Petrochemical tank field multi-mode detection anti-explosion method and petrochemical tank field inspection robot

Abstract

The invention provides a petrochemical tank field multi-mode detection explosion-proof method and a petrochemical tank field inspection robot, and belongs to the technical field of industrial safety detection. The method comprises the steps of performing multi-modal detection on a target petrochemical tank field according to a preset inspection path to obtain multi-modal detection data, performing item-by-item analysis on the multi-modal detection data, mapping a corresponding analysis result to a time-space unified coordinate system pre-constructed according to a timestamp calibrated by the multi-modal detection data, and performing multi-modal detection on the target petrochemical tank field according to the time-space unified coordinate system. According to the position of the analysis result in the space-time unified coordinate system, carrying out cross-cooperation verification of the recheck analysis result, judging whether the cross-cooperation verification is established or not, if the cross-cooperation verification is established, calculating an abnormal index association weight coefficient, and according to a calculation result, judging an association degree; and outputting a corresponding explosion-proof disposal plan according to the correlation degree. Early recognition of tiny leakage of hazardous gas is achieved, the judgment precision of gas-liquid two-phase leakage characteristics is improved, the false alarm rate is reduced, and maintenance resources are saved.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial safety detection, and relates to a multi-modal detection explosion-proof method for a petrochemical tank area and a petrochemical tank area inspection robot. Background Art

[0002] Petrochemical tank farms, containing explosive and other hazardous gases, present safety risks and require continuous monitoring. However, existing inspection equipment used in these areas typically utilizes a single sensor. This makes it impossible to simultaneously capture optical imaging, infrared temperature field distribution, and ultrasonic spectrum characteristics of gas leaks, nor is it possible to establish dynamic correlation models for multi-physics field data. These issues hinder the early identification of small leaks and the accuracy of gas-liquid two-phase leakage characteristics. This not only significantly increases the frequency of manual re-inspections in high-risk areas, but also wastes maintenance resources due to high false alarm rates, severely hindering the advancement of unmanned inspections across petrochemical facilities. Summary of the Invention

[0003] The purpose of the present invention is to address the above problems in the existing technology and propose a multi-modal detection explosion-proof method for petrochemical tank areas.

[0004] The purpose of the present invention can be achieved through the following technical solutions: A multi-modal detection explosion-proof method for a petrochemical tank area, comprising: According to the preset inspection route, multimodal inspection is carried out on the target petrochemical tank area to obtain multimodal inspection data; Analyzing the multimodal detection data item by item, and mapping the corresponding analysis results to a unified spatiotemporal coordinate system constructed in advance based on the timestamps calibrated for the multimodal detection data; Performing cross-cooperative verification of the re-examination analysis results according to the position of the analysis results in the unified time-space coordinate system, and determining whether the cross-cooperative verification is established; If the cross-cooperative verification is established, the abnormal indicator correlation weight coefficient is calculated, the correlation degree is determined according to the calculation result, and the corresponding explosion-proof disposal plan is output according to the correlation degree.

[0005] As an optional embodiment of the present invention, multimodal detection is performed on the target petrochemical tank farm to obtain multimodal detection data, including: The multimodal detection data includes visible light images, gas distribution data, and acoustic wave signals in a preset frequency band, wherein the gas distribution data includes spectral absorption data and plume distribution data; Using dual visible light cameras to collect visible light images of the target petrochemical tank area; An optical gas imager is used to collect spectral absorption data and plume distribution data of the target petrochemical tank area; Ultrasonic array sensors are used to collect sound wave signals in a preset frequency band.

[0006] As an optional embodiment of the present invention, cross-cooperative verification of the re-examination analysis results is performed based on the position of the analysis results in the unified spatiotemporal coordinate system to determine whether the cross-cooperative verification is established, including: If an analysis result is abnormal, rechecking and analyzing the multimodal detection data based on the position of the abnormality in the unified time and space coordinate system to obtain a recheck analysis result; Calculating the correlation between the analysis result and the re-inspection analysis result, determining whether the correlation is successful based on whether the correlation meets a preset correlation standard, and setting a graded corresponding measure based on the successfully correlated multimodal detection data; If the analysis result and the re-analysis result are both successfully associated, it is determined that the cross-cooperative validation is established.

[0007] As an optional embodiment of the present invention, analyzing the multimodal detection data item by item includes: analyzing whether the visible light image contains an abnormal area; if so, the visible light image is abnormal, and extracting an edge contour of the abnormal area; The position of the edge contour in the unified time-space coordinate system is marked, and based on the pre-constructed three-dimensional digital model of the petrochemical tank area, the detection point position within the preset range is locked according to the mark.

