similarity between a current frame and a feature library frame is calculated: , Is a dictionary of word bags, The word frequency-inverse document frequency weight; representing text And (3) with Similarity of (2); Is a word In the target text TF-IDF value of (a); Is a word In the reference text TF-IDF value of (b).
4. The multi-mode sensing inspection robot for the power distribution room based on visual navigation according to claim 1, wherein the multi-mode sensing module comprises an infrared thermal image sensing unit, a sound sensing unit, a gas sensing unit, a temperature and humidity sensing unit and a synchronous control unit, the infrared thermal image sensing unit is used for detecting surface temperature distribution of power distribution equipment, the sound sensing unit is used for collecting equipment operation abnormal sound signals, the gas sensing unit is used for detecting gas concentrations of SF6, O2, CO and the like, the temperature and humidity sensing unit is used for collecting environmental temperature and humidity data, and the synchronous control unit is used for controlling all sensing units to synchronously collect data.
5. The visual navigation-based multi-mode sensing inspection robot for a power distribution room according to claim 1, wherein the algorithm of the infrared thermal image sensing unit comprises the following steps: S11, preprocessing original thermal image data, median filtering to reduce noise, and setting an original pixel matrix of the thermal image as The median filtering output is: mapping the pixel gray value to the [0,255] interval by adopting linear stretching, wherein the formula is as follows: ; Wherein the method comprises the steps of Is the minimum gray value of a single-frame thermal image, For the maximum gray value of a single-frame thermal image, the hot spot area of the equipment is ensured to be clearly distinguished Is the pixel value of the filtered image at pixel coordinates (m, n); pixel values of the original image at coordinates m+i, n+j; Representation taking A3 x 3 neighborhood that is the center; pixel values at pixel coordinates (m, n) for the gray stretched image; S12, gray level-temperature calibration conversion, wherein the gray level value output by the infrared sensor and the target temperature are in a nonlinear relation, gray and temperature mapping are realized by adopting quadratic polynomial fitting based on standard blackbody furnace calibration data: wherein Is a pixel Corresponding actual temperature; Is a calibration coefficient; S13, segmenting the area of the power distribution room equipment, separating target equipment such as a power distribution cabinet and a switch from the environment background by adopting an adaptive threshold segmentation algorithm, calculating an adaptive threshold, and segmenting the threshold Wherein Is a locally adaptive threshold at pixel coordinates (m, n); In the case of a pixel which is a pixel, The average temperature of the mixture is calculated, Is the variance; for adjusting the coefficients; s14, extracting temperature characteristics, namely extracting key temperature characteristics aiming at the segmented equipment area, and providing input for abnormality detection.
6. The multi-mode sensing inspection robot for the power distribution room based on visual navigation according to claim 1, wherein the control module comprises a main controller and a data processing unit, the data processing unit adopts a Kalman filtering algorithm to conduct fusion processing on multi-dimensional data output by the multi-mode sensing module, data noise interference is eliminated, accuracy of sensing data is improved, an anomaly detection model is built in the main controller and is obtained based on training of a random forest machine learning algorithm, and abnormal states such as equipment overtemperature, gas leakage, abnormal operation noise, excessive temperature and humidity are identified through comparison of the fused data and a preset threshold value, and alarm information including abnormal positions, types and degrees is generated.
7. The multi-mode sensing inspection robot for the power distribution room based on visual navigation according to claim 1, wherein the moving platform adopts a crawler-type moving mechanism and comprises a driving motor, an anti-skid crawler, a damping component and a bearing chassis, wherein the damping component is of a spring damping structure and is used for buffering vibration generated by uneven ground of the power distribution room, and anti-skid lines are arranged on the surface of the anti-skid crawler to improve the traffic capacity of complex ground.
8. The multi-mode sensing inspection robot for the power distribution room based on visual navigation according to claim 2, wherein the navigation processor is further provided with a visual feature matching unit, the visual feature matching unit extracts corner points and edge features in an environment image, and the corner points and the edge features are compared with a pre-stored key position feature library of the power distribution room, so that the positioning error of an SLAM algorithm is corrected, and the positioning accuracy of the robot is less than or equal to 5cm.
9. The multi-mode sensing inspection robot for the power distribution room based on visual navigation according to claim 3, wherein the gas sensing unit is an electrochemical sensor array, the electrochemical sensor array comprises an SF6 sensor, an O2 sensor and a CO sensor, the detection precision is +/-10 ppm, +/-0.5% VOL and +/-5 ppm, and the response time is less than or equal to 3s.
