A multi-type navigation mark inspection device and inspection method based on artificial intelligence

Abstract

The present invention is a device and method based on artificial intelligence that can be used for multi-type navigation mark inspection, aiming to improve detection efficiency and accuracy. For visual navigation lights, the Mask R-CNN algorithm of artificial intelligence is used to improve the recognition ability of navigation mark object features through video key frame extraction and data enhancement technology, combined with illumination correction, feature fusion and attention mechanism, and the loss function that is insensitive to illumination is designed and image processing is performed after segmentation to remove illumination noise. The detection module can automatically calculate the flashing cycle, interval and light intensity data of the navigation mark, greatly improving the detection accuracy. The radar navigation mark detection module automatically calculates the frequency range, effective distance and characteristic value of the electromagnetic wave signal; the acoustic navigation mark detection collects the sound signal through the sound pressure sensor and performs digital processing, and outputs the characteristic data for judging the navigation mark status. The device not only improves the detection efficiency and accuracy, but the accumulated data also helps to improve product design. Its wind curtain structure design reduces the corrosion of the marine environment to the equipment and prolongs the service life. At the same time, it uses the external natural wind to drive the air supply component, which has the effect of energy saving and environmental protection.

Description

Technical Field

[0001] The present invention relates to the technical field of navigation mark inspection, and in particular to a multi-type navigation mark inspection device and inspection method based on artificial intelligence. Background Art

[0002] Navigation beacon inspection is to conduct regular or irregular inspection and maintenance work on navigation beacons to ensure their normal operation and play a role in navigation. In the prior art, the inspection of navigation beacons usually requires inspectors to carry bulky equipment and tools, which is inefficient and difficult to ensure the comprehensiveness and accuracy of the inspection. For example, the flashing cycle and flashing time interval of visual navigation beacons are currently calculated by stopwatches, and for acoustic navigation beacons, they are mainly detected by artificial hearing, and the above detection efficiency and accuracy are low. For radar navigation beacons, it is usually necessary to use ships with radar receiving equipment for detection, and the cost of using ships increases the inspection cost, and the inspection devices currently on the market often have a single function and cannot meet the inspection needs of different types of navigation beacons. Therefore, it is necessary to design a portable inspection device suitable for a variety of navigation beacons to improve the inspection efficiency and accuracy. In addition, the general working environment of navigation beacon inspection devices usually has a high humidity and salt spray concentration. Moisture and salt spray can easily cause problems such as rust and short circuit on the surface of the device, affecting its service life.

[0003] In order to solve the above problems, the present invention proposes a multi-type navigation mark inspection device and inspection method based on artificial intelligence. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies of the prior art, adapt to actual needs, and provide a multi-type navigation mark inspection device and inspection method based on artificial intelligence to solve the above-mentioned technical problems.

[0005] In order to achieve the purpose of the present invention, the technical solution adopted by the present invention is:

[0006] A multi-type navigation mark inspection device based on artificial intelligence includes an inspection equipment body, an air supply box is arranged at the bottom of the inspection equipment body, the inner wall of the air supply box is fixedly connected to a box cover fixedly connected to the inspection equipment body, the upper surface of the box cover is provided with a first air outlet surrounding the inspection equipment body, the upper surface of the box cover is fixedly connected to a fixed pipe connected to the air supply box, the surface of the fixed pipe is provided with a second air outlet facing the inspection equipment body, the first air outlet and the second air outlet cooperate to form a wind curtain to wrap the inspection equipment body, the lower surface of the air supply box is fixedly connected to an air supply component, and the air supply component supplies dry and low-salinity air into the air supply box.

[0007] Furthermore, a triangular bracket is fixedly connected to the lower surface of the air supply assembly, and the triangular bracket has a tray, the surface of the tray is rotatably connected to three supporting legs, the bottom ends of the supporting legs are hemispherical, and the lower surface of the tray is fixedly connected to a loading part, and a counterweight block is arranged in the loading part.