[0008] As an optional embodiment of the present invention, the item-by-item analysis of the multimodal detection data further includes: Analyzing the gas distribution data; if the gas distribution data contains a gas concentration higher than a historical normal concentration range, the gas distribution data is considered abnormal, and the corresponding gas is used as a target gas; Analyzing the spectral absorption data of the target gas to identify whether the target gas is a leak-proof gas; The plume distribution data of the target gas is analyzed, and the diffusion parameters of the target gas are calculated.

[0009] As an optional embodiment of the present invention, if an analysis result is abnormal, the method further includes: Extracting target data from the multimodal detection data to construct a multimodal association matrix; The target data association is analyzed, and a data consistency index of the multimodal association matrix is ​​updated, wherein updating the data consistency index includes updating time synchronization, spatial consistency, and intensity correlation.

[0010] As an optional embodiment of the present invention, performing association analysis on the target data and updating the data consistency index of the multimodal association matrix includes: Calculating the time deviation between target data, wherein the time synchronization is determined according to the time deviation; Calculating the Euclidean distance between the center point of the gas diffusion area and the gas source positioning point according to the target data, wherein the spatial consistency is determined according to the corresponding value of the Euclidean distance; Dynamically comparing the changing trends of the gas distribution data and the spectrum of the acoustic wave signal to determine whether they are positively correlated or negatively correlated, wherein the intensity correlation is determined based on the result of the determination; The data consistency index is updated according to the time synchronization, spatial consistency and intensity correlation after the corresponding values ​​are taken.

[0011] The present invention also provides a petrochemical tank area inspection robot, which is applied to the above-mentioned petrochemical tank area multimodal detection explosion-proof method, comprising: A vehicle body, wherein two rollers are symmetrically provided on both sides of the vehicle body, and the rollers are used to connect with the load-bearing cables; a detection unit comprising a dual visible light camera and an ultrasonic array sensor fixedly mounted on the bottom of the vehicle body, and an optical gas imager fixedly mounted on the front end of the vehicle body along the forward direction of the vehicle body; A drive connection portion is fixedly arranged on the top of the vehicle body and is used for connecting to the traction cable.

[0012] As an optional embodiment of the present invention, the outer peripheral surface of the roller is provided with a groove, and the groove includes two symmetrically arranged abutment surfaces, and the abutment surfaces are used to abut against the load-bearing cable. Along the circumferential direction of the roller, a plurality of magnets are spaced apart on the abutment surfaces.

[0013] The present invention also provides a petrochemical tank area multi-modal detection explosion-proof system, comprising: The data acquisition module is used to perform multimodal detection on the target petrochemical tank area according to the preset inspection path and obtain multimodal detection data; A mapping module, configured to analyze the multimodal detection data item by item and map the corresponding analysis results to a unified spatiotemporal coordinate system constructed in advance based on the timestamps calibrated for the multimodal detection data; A cross-cooperative verification module is used to perform cross-cooperative verification of the re-examination analysis results according to the position of the analysis results in the unified time-space coordinate system, and determine whether the cross-cooperative verification is established; The output data module is used to calculate the abnormal indicator correlation weight coefficient if the cross-cooperative verification is established, determine the correlation degree according to the calculation result, and output the corresponding explosion-proof disposal plan according to the correlation degree.

[0014] The present invention further provides an electronic device, comprising: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to implement the above-mentioned petrochemical tank area multimodal detection explosion prevention method when executing the executable instructions.

[0015] Compared with the existing technology, the present invention achieves early identification of tiny leaks of hazardous gases by synchronously acquiring multimodal detection data and dynamically correlating and analyzing them, improves the accuracy of distinguishing gas-liquid two-phase leakage characteristics, significantly reduces the frequency of manual re-inspections in high-risk areas, reduces false alarm rates, saves maintenance resources, and promotes unmanned inspections of petrochemical tank areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flow chart of a multi-modal detection explosion-proof method for a petrochemical tank area according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the three-dimensional structure of the petrochemical tank area inspection robot; Figure 3 yes Figure 2 Schematic diagram of the three-dimensional structure after adding the anti-slip plate; Figure 4 It is a schematic diagram of the three-dimensional structure of the roller; Figure 5 This is a block diagram of a multi-modal detection explosion-proof system for a petrochemical tank area according to an embodiment of the present invention; In the figure, the vehicle body 100; the roller 101; the anti-slip plate 102; the magnet 103; the drive connection part 104; the traction cable 200; the load-bearing cable 201; the dual visible light camera 300; the ultrasonic array sensor 301; the optical gas imager 302; the radio frequency read / write sensor 303; the petrochemical tank area multimodal detection explosion-proof system 10; the data acquisition module 11; the mapping module 12; the cross-collaborative verification module 13; and the output data module 14. DETAILED DESCRIPTION

[0017] The following are specific embodiments of the present invention and the accompanying drawings to further describe the technical solutions of the present invention, but the present invention is not limited to these embodiments.

