Search

CN-122015967-A - Multi-mode fusion-based unmanned electric equipment inspection vehicle and state evaluation method

CN122015967ACN 122015967 ACN122015967 ACN 122015967ACN-122015967-A

Abstract

The invention provides an unmanned inspection system and a state evaluation method for substation electrical equipment, which take an autonomous mobile inspection vehicle as a platform and integrate infrared, voiceprint, ultraviolet and visible light multi-mode perception. The multi-mode state evaluation method based on the equipment structure space mapping, abnormal region consistency check and mode reliability self-adaptive correction realizes one-to-one correspondence between the health state result and the specific equipment structure part, automatically reduces the fusion weight under the single-mode distortion condition, eliminates the misjudgment caused by the dislocation of the multi-mode abnormal region, and remarkably reduces the misinformation rate and the state evaluation fluctuation.

Inventors

  • GONG YINYING
  • LIU LI

Assignees

  • 北京北创芯通科技有限公司

Dates

Publication Date
20260512
Application Date
20260209

Claims (9)

  1. 1. The unmanned inspection vehicle is characterized by comprising a multi-mode sensing acquisition unit and a vehicle-mounted calculation and control unit, The multi-mode sensing acquisition unit comprises a thermal infrared imager, a voiceprint imaging sensor, an ultraviolet imaging camera and a high-definition visible light camera; The vehicle-mounted calculation and control unit preprocesses the acquired data and extracts corresponding characteristics, namely an infrared temperature characteristic vector (F ir ), a voiceprint frequency spectrum characteristic vector (F ac ), an ultraviolet discharge characteristic vector (F uv ) and an appearance defect characteristic vector (F vis ) respectively Inputting the characteristics into a multi-mode fusion state evaluation model to obtain a unified fusion characteristic representation: The method comprises the steps of obtaining a model, wherein alpha 1, alpha 2, alpha 3 and alpha 4 represent adaptive modal attention weights, the weights are dynamically calculated by signal-to-noise ratios or confidence degrees of various modal features, W fusion is a fusion transformation matrix used for mapping spliced high-dimensional features to unified fusion feature space dimensions M, b fusion is a bias vector, and sigma () is a nonlinear activation function so as to enhance the expression capability of the model on complex fault modes. Based on the fusion features, the state assessment model outputs a device health state vector.
  2. 2. The inspection vehicle of claim 1, wherein the thermal infrared imager is configured to collect temperature field distribution on a surface of an electrical device, the voiceprint imaging sensor is configured to collect acoustic signals generated by partial discharge, the ultraviolet imaging camera is configured to detect ultraviolet radiation characteristics generated by corona discharge, and the high-definition visible light camera is configured to collect an appearance state image of the device.
  3. 3. The inspection vehicle of claim 1, wherein the infrared temperature feature vector, the voiceprint spectral feature vector, the ultraviolet discharge feature vector, and the appearance defect feature vector are all provided with uniform time stamp and pose information.
  4. 4. The inspection vehicle of claim 1, wherein the multi-modal fusion includes spatial correlation of infrared temperature anomaly regions and visible light images, correspondence of voiceprint anomaly signals to device structure locations, and joint determination of ultraviolet discharge characteristics and device appearance defects.
  5. 5. The inspection vehicle according to claim 1, wherein the on-board computing and control unit performs the steps of 301 spatial modeling of the equipment to be inspected, S302 positioning of the multi-modal anomaly equipment, S303 spatial consistency constraint meter based on multi-modal data.
  6. 6. A method for evaluating the status of an electrical device implemented with the inspection vehicle according to any one of claims 1 to 5, characterized in that it comprises the steps of: S1, the unmanned inspection vehicle automatically runs in a transformer substation according to preset inspection route planning information; s2, after the unmanned inspection vehicle reaches a designated equipment area, automatically switching to a low-speed inspection mode or a fixed-point parking mode, and carrying out joint acquisition on the same equipment by using the multi-mode sensing acquisition unit under a unified time reference to obtain multi-mode data; s3, based on multi-mode data, the vehicle-mounted calculation and control unit completes data preprocessing, feature extraction and fusion analysis in real time; The step S3 specifically comprises the steps of modeling the structural space of equipment to be inspected, locating multi-mode abnormal equipment, calculating space consistency constraint, adaptively correcting mode credibility, and evaluating constraint fusion state, wherein the step S301 comprises the step S302 comprises the step S303 of carrying out spatial consistency constraint calculation, the step S304 comprises the step S305 of carrying out adaptive correction on mode credibility.
  7. 7. The method of claim 6, wherein the thermal infrared imager is configured to collect a temperature field distribution of a surface of an electrical device, the voiceprint imaging sensor is configured to collect acoustic signals generated by partial discharge, the ultraviolet imaging camera is configured to detect ultraviolet radiation characteristics generated by corona discharge, and the high definition visible camera is configured to collect an appearance state image of the device.
  8. 8. The method of claim 6, wherein the infrared temperature feature vector, the voiceprint spectral feature vector, the ultraviolet discharge feature vector, and the appearance defect feature vector are each with uniform time stamp and pose information.
  9. 9. The method of claim 6, wherein the multi-modal fusion includes spatial correlation of infrared temperature anomaly regions with visible light images, correspondence of voiceprint anomaly signals with device structure locations, and joint determination of ultraviolet discharge characteristics and device appearance defects.