[0008] Furthermore, an end cover is provided at the bottom end of the loading part, an external threaded tube is fixedly connected to the upper surface of the end cover, an internal thread matching the external threaded tube is provided on the inner wall of the loading part, the external threaded tube is threadedly connected to the loading part, a clamping block is fixedly connected to the surface of the end cover, and a clamping groove corresponding to the supporting leg is provided on the surface of the clamping block, and the clamping block is elastic.

[0009] Furthermore, the loading part includes an upper connecting tube fixedly connected to the lower surface of the tray, the bottom end of the upper connecting tube is rotatably connected to a driving member, the driving member is used to drive the air supply assembly to operate, the bottom end of the driving member is rotatably connected to a lower connecting tube, the counterweight block is arranged in the lower connecting tube, and the external threaded tube is threadedly connected to the lower connecting tube.

[0010] Furthermore, the driving member includes a rotating column, both ends of which are rotatably connected to the upper connecting tube and the lower connecting tube respectively, a concave mounting groove is formed on the surface of the rotating column, and a plurality of rotating blades are rotatably connected to the inner wall of the mounting groove, and the rotating blades can drive the rotating column to rotate under the influence of external wind force, and a transmission shaft is fixedly connected to the top end of the rotating column, and the transmission shaft is connected to the air supply assembly and drives the air supply assembly to operate.

[0011] Furthermore, a groove is provided at the top of the rotating column, and a groove cover is fixedly connected to the inner wall of the groove, the groove cover is fixedly connected to the transmission shaft, the inner bottom wall of the groove is rotatably connected to a driving gear, and the inner walls of the groove are respectively rotatably connected to a plurality of driven gears, the driven gears correspond to the rotating leaves, the driving gears are meshed with the driven gears, the lower surface of the driven gears is fixedly connected to a rotating shaft, the rotating shaft extends into the mounting groove, and the bottom end of the rotating shaft is rotatably connected to the inner bottom wall of the mounting groove, and the rotating leaves are fixedly connected to the surface of the rotating shaft.

[0012] Furthermore, a sliding groove is provided at the bottom end of the rotating column, and the inner wall of the sliding groove is slidably connected to a starting piece, the starting piece includes a sliding cylinder slidably connected to the inner wall of the sliding groove, the inner wall of the sliding cylinder is fixedly connected to a cylindrical pressure rod, a rotating hole is provided at the center position of the inner wall of the rotating column, and the inner wall of the rotating hole is rotatably connected to the starting column, the starting column is fixedly connected to the driving gear, the sliding cylinder is sleeved on the outside of the starting column, a limiting groove is provided on the surface of the starting column, the pressure rod is arranged on the inner wall of the limiting groove, and the surface of the pressure rod is fitted with the limiting groove, a spring groove is provided on the upper surface of the sliding cylinder, and a telescopic spring is fixedly connected to the inner bottom wall of the spring groove, the other end of the telescopic spring is fixedly connected to the inner top wall of the sliding groove, and the counterweight block is pushed by the end cover to drive the sliding cylinder to rise.

[0013] Furthermore, an embedding groove is provided on the lower surface of the sliding cylinder, and a ball is movably arranged in the embedding groove.

[0014] Furthermore, the air supply assembly includes a fixed cylinder fixedly connected to the lower surface of the air supply box, the fixed cylinder is communicated with the air supply box, the top end of the transmission shaft extends to the interior of the air supply box and is connected to the fan blades through a gear box, an air inlet is opened on the surface of the fixed cylinder, and a filter is fixedly connected to the inner wall of the air inlet, and an electric heating wire is arranged in the fixed cylinder.

[0015] Furthermore, a communication port is provided on the inner bottom wall of the air supply box, the fixed cylinder is connected with the air supply box through the communication port, and a one-way valve is fixedly connected to the inner wall of the communication port.