[0018] Example 1

[0019] Crude oil, refined oil, chemicals, etc. stored in petrochemical tank farms are mostly flammable and explosive substances. When their vapors mix with air, they may form an explosive environment, which directly threatens personal safety. Therefore, explosion protection in petrochemical tank farms is the core link of the multi-layer protection system. In this embodiment, the following methods are proposed: Figure 1 The multi-modal detection and explosion prevention method for the petrochemical tank area shown includes: S1, according to the preset inspection route, conduct multimodal inspection on the target petrochemical tank area and obtain multimodal inspection data; S2, analyzing the multimodal detection data item by item, and mapping the corresponding analysis results to a unified spatiotemporal coordinate system constructed in advance based on the timestamps calibrated for the multimodal detection data; S3, performing cross-cooperative verification of the re-examination analysis results according to the position of the analysis results in the unified spatiotemporal coordinate system, and determining whether the cross-cooperative verification is established; S4: If the cross-cooperative verification is established, calculate the abnormal indicator correlation weight coefficient, determine the correlation degree according to the calculation result, and output the corresponding explosion-proof disposal plan according to the correlation degree.

[0020] According to the distribution of various facilities in the petrochemical tank farm, an inspection route is formulated in advance, and an inspection device is used and a control signal is sent to it to perform multimodal detection of the petrochemical tank farm according to the inspection route. Among them, the inspection device can adopt an explosion-proof robot, etc., and use its flexible and stable structure to smoothly complete the inspection according to the inspection route, and transmit the detected multimodal detection data for analysis. Multimodal data is information composed of multiple different types of data modes, which can be text, images, audio, video, sensor data, time series data, etc. In this embodiment, the inspection data of the petrochemical tank farm by the inspection device is multimodal, and all the detection data obtained are used as multimodal detection data, and then the multimodal detection data is transmitted to the control center of the petrochemical tank farm for subsequent analysis.

[0021] Based on the timestamp of each modal detection data, the spatial and temporal dimensions are integrated to construct a unified spatiotemporal coordinate system, achieving time axis alignment of multimodal data. Each modal detection data is analyzed individually to determine whether it is normal or abnormal, and the analysis results are mapped to the unified spatiotemporal coordinate system based on the timestamp.

[0022] Based on the analysis results of each modal detection data, cross-collaborative verification is performed to recheck whether each modal detection data at the same petrochemical tank area is normal, abnormal, or partially abnormal. On the one hand, the real-time nature of the detection data is guaranteed. On the other hand, based on the situations of all being normal, all being abnormal, or partially being abnormal, the correlation between each modal detection data is calculated to see whether it exceeds the preset correlation standard. When it exceeds the preset correlation standard, the cross-collaborative verification is established, and the correlation weight coefficient of the abnormal indicator is further calculated. The degree of correlation is determined based on the specific calculated value, triggering a remote alarm and pushing the emergency response plan.

[0023] Preferably, multimodal detection is performed on the target petrochemical tank farm to obtain multimodal detection data, including: The multimodal detection data includes visible light images, gas distribution data, and acoustic wave signals in a preset frequency band, wherein the gas distribution data includes spectral absorption data and plume distribution data; Using dual visible light cameras to collect visible light images of the target petrochemical tank area; An optical gas imager is used to collect spectral absorption data and plume distribution data of the target petrochemical tank area; Ultrasonic array sensors are used to collect sound wave signals in a preset frequency band.

[0024] The multimodal detection data collected in this embodiment includes visible light images, spectral absorption data about the gas, plume distribution data, and acoustic wave signal data.