Description

Multi-mode fusion-based unmanned electric equipment inspection vehicle and state evaluation method Technical Field The invention relates to the technical field of intelligent operation and maintenance of an electric power system and inspection of robots, in particular to an unmanned inspection system and a state evaluation method for electric equipment of a transformer substation, which are integrated with infrared, voiceprint, ultraviolet and visible light multi-mode perception by taking an autonomous mobile inspection vehicle as a platform. Background In the long-term operation process of electrical equipment such as transformers, circuit breakers, isolating switches and the like in a transformer substation, fault hidden dangers such as overheating, partial discharge, insulation degradation, appearance damage and the like are easily generated due to load change, environmental factors or material aging. The existing substation inspection method mainly comprises manual inspection or single sensor auxiliary inspection, and has the defects that the environment of a substation is complex, high-pressure risk exists, high-frequency and full-coverage detection is difficult to realize by manual inspection, misjudgment or missed judgment is easy to occur only by means of infrared, visible light or acoustic sensing, real health states of equipment are difficult to accurately reflect, space deviation exists on observation positions of the same equipment by different mode sensors (infrared, voiceprint, ultraviolet and visible light), and single mode abnormal probability is high due to local environmental noise (wind noise, electromagnetic interference and reflected light), so that fusion results are interfered. Therefore, it is necessary to provide a comprehensive inspection system which takes an unmanned inspection vehicle as a carrier, takes multi-mode sensing as a means and takes state evaluation as a target. Disclosure of Invention In view of the problems existing in the prior art, the invention provides an unmanned inspection vehicle, which comprises a multi-mode sensing acquisition unit and a vehicle-mounted calculation and control unit, The multi-mode sensing acquisition unit comprises a thermal infrared imager, a voiceprint imaging sensor, an ultraviolet imaging camera and a high-definition visible light camera; The vehicle-mounted calculation and control unit preprocesses the acquired data and extracts corresponding characteristics, namely an infrared temperature characteristic vector (Fir), a voiceprint frequency spectrum characteristic vector (Fac), an ultraviolet discharge characteristic vector (Fuv) and an appearance defect characteristic vector (Fvis) respectively Inputting the characteristics into a multi-mode fusion state evaluation model to obtain a unified fusion characteristic representation: Wherein, alpha 1, alpha 2, alpha 3 and alpha 4 represent self-adaptive modal attention weights which are dynamically calculated by the signal-to-noise ratio or confidence coefficient of each modal feature, W fusion is a fusion transformation matrix which is used for mapping the spliced high-dimensional features to a unified fusion feature space dimension M, b fusion is a bias vector; is a nonlinear activation function to enhance the expressive power of the model on complex failure modes. Based on the fusion features, the state assessment model outputs a device health state vector. The infrared thermal imager is used for collecting temperature field distribution on the surface of the electrical equipment, the voiceprint imaging sensor is used for collecting acoustic signals generated by partial discharge, the ultraviolet imaging camera is used for detecting ultraviolet radiation characteristics generated by corona discharge, and the high-definition visible light camera is used for collecting appearance state images of the equipment. Preferably, the infrared temperature feature vector, the voiceprint spectrum feature vector, the ultraviolet discharge feature vector and the appearance defect feature vector all carry unified timestamp and pose information. Preferably, the multi-mode fusion comprises spatial association of infrared temperature anomaly regions and visible light images, correspondence of voiceprint anomaly signals and equipment structure positions, and joint judgment of ultraviolet discharge characteristics and equipment appearance defects. Preferably, the vehicle-mounted computing and control unit performs the following steps of 301, modeling the structural space of equipment to be patrolled and examined, 302, positioning abnormal equipment in multiple modes, and 303, a space consistency constraint meter. The invention also provides an electrical equipment state evaluation method implemented by using the inspection vehicle, which comprises the following steps: S1, the unmanned inspection vehicle automatically runs in a transformer substation according to preset inspection route planning information;