[0016] A multi-type navigation mark inspection device based on artificial intelligence includes an inspection device body 5, the inspection device body 5 includes a detection platform 1, a visual detection module 17, an audio detection module 18, a radar detection module 19, and a processor 20 are fixedly installed inside the detection platform 1, an LCD display screen 21 is fixedly installed on the front surface of the detection platform 1, a sound pickup area 22, a Beidou communication module 23, a radar transmitting device 24 and a radar receiving device 25 are respectively provided on the upper surface of the detection platform 1, and a photosensitive area 26, a temperature sensor 27 and a laser rangefinder 28 are provided on the rear surface of the detection platform 1.

[0017] Furthermore, a visual detection switch button 14, an audio detection switch button 15 and a radar detection switch button 16 are fixedly installed on the front surface of the detection platform 1. By setting the visual detection switch button 14, the audio detection switch button 15 and the radar detection switch button 16, they are used to start the visual detection module 2, the audio detection module 3 and the radar detection module 4 respectively.

[0018] Furthermore, the visual detection module 2 includes a CMOS image sensor, a lens and a light intensity sensor, and a filter film is arranged in front of the lens.

[0019] Furthermore, the sound detection module 3 includes a sound pressure sensor 301 and a detection unit 302 . The sound pressure sensor 301 is connected to the detection unit 302 . The output end of the detection unit 302 is connected to the input end of the LCD display screen 21 .

[0020] Furthermore, the radar detection module 4 includes a transmitting unit 401, a receiving unit 402, an intermediate frequency unit 403, a digital signal processing unit 404 and a main control unit 405. The transmitting unit 401 is used to generate an analog radar signal to trigger the transponder, the output end of the receiving unit 402 is connected to the signal input end of the intermediate frequency unit 403, the output end of the intermediate frequency unit 403 is connected to the signal input end of the digital signal processing unit 404, the output end of the digital signal processing unit 404 is connected to the signal input end of the main control unit 405, and the main control unit 405 is connected to the LCD display 21.

[0021] Furthermore, the source module 5 is a ternary lithium battery, the rated voltage of the ternary lithium battery is 48V, 10Ah, and the maximum continuous working hours are 12 hours.

[0022] An inspection method for multiple types of navigation mark inspection devices based on artificial intelligence comprises the following steps:

[0023] S1: Choose good weather (atmospheric transparency coefficient is greater than or equal to 0.74, ambient temperature is -25 to +55 degrees Celsius, relative humidity is not more than 95%);

[0024] S2: Move the inspection device to the vicinity of the navigation mark, and then start the inspection device for self-inspection;

[0025] S3: Determine the type of navigation mark to be tested and select the appropriate detection method:

[0026] S31: When the measured beacon is a visual beacon, a visual detection module is used for detection;

[0027] S32: When the navigation mark to be measured is an acoustic navigation mark, an acoustic detection module is used for detection;

[0028] S33: When the measured beacon is a radar beacon, a radar detection module is used for detection.

[0029] Furthermore, S31 is specifically:

[0030] S311: Turn on the visual detection switch button 29, open the protective cover of the photosensitive area 26, and use the camera to aim at the direction of the navigation mark to shoot a video. When shooting, take the navigation mark as the center, shoot at six directions at intervals of 60 degrees, and measure twice at each test position; at the same time, the laser rangefinder 28 also measures the distance between the measured navigation mark and the inspection device in real time;

[0031] S312: separating the navigation light in the image from the background;

[0032] S3121: Generate training model

[0033] Based on multiple pre-prepared images of navigation lights as training data sets, the pre-trained convolutional neural network ResNet is used as the backbone network to extract feature maps.

[0034] The results of compressing the length and width twice, three times, four times, and five times are taken out, namely p1, p2, p3, and p4 to construct the feature pyramid structure.

[0035] The candidate regions are generated through RPN, and classification and bounding box regression are performed to determine whether the captured content contains the target, and the suggestion box is adjusted to achieve pixel-level object segmentation.

[0036] The candidate region is generated using the following formula:

[0037] a x，y =(x+w a ·r·cos(θ),y+h a ·r·sin(θ))

[0038] w a and h a are the width and height of the candidate region, r and θ are the scale and rotation angle respectively.