[0025] Dual-visible cameras are installed on the inspection device to scan the pipelines, flanges, and other facilities within the target petrochemical tank area, as well as the ground, in real time according to the inspection route. The dual-visible cameras are equipped with adaptive lighting whose brightness dynamically adjusts based on ambient illumination, ensuring clear images with a resolution of 1920×1080 even at night or in low-light petrochemical tank areas. An optical gas imager is installed on the inspection device to collect spectral absorption data from the target petrochemical tank area. The optical gas imager features multiple built-in spectral filter adjustment slots, each corresponding to a spectral filter of a different wavelength. Control signals are used to focus the corresponding spectral filter on the characteristic absorption spectrum of the target gas. The optical gas imager also uses an uncooled infrared detector to scan the gas plume distribution. The gas plume distribution data is analyzed to obtain gas concentration data and diffusion velocity parameters. An ultrasonic array sensor, consisting of multiple sensing units in a rectangular array structure, is installed on the inspection device to capture acoustic signals in the 20-40kHz frequency band.

[0026] Preferably, performing cross-cooperative verification of the re-examination analysis results according to the position of the analysis results in the unified time-space coordinate system, and judging whether the cross-cooperative verification is established, includes: If an analysis result is abnormal, rechecking the multimodal detection data based on the position of the abnormality in the unified time and space coordinate system and analyzing it to obtain a recheck analysis result; Calculating the correlation between the analysis result and the re-analysis result, and determining whether the correlation is successful based on whether the correlation meets a preset correlation standard; If the analysis result and the re-analysis result are both successfully associated, it is determined that the cross-cooperative validation is established.

[0027] After analyzing each modal detection data separately, if the analysis result of the detection data of one of the modes is abnormal, the remaining modal detection data of the coordinate position will be rechecked according to the coordinate position of the modal detection data in the unified time and space coordinate system. For example, when the analysis result of the visible light image contains an abnormal area, the hierarchical response mechanism is triggered. According to the position of the abnormality in the unified time and space coordinate system, the actual position of the abnormality in the petrochemical tank area is determined. The gas distribution data and the preset frequency band sound wave signal of the actual position are detected again in real time and analyzed to see whether they are all normal, all abnormal or partially abnormal. Then, the correlation between the two modal detection data of visible light image and gas distribution data is calculated based on whether the gas distribution data is all normal, all abnormal or partially abnormal. If the correlation meets the standard, it indicates that the possibility of suspected gas leakage is high. At this time, measures are taken to mark the coordinates of the leakage point. Based on the correlation between the visible light image and the re-inspected gas distribution data, the correlation with the re-inspected preset frequency band sound wave signal is calculated. If the correlation meets the standard, it indicates that the possibility of suspected gas leakage is even higher. When the correlation between the visible light image, the re-inspected gas distribution data, and the acoustic wave signal in the preset frequency band meets the required criteria, cross-validation is established. The correlation weight coefficient W for these three abnormality indicators is then calculated and compared with the preset threshold. In this embodiment, the threshold T is set at 0.85. If W ≥ T, a strong correlation leak event is identified, triggering a remote alarm and disseminating the emergency response plan. Among them, the hierarchical response mechanism is triggered when the analysis result contains an anomaly, and after the trigger, the re-inspection of the remaining modal detection data is started, and relevant measures are taken according to whether the correlation meets the standard. In this embodiment, the hierarchical response mechanism is that when the analysis result of one of the modal detection data contains an anomaly, the relevant measures taken at this time are to start the re-inspection of the remaining modal detection data; and based on one of the re-inspection analysis results, that is, the analysis result of the gas distribution data or the preset frequency band sound wave signal, the correlation with the analysis result containing the anomaly is calculated. If the correlation meets the standard, the relevant measures taken at this time are to automatically mark the coordinates of the leakage point; then, based on the modal detection data that meets the correlation standard, the correlation with the analysis result of the preset frequency band sound wave signal or gas distribution data is calculated. If all correlations meet the standard, the relevant measures taken at this time are to calculate the abnormality indicator correlation weight coefficient based on the updated data consistency index, that is, the updated value of time synchronization, spatial consistency and intensity correlation, to determine whether it is a strongly correlated leakage event. If so, a remote alarm is triggered and an emergency response plan is pushed. It should be noted that the measures taken in the hierarchical response mechanism can be set according to the modal type of the actual detection data and whether the correlation meets the standard.

[0028] Preferably, analyzing the multimodal detection data item by item includes: analyzing whether the visible light image contains an abnormal area; if so, the visible light image is abnormal, and extracting an edge contour of the abnormal area; The position of the edge contour in the unified time-space coordinate system is marked, and based on the pre-constructed three-dimensional digital model of the petrochemical tank area, the detection point position within the preset range is locked according to the mark.