10. The visual navigation-based multi-mode sensing inspection robot for a power distribution room is characterized in that the communication module adopts a wired communication and wireless communication redundancy design, the wireless communication module supports 5G and Wi-Fi dual-mode switching, the wired communication module is provided with an Ethernet interface, a data encryption unit is further arranged in the communication module, and an AES-256 encryption algorithm is adopted to encrypt transmission data.

Description

Multi-mode sensing inspection robot for power distribution room based on visual navigation Technical Field The invention relates to the technical field of power equipment inspection, in particular to a power distribution room multi-mode sensing inspection robot based on visual navigation. Background The power distribution room is taken as a key component of the power system, and the running state of equipment directly influences the stability and safety of power supply. The traditional power distribution room inspection is mainly finished manually, inspection personnel need to contact high-voltage equipment in a close range and face safety risks such as electric shock and gas poisoning, meanwhile, the manual inspection efficiency is low, faults are easy to miss due to influences of subjective factors, and 24-hour uninterrupted inspection cannot be achieved. The prior art has the defects that 1) a navigation mode is single, magnetic track navigation or single vision navigation is adopted, the magnetic track navigation needs to be preset, flexibility is poor, the single vision navigation has low positioning precision in the light shortage or complex environment, 2) a sensing mode is limited, and the system has temperature or image acquisition functions, cannot comprehensively detect multiple types of faults such as abnormal sound and gas leakage of equipment, 3) the data processing capability is weak, a data fusion technology is not adopted, the sensing data error is large, the abnormal detection accuracy is low, 4) the environment adaptability is poor, a moving mechanism is difficult to adapt to the uneven ground of a power distribution room, and communication is easy to be disturbed. Therefore, there is a need for a multi-mode sensing inspection robot for a power distribution room based on visual navigation. Disclosure of Invention Aiming at the defects in the prior art, the invention aims to provide a multi-mode sensing inspection robot for a power distribution room based on visual navigation. The multi-mode sensing inspection robot for the power distribution room based on visual navigation comprises a mobile platform, a visual navigation module, a power distribution room control module and a power distribution room control module, wherein the visual navigation module is arranged on the mobile platform and used for collecting visual and distance information of the environment of the power distribution room and realizing the positioning and autonomous path planning of the robot; The multi-mode sensing module is arranged on the mobile platform and used for synchronously collecting the state of the power distribution room equipment and the environmental multi-dimensional data; The control module is respectively and electrically connected with the mobile platform, the visual navigation module and the multi-mode sensing module and is used for receiving positioning path information and sensing data, controlling the motion of the mobile platform, and carrying out fusion processing and anomaly detection on the sensing data; And the communication module is electrically connected with the control module and is used for transmitting the processed inspection data and the processed alarm information to a remote monitoring center. The visual navigation module comprises a binocular camera, a laser radar and a navigation processor, wherein the binocular camera acquires environment image information, the laser radar acquires environment distance information, and the navigation processor constructs a three-dimensional map of the power distribution room environment through an SLAM algorithm and realizes real-time positioning and path planning based on the three-dimensional map. The SLAM algorithm comprises the following steps: S1, ORB feature extraction and descriptor generation; setting image pixel point The gray value isToSelecting 16 sampling points on a circle with radius of 3 pixels as the center, if continuous existsThe gray value of each point satisfies,; Wherein the method comprises the steps ofThe sampling point is t, the gray threshold value is t, and the corner point is p; s2, inter-frame feature matching and outlier rejection, and framing Feature point descriptorFrame ofFeature point descriptorMatching cost:; to be exclusive-or operation S3, estimating the gesture, solving an essential matrix E, decomposing to obtain (R, t), and correcting the scale by the laser radar; interior point matching pairs satisfy the essential matrix constraint ; Wherein the method comprises the steps ofIs pixel coordinates; matching point coordinates; Is an antisymmetric matrix of translation vectors; Wherein the method comprises the steps of The rotation matrix is adopted, and t is a translation vector;; s4, three-dimensional map construction and beam method adjustment (BA) optimization, and based on adjacent frame posture Triangularizing to obtain three-dimensional coordinates of the feature points;: ; Wherein the method comprises t