[0039] Use the labelme package in Python to annotate the obtained pixel-level images. After completing the annotation of all images, use the labelme2coco package in Python to convert the labelme format dataset into COCO format to obtain the training model.

[0040] S3122: Based on the captured video, the processor 20 extracts key frames from the video to obtain multiple navigation mark images, and uses the artificial intelligence Mask R-CNN algorithm to perform instance segmentation. Then, based on the generated training model, the processor 20 performs real-time navigation mark light detection on the image extracted from the video and generates a COCO format image, thereby achieving pixel-level object segmentation.

[0041] S313: performing image processing;

[0042] Calculate the stroboscopic period: Based on the captured video, use the CMOS image sensor to convert the light signal of the beacon into an electrical signal; use the A / D converter to convert the electrical signal into a digital signal;

[0043] Calculate the time interval: The digital signal is processed by the image processing algorithm to obtain the illumination and chromaticity of the light source; the frequency and duration of the flash signal are detected to calculate the interval between flashes;

[0044] Calculate the light intensity: The laser rangefinder 28 of the inspection device collects the relative distance information between the inspection device and the navigation light as the distance parameter l, and then calculates the light intensity:

[0045] I0=El 2 cosθ

[0046] Where I0 is the light intensity of the beacon light being measured; E is the illuminance on the receiving surface of the illuminometer; l is the measuring distance; θ is the angle between the light beam and the normal of the receiving surface of the illuminometer.

[0047] S314: Output the result to the display screen 21.

[0048] Further, S32 is specifically:

[0049] S321: Turn on the audio detection switch button 30;

[0050] S322: Turn on the test release switch of the acoustic beacon under test;

[0051] S323: The sound pressure sensor 301 collects sound, and the detection unit 302 performs digital signal (DSP) algorithm processing on the collected sound signal. The process includes sampling, quantization, time domain processing (filtering), frequency domain processing (Fourier transform), and feature extraction. The calculation formula for frequency domain processing is:

[0052]

[0053] Among them, F(ω) is a complex function in the frequency domain, f(t) is a function in the time domain, ω represents the frequency, and e -jωt is a complex exponential function. This formula decomposes the time domain signal f(t) into a series of complex amplitude and phase combinations that describe the contribution of the signal at different frequencies.

[0054] SS24: The extracted features are output to the display screen. Based on the displayed results, the staff determines whether they are consistent with the data published when the beacon was originally shipped from the factory, thereby determining whether the beacon is in a valid state.

[0055] Further, S33 is specifically:

[0056] S331: Turn on the radar detection switch button 31;

[0057] S332: Start detection and output the detection result to the display screen 21;

[0058] a) When the measured beacon is a radar transponder:

[0059] a1): Measure whether the frequency bandwidth range is normal: within the frequency range of the radar transponder, the radar transmitting device 24 sends a radar signal of the corresponding frequency. If the radar receiving device 25 can receive the response signal of the measured navigation mark at any frequency point within the bandwidth, it means that the navigation mark meets the frequency response requirements and is in normal working condition;

[0060] a2): Measure the effective distance: According to the effective distance published by the navigation mark, perform a pull-distance test and move the inspection device to the maximum effective distance; determine whether the radar receiving device 25 can receive the radar signal fed back by the navigation mark;

[0061] a3): Measure the agile frequency: change the transmission frequency. If the received signal frequency also changes and the frequencies are the same, it is determined that the radar transponder under test has the agile frequency characteristic.

[0062] a4): Measure characteristic values: According to the obtained signal packet, obtain the characteristic values ​​contained in the signal;

[0063] a5): Compare the measurement result with the data published by the navigation mark to determine whether the navigation mark is qualified, and output the result to the display screen 21.

[0064] b) When the measured beacon is a radar reflector:

[0065] b1): The radar transmitting device 24 transmits a 9.4Ghz (X-band) radar signal;

[0066] b2): Determine whether the radar receiving device 25 can receive the reflected radar signal;

[0067] b3): If the reflected signal can be received, the radar reflection area RCS is calculated and compared with the data published by the beacon to determine whether the beacon is qualified and output the result to the display; if the reflected signal cannot be received, it is directly judged as unqualified and the result is output to the display.