[0029] The dual visible light cameras continuously capture images of the tank and pipeline surfaces of the target petrochemical tank farm. When an abnormal area is identified in the visible light image, such as abnormal reflective patches or wet marks in a dry area, it is determined to be a suspected leakage area, and edge enhancement processing is automatically triggered to extract the edge contour features of the abnormal area, and mark the true location coordinates of the suspected leakage area. Based on the pre-built three-dimensional digital model of the petrochemical tank farm, the detection points within the preset range are locked, such as the associated tank flanges, pipeline welds or valve connections within the range directly above the true location coordinates.

[0030] Preferably, analyzing the multimodal detection data item by item further includes: Analyzing the gas distribution data; if the gas distribution data contains a gas concentration higher than a historical normal concentration range, the gas distribution data is considered abnormal, and the corresponding gas is used as a target gas; Analyzing the spectral absorption data of the target gas to identify whether the target gas is a leak-proof gas; The plume distribution data of the target gas is analyzed, and the diffusion parameters of the target gas are calculated.

[0031] The optical gas imager continuously captures infrared images of the tank and pipeline surfaces. If it detects gas concentrations outside the historically normal range, it indicates an anomaly in the gas distribution data. The gas outside the historically normal range is identified as a target gas for further identification. Active spectral enhancement technology improves the ability to identify the molecular absorption spectral characteristics of the target gas, effectively distinguishing environmental interference signals from actual gas leak signatures. This allows the determination of whether the target gas is a containment gas, which includes flammable gases such as methane. If it is a containment gas, the target petrochemical tank area is leaking containment gas. The detected leak area is pseudo-colored, indicating the need for timely intervention. Once confirmed as a containment gas, the plume distribution data from the uncooled infrared detector is further extracted to trace the source of the containment gas back to the source. Concentration data and diffusion velocity parameters are analyzed and output. Kalman filtering is applied to the containment gas concentration data output by the optical gas imager for noise reduction, with a 5-second sliding average filter window. A concentration trend curve is generated, and periods of abnormal fluctuation exceeding 15% are marked.

[0032] In addition, when abnormal areas are found in the visible light image through analysis, the actual position of the abnormal area is focused and scanned to obtain re-inspected gas distribution data for subsequent cross-cooperative verification.

[0033] Preferably, analyzing the multimodal detection data item by item further includes: Performing a fast Fourier transform on the sound wave signal of the preset frequency band to extract a spectrum diagram of the target frequency band; Identify whether the spectrum graph contains the peak feature of anti-leakage gas. If so, the sound wave signal in the preset frequency band is abnormal, and calculate the phase difference of the sound wave signal corresponding to the peak feature received by the sensor through the beamforming algorithm to locate the sound source direction of the sound wave signal.

[0034] In this embodiment, the preset frequency band acoustic wave signal is an acoustic wave signal in the 20-40kHz frequency band, and the sampling frequency of the ultrasonic array sensor is 1MHz. The acoustic wave signal in the 20-40kHz frequency band is fast Fourier transformed, and the spectrum of the 25kHz frequency band is extracted. The spectrum is then identified to see whether it has peak characteristics of anti-leakage gas, such as whether it has peak characteristics of methane gas. The phase difference of the signals received by each sensor unit is calculated using a beamforming algorithm to achieve preliminary positioning of the sound source direction of the acoustic wave signal corresponding to the anti-leakage gas. It should be noted that the original state of the anti-leakage gas mentioned can be gas or evaporated from liquid.

[0035] It should also be noted that item-by-item analysis of multimodal detection data refers to analyzing each modality's detection data separately. The analysis process is parallel, and the analysis results can be transferred between them. This allows the analysis results of each modality's detection data to be integrated and analyzed for cross-validation. Specifically, when a visible light image contains an abnormal area, the coordinates of the actual location of the abnormal area are automatically synchronized with the optical gas imager, guiding its focused scanning. Molecular absorption spectral characteristics are then used to secondary verify whether the abnormal area contains leaks of anti-leakage gas. Ultrasonic array sensors collect acoustic signals in specific frequency bands in the abnormal area in real time, further verifying the presence of leaks by identifying peak characteristics. Alternatively, when the optical gas imager captures a gas concentration that exceeds the historical normal concentration range, the visible light image, acoustic signal, and molecular absorption spectral characteristics corresponding to the location in a unified spatiotemporal coordinate system are linked for cross-validation, generating a three-dimensional heat map that includes the leak location, leaked material type, and diffusion trend.

[0036] Based on the spatio-temporal unified coordinate system, the target data corresponds to the time point of the sudden increase in gas concentration in the multi-modal detection data, the contour position of the target gas diffusion area in the visible light image at the same moment, and the intensity change value of the ultrasonic wave signal in the corresponding period. Align these multi-modal detection data along the time axis to construct a multi-modal correlation matrix. It should be noted that the relationship between the extracted target data and the abnormal analysis result can be inclusive or exclusive. For example, the data of the sudden increase in gas concentration may belong to or not belong to the gas concentration data higher than the historical normal concentration range. This matrix adopts a sliding time window mechanism with a window width of 30 seconds, and updates the data consistency indicators: time synchronization α, spatial consistency β, and intensity correlation γ, for calculating the abnormal index correlation weight coefficient W, where W = (0.4α + 0.3β + 0.3γ). The value range of the abnormal index correlation weight coefficient W is [0, 1], and the closer its value is to 1, the stronger the collaborative correlation between the multi-modal data.

[0037] Preferably, for the correlation analysis of the target data, updating the data consistency indicators of the multi-modal correlation matrix includes: Calculating the time deviation between the target data, and the time synchronization takes corresponding values according to the time deviation; Calculating the Euclidean distance between the center point of the gas diffusion area and the gas source positioning point according to the target data, and the spatial consistency takes corresponding values according to the Euclidean distance; Dynamically comparing the change trends of the gas distribution data and the spectrum of the acoustic wave signal to determine positive or negative correlation, and the intensity correlation takes corresponding values according to the determination result; Updating the data consistency indicators according to the time synchronization, spatial consistency, and intensity correlation after corresponding value taking.

[0038] For the value taking of the time synchronization α, compare the moment of the sudden increase in gas concentration, the plume generation time, and the time of the sudden change in the ultrasonic amplitude. If the time deviation among the three is within 2 seconds, it is determined as highly time synchronized, and α takes 1; when the time deviation 2 seconds < |Δt| ≤ 6 seconds, linear deduction is performed according to the time deviation amount, and at this time α = 1 - (|Δt| - 2) / 4.

[0039] For the value taking of the spatial consistency β, it is determined by calculating the Euclidean distance between the center point of the gas diffusion area in the simultaneously acquired visible light image and the gas sound source positioning point of the ultrasonic wave; when the distance between the two is less than 0.5 meters, it is considered that the spatial positions are completely matched, and β takes 1; when the distance 0.5 meters < d ≤ 2.0 meters, β = 1 - (d - 0.5) / 1.5.

[0040] For the value of intensity correlation γ, the trend comparison method is used to dynamically compare the gas concentration change rate and the ultrasonic signal intensity increase. When the two show a trend of change in the same direction, a positive correlation score is given. If the concentration increases but the signal intensity decreases, it is judged to be negatively correlated. In this case, γ = ΔC × ΔS, where ΔC is the gas concentration change rate and ΔS is the percentage increase in ultrasonic signal intensity. When the calculated result exceeds 1.0, γ = 1, and when ΔC and ΔS change in opposite directions, γ = 0.

[0041] By applying the above method to explosion-proof detection in petrochemical tank areas, multimodal detection data can be obtained simultaneously, and dynamic correlation analysis can be performed to achieve early identification of small leaks of hazardous gases, improve the accuracy of distinguishing gas-liquid two-phase leakage characteristics, significantly reduce the frequency of manual re-inspections in high-risk areas, reduce the false alarm rate, save maintenance resources, and promote unmanned inspections in petrochemical tank areas.

[0042] Example 2 This embodiment also proposes a petrochemical tank area inspection robot, which is applied to the petrochemical tank area multimodal detection explosion-proof method described in Example 1. Figure 2-Figure 4 As shown, a petrochemical tank area inspection robot includes: The vehicle body 100 has two rollers 101 symmetrically disposed on both sides thereof. The rollers 101 are used to connect with the load-bearing cables 201 to provide support for the vehicle body 100 and to roll along the extension direction of the load-bearing cables 201. The detection unit includes a dual visible light camera 300 and an ultrasonic array sensor 301 fixedly mounted on the bottom of the vehicle body 100 , and an optical gas imager 302 fixedly mounted on the front end of the vehicle body 100 in the forward direction of the vehicle body 100 ; The driving connection part 104 is fixedly arranged on the top of the vehicle body 100 , and the driving connection part 104 is fixed to the traction cable 200 .

[0043] It should be noted that the dual visible light cameras 300 refer to two visible light imaging units integrated on the explosion-proof inspection robot, and the optical gas imager 302 includes a non-cooled infrared detector.