[0068] The calculation formula for the spherical radar reflector RCS is:

[0069] RCS=4π×r 2 / λ(m 2 )

[0070] r is the radius of the sphere, λ is the radar wavelength

[0071] The calculation formula for the RCS of a square radar reflector is:

[0072] RCS=4πLC / λ 2 (m 2 )

[0073] L, C are the length and width of the rectangle, λ is the radar wavelength;

[0074] The calculation formula for the RCS of a cylindrical radar reflector is:

[0075] RCS=2πrh 2 / λ(m 2 )

[0076] r is the radius of the cylinder, h is the height of the cylinder, and λ is the radar wavelength;

[0077] The calculation formula for the RCS of a triangular radar reflector is:

[0078] RCS=4πL 4 / 3λ 2 (m 2 )

[0079] Where L is the side length of the triangle and λ is the radar wavelength.

[0080] c) When the measured navigation mark is a radar beacon:

[0081] c1): Measure the effective distance: According to the effective distance published by the beacon, perform a pull-distance test and move the inspection device to the maximum effective distance; determine whether the radar receiving device 25 can receive the radar signal sent by the beacon;

[0082] c2): Measurement frequency range: Check whether the received signal is within the X-band range (9.3Ghz~9.5Ghz). If it is within this band, it proves that the measured beacon is working normally; if not, it proves that it is working abnormally.

[0083] c3): Output C2 result to display screen 21.

[0084] Beneficial effects:

[0085] 1. In the present invention, according to the type of navigation mark, the corresponding detection mode can be turned on to realize automatic detection of navigation mark. When detecting the visual navigation mark light, the detection module will automatically calculate the flashing cycle and flashing interval of the navigation mark light, replacing the existing traditional method of manual countdown or using a stopwatch; the light intensity data of the navigation mark light can be detected in real time; when detecting the radar navigation mark, the detection module will automatically calculate the key data such as the frequency range, working distance, and characteristic value of the navigation mark; when detecting the acoustic navigation mark, the sound pressure sensor collects the sound, and the detection unit processes the collected sound signal with a digital signal algorithm, and outputs the extracted features to the display screen. The staff judges whether the displayed result is consistent with the data published when the navigation mark was originally shipped from the factory, thereby judging whether the navigation mark is in a valid state. The inspection device can not only improve the detection efficiency, but also greatly improve the detection accuracy.

[0086] 2. In the present invention, dry low-salinity air is delivered into the air supply box through the air supply assembly, and a first air outlet and a second air outlet are provided to form an air curtain to surround the entire inspection equipment body, thereby reducing the corrosion of the inspection equipment body by marine environmental factors such as humid air and salt spray, so that the navigation mark inspection device has the effect of extending the service life of the equipment.

[0087] 3. In the present invention, by setting the rotating blades, the rotating column can be driven to rotate under the action of external wind force, and then the transmission shaft can be driven to rotate. The fan blades can be driven to rotate rapidly through the cooperation of the transmission shaft and the gear box, so that the external air passes through the filter to reduce the salinity and then enters the fixed cylinder. The air is heated by the electric heating wire to make it dry and then enters the air supply box. The external wind force is used as the driving force, which has the effect of energy saving.

[0088] 4. In the present invention, the sliding cylinder can be driven to move downward by the spring reset, and then the pressure column is driven to move downward, and then the starting column is rotated clockwise with the cooperation of the limit groove, and then the driving gear is driven to rotate, and then the driven gear is driven to rotate counterclockwise, and then the rotating shaft is driven to rotate, and then the rotating leaves are driven to rotate and fit against the inner wall of the installation groove, and the rotating leaves are folded to facilitate carrying the entire device. BRIEF DESCRIPTION OF THE DRAWINGS

[0089] Figure 1 It is a schematic diagram of the three-dimensional structure of the present invention in working state;

[0090] Figure 2 It is a schematic diagram of the three-dimensional structure of the present invention in a folded state;

[0091] Figure 3 It is a schematic diagram of the exploded structure of the air supply box, the box cover and the fixing frame of the present invention;

[0092] Figure 4 For the present invention Figure 3 A schematic diagram of the enlarged structure at A in the middle;

[0093] Figure 5 For the present invention Figure 3 A schematic diagram of the enlarged structure at B in the middle;

[0094] Figure 6 It is a schematic diagram of the three-dimensional structure of the loading part of the present invention;

[0095] Figure 7 It is a schematic diagram of the three-dimensional structure of the air supply assembly of the present invention;

[0096] Figure 8 It is a schematic cross-sectional structural diagram of the upper connecting tube of the present invention;

[0097] Fig. 9 It is a three-dimensional structural schematic diagram of the driving member of the present invention;

[0098] Fig.10 It is a schematic cross-sectional structural diagram of the rotating column of the present invention;

[0099] Fig.11 It is a schematic diagram of the explosion structure of the starting member of the present invention;

[0100] Fig.12 It is a schematic cross-sectional structural diagram of the sliding cylinder of the present invention;

[0101] Fig.13 It is a cross-sectional structural schematic diagram of the lower connecting cylinder and the clamping block of the present invention;

[0102] Fig.14 It is a schematic diagram of the three-dimensional structure of the inspection device body of the present invention;

[0103] Fig.15 It is a characteristic pyramid structure. DETAILED DESCRIPTION

[0104] The following is combined with Figure 1-15 The present invention is further described with examples:

[0105] Example 1

[0106] like Figure 1-15 As shown, a multi-type navigation mark inspection device based on artificial intelligence includes an inspection equipment body 5, an air supply box 6 is arranged at the bottom of the inspection equipment body 5, the inner wall of the air supply box 6 is fixedly connected with a box cover 7 fixedly connected to the inspection equipment body 5, the upper surface of the box cover 7 is provided with a first air vent 8 surrounding the inspection equipment body 5, the upper surface of the box cover 7 is fixedly connected with a fixed pipe 9 connected to the air supply box 6, the surface of the fixed pipe 9 is provided with a second air vent 10 facing the inspection equipment body 5, the first air vent 8 and the second air vent 10 cooperate to form a wind curtain to wrap the inspection equipment body 5, the lower surface of the air supply box 6 is fixedly connected with an air supply component 1, and the air supply component 1 supplies dry and low-salinity air into the air supply box 6.

[0107] Specifically, dry low-salinity air is delivered into the air supply box 6 through the air supply assembly 1, and a first air outlet 8 and a second air outlet 10 are provided to form a wind curtain to surround the entire inspection equipment body 5, thereby reducing the corrosion of the inspection equipment body 5 by marine environmental factors such as humid air and salt spray, so that the navigation mark inspection device has the effect of extending the service life of the equipment.

[0108] A triangular bracket is fixedly connected to the lower surface of the air supply component 1, and the triangular bracket has a tray 11. Three supporting legs 12 are rotatably connected to the surface of the tray 11, and the bottom ends of the supporting legs 12 are hemispherical. A loading part 2 is fixedly connected to the lower surface of the tray 11, and a counterweight 13 is arranged in the loading part 2. The arrangement of the tray 11 and the supporting legs 12 facilitates the placement of the device. The arrangement of the loading part 2 and the counterweight 13 improves the stability of the device and makes it have a higher wind resistance. The triangular bracket also has a locking structure for positioning the angle of the supporting legs 12. The supporting legs 12 are made of high-strength aluminum alloy, and the surface of the supporting legs 12 is coated with a corrosion-resistant layer, so that the inspection device is suitable for corrosive environments such as the seaside.