[0044] It is further defined that a groove is provided on the outer peripheral surface of the roller 101, and the groove includes two symmetrically arranged abutment surfaces, which are used to abut against the load-bearing cable 201. Along the circumferential direction of the roller 101, a plurality of installation grooves are spaced apart on the abutment surface, and a magnet 103 is provided in each installation groove. The magnet 103 does not protrude from the abutment surface, and the magnet 103 can generate magnetic attraction between the load-bearing cable 201. Along the circumferential direction of the roller 101, the gap distance between two adjacent magnets 103 is less than the length of the magnet 103, thereby realizing continuous coverage of the magnetic field and improving the adsorption stability.

[0045] A plurality of U-shaped anti-slip plates 102 are fixedly provided on the vehicle body 100 , and a cable space is formed between the anti-slip plates 102 and the corresponding rollers 101 . The load-bearing cables 201 extend through the cable space without contacting the anti-slip plates 102 .

[0046] Further preferably, a radio frequency location tag is provided on the carrier cable 201 , and a radio frequency read / write sensor 303 is provided on the vehicle body 100 . The radio frequency read / write sensor 303 can read the location information on the radio frequency location tag.

[0047] The vehicle body is also equipped with a wireless charging module, an industrial switch and a wireless communication module. An explosion-proof antenna is installed on the vehicle body. The wireless charging module is axially aligned with the external charging base station for electromagnetic coupling.

[0048] Using this robot, one can not only simultaneously obtain optical imaging, infrared temperature field distribution and ultrasonic spectrum characteristics of gas leakage morphology, but also establish a dynamic correlation model of multi-physical field data to improve the accuracy of distinguishing tiny leaks and gas-liquid two-phase leakage characteristics.

[0049] Example 3

[0050] Based on the same principle as the above method, a multi-modal detection explosion-proof system for petrochemical tank area 10 is also proposed. Figure 5 Shown, including: The data acquisition module 11 is used to perform multimodal detection on the target petrochemical tank area according to the preset inspection path and obtain multimodal detection data; A mapping module 12 is configured to analyze the multimodal detection data item by item and map the corresponding analysis results to a unified spatiotemporal coordinate system constructed in advance based on the timestamps calibrated for the multimodal detection data; A cross-cooperative verification module 13 is configured to perform cross-cooperative verification of the re-examination analysis results according to the position of the analysis results in the unified spatiotemporal coordinate system, and determine whether the cross-cooperative verification is established; The output data module 14 is used to calculate the abnormal indicator correlation weight coefficient if the cross-cooperative verification is established, determine the correlation degree according to the calculation result, and output the corresponding explosion-proof disposal plan according to the correlation degree.

[0051] Example 4

[0052] Furthermore, an electronic device is proposed, comprising: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to implement the petrochemical tank area multimodal detection explosion prevention method described in Example 1 when executing the executable instructions.

[0053] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0054] In addition, it should be noted that the descriptions of "first", "second", "one", etc. in the present invention are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly defined. The terms "connected", "fixed", etc. should be understood in a broad sense. For example, "fixed" can be a fixed connection, a detachable connection, or an integral whole; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements, unless otherwise clearly defined. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.

[0055] In addition, the technical solutions between the various embodiments of the present invention can be combined with each other, but it must be based on the fact that ordinary technicians in this field can implement it. When the combination of technical solutions is mutually contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0056] The specific embodiments described herein are merely illustrative of the spirit of the present invention. Persons skilled in the art may make various modifications, additions, or substitutions to the described specific embodiments without departing from the spirit of the present invention or exceeding the scope of the appended claims.

Claims

1. A multi-modal detection explosion-proof method for a petrochemical tank area, characterized in that: include: According to the preset inspection route, multimodal inspection is carried out on the target petrochemical tank area to obtain multimodal inspection data; Analyzing the multimodal detection data item by item, and mapping the corresponding analysis results to a unified spatiotemporal coordinate system constructed in advance based on the timestamps calibrated for the multimodal detection data; Performing cross-cooperative verification of the re-examination analysis results according to the position of the analysis results in the unified time-space coordinate system, and determining whether the cross-cooperative verification is established; If the cross-cooperative verification is established, the abnormal indicator correlation weight coefficient is calculated, the correlation degree is determined according to the calculation result, and the corresponding explosion-proof disposal plan is output according to the correlation degree.

2. The multi-modal detection explosion-proof method for a petrochemical tank area according to claim 1 is characterized in that: Perform multimodal inspection on the target petrochemical tank farm and obtain multimodal inspection data, including: The multimodal detection data includes visible light images, gas distribution data, and acoustic wave signals in a preset frequency band, wherein the gas distribution data includes spectral absorption data and plume distribution data; Using dual visible light cameras to collect visible light images of the target petrochemical tank area; An optical gas imager is used to collect spectral absorption data and plume distribution data of the target petrochemical tank area; Ultrasonic array sensors are used to collect sound wave signals in a preset frequency band.

3. The multi-modal detection explosion-proof method for a petrochemical tank area according to claim 2 is characterized in that: Performing cross-cooperative verification of the re-examination analysis results according to the position of the analysis results in the unified time-space coordinate system, and determining whether the cross-cooperative verification is established, includes: If an analysis result is abnormal, rechecking and analyzing the multimodal detection data based on the position of the abnormality in the unified time and space coordinate system to obtain a recheck analysis result; Calculating the correlation between the analysis result and the re-inspection analysis result, determining whether the correlation is successful based on whether the correlation meets a preset correlation standard, and setting a graded corresponding measure based on the successfully correlated multimodal detection data; If the analysis result and the re-analysis result are both successfully associated, it is determined that the cross-cooperative validation is established.

4. The multi-modal detection explosion-proof method for a petrochemical tank area according to claim 3 is characterized in that: Analyze the multimodal detection data item by item, including: analyzing whether the visible light image contains an abnormal area; if so, the visible light image is abnormal, and extracting an edge contour of the abnormal area; The position of the edge contour in the unified time-space coordinate system is marked, and based on the pre-constructed three-dimensional digital model of the petrochemical tank area, the detection point position within the preset range is locked according to the mark.

5. The multi-modal detection explosion-proof method for a petrochemical tank area according to claim 3 is characterized in that: Analyzing the multimodal detection data item by item also includes: Analyzing the gas distribution data; if the gas distribution data contains a gas concentration higher than a historical normal concentration range, the gas distribution data is considered abnormal, and the corresponding gas is used as a target gas; Analyzing the spectral absorption data of the target gas to identify whether the target gas is a leak-proof gas; Analyze the plume distribution data of the target gas and calculate the diffusion parameters of the target gas.

6. The multi-modal detection explosion-proof method for a petrochemical tank area according to claim 3 is characterized in that: Analyzing the multimodal detection data item by item also includes: Performing a fast Fourier transform on the sound wave signal in the preset frequency band to extract a spectrum diagram of the target frequency band; Identify whether the spectrum graph contains the peak feature of anti-leakage gas. If so, the sound wave signal in the preset frequency band is abnormal, and calculate the phase difference of the sound wave signal corresponding to the peak feature received by the sensor through the beamforming algorithm to locate the sound source direction of the sound wave signal.

7. The multi-modal detection explosion-proof method for a petrochemical tank area according to claim 3 is characterized in that: If the analysis result is abnormal, it also includes: Extracting target data from the multimodal detection data to construct a multimodal association matrix; The target data association analysis is performed to update the data consistency index of the multimodal association matrix, wherein updating the data consistency index includes updating time synchronization, spatial consistency and intensity correlation.

8. The multi-modal detection explosion-proof method for a petrochemical tank area according to claim 7, characterized in that: Analyzing the association of the target data and updating the data consistency index of the multimodal association matrix include: Calculating the time deviation between target data, wherein the time synchronization is determined according to the time deviation; Calculating the Euclidean distance between the center point of the gas diffusion area and the gas source positioning point according to the target data, wherein the spatial consistency is determined according to the corresponding value of the Euclidean distance; Dynamically comparing the changing trends of the gas distribution data and the spectrum of the acoustic wave signal to determine whether they are positively correlated or negatively correlated, wherein the intensity correlation is determined based on the result of the determination; The data consistency index is updated according to the time synchronization, spatial consistency and intensity correlation after the corresponding values ​​are taken.

9. A petrochemical tank area inspection robot, applied to the petrochemical tank area multimodal detection explosion-proof method according to any one of claims 1 to 8, characterized in that: include: A vehicle body, wherein two rollers are symmetrically provided on both sides of the vehicle body, and the rollers are used to connect with the load-bearing cables; a detection unit comprising a dual visible light camera and an ultrasonic array sensor fixedly mounted on the bottom of the vehicle body, and an optical gas imager fixedly mounted on the front end of the vehicle body along the forward direction of the vehicle body; A drive connection portion is fixedly arranged on the top of the vehicle body and is used for connecting to the traction cable.

10. The petrochemical tank area inspection robot according to claim 9, characterized in that: The outer circumferential surface of the roller is provided with a groove, and the groove includes two symmetrically arranged abutting surfaces, and the abutting surfaces are used to abut against the load-bearing cable. Along the circumferential direction of the roller, a plurality of magnets are spaced apart on the abutting surfaces.

Images