[0109] An end cover 14 is provided at the bottom end of the loading part 2, and an external threaded tube 15 is fixedly connected to the upper surface of the end cover 14. An internal thread matching the external threaded tube 15 is provided on the inner wall of the loading part 2, and the external threaded tube 15 is threadedly connected to the loading part 2. A clamping block 16 is fixedly connected to the surface of the end cover 14, and a clamping groove corresponding to the supporting leg 12 is provided on the surface of the clamping block 16. The clamping block 16 is elastic. The counterweight 13 is supported by the end cover 14. The external threaded tube 15 is provided to facilitate the removal of the counterweight 13 and the separate carrying when the device is moved. The supporting leg 12 can be fixed through the clamping groove after the supporting leg 12 is folded. At the same time, the cooperation between the clamping block 16 and the clamping groove facilitates the rotation of the external threaded tube 15.

[0110] The loading part 2 includes an upper connecting tube 201 fixedly connected to the lower surface of the tray 11, and the bottom end of the upper connecting tube 201 is rotatably connected to a driving member 202, and the driving member 202 is used to drive the air supply component 1 to operate. The bottom end of the driving member 202 is rotatably connected to a lower connecting tube 203, and the counterweight block 13 is arranged in the lower connecting tube 203. The external threaded tube 15 is threadedly connected to the lower connecting tube 203. By setting the driving member 202, the air supply component 1 is driven to operate and supply air.

[0111] The driving member 202 includes a rotating column 2021, and the two ends of the rotating column 2021 are rotatably connected to the upper connecting tube 201 and the lower connecting tube 203 respectively. The surface of the rotating column 2021 is formed with a concave installation groove, and the inner wall of the installation groove is rotatably connected with a plurality of rotating blades 2022. The rotating blades 2022 can drive the rotating column 2021 to rotate under the influence of external wind force. The top of the rotating column 2021 is fixedly connected with a transmission shaft 2023, and the transmission shaft 2023 is connected to the air supply component 1 and drives the air supply component 1 to operate. By setting the rotating blades 2022, the rotating column 2021 can be driven to rotate under the action of external wind force, thereby driving the transmission shaft 2023 to rotate, and the air supply component 1 is driven to operate through the transmission shaft 2023.

[0112] A groove is provided at the top of the rotating column 2021, and a groove cover 2024 is fixedly connected to the inner wall of the groove, and the groove cover 2024 is fixedly connected to the transmission shaft 2023. A driving gear 2025 is rotatably connected to the inner bottom wall of the groove, and a plurality of driven gears 2026 are rotatably connected to the inner walls of the groove, respectively. The driven gears 2026 correspond to the rotating blades 2022, and the driving gear 2025 meshes with the driven gear 2026. A rotating shaft 2027 is fixedly connected to the lower surface of the driven gear 2026, and the rotating shaft 2027 extends into the mounting groove, and the bottom end of the rotating shaft 2027 is rotatably connected to the inner bottom wall of the mounting groove. The rotating blade 2022 is fixedly connected to the surface of the rotating shaft 2027, and the driving gear 2025 rotates counterclockwise, thereby driving the driven gear 2026 to rotate clockwise, thereby driving the rotating shaft 2027 to rotate clockwise, and then driving the rotating blade 2022 to rotate clockwise and then unfold so as to be pushed by wind.

[0113] A sliding groove is provided at the bottom end of the rotating column 2021, and the inner wall of the sliding groove is slidably connected with a starting member 4, the starting member 4 includes a sliding cylinder 401 slidably connected to the inner wall of the sliding groove, the inner wall of the sliding cylinder 401 is fixedly connected with a cylindrical pressure rod 402, a rotating hole is provided at the center of the inner wall of the rotating column 2021, and the inner wall of the rotating hole is rotatably connected with a starting column 403, the starting column 403 is fixedly connected to the driving gear 2025, the sliding cylinder 401 is sleeved on the outside of the starting column 403, the surface of the starting column 403 is provided with a limiting groove 404, the pressure rod 402 is arranged on the inner wall of the limiting groove 404, and the surface of the pressure rod 402 is fitted with the limiting groove 404, the upper surface of the sliding cylinder 401 is provided with a spring groove, and the inner bottom wall of the spring groove is fixedly connected with a telescopic spring 405, the other end of the telescopic spring 405 is connected to the sliding groove The inner top wall of the groove is fixedly connected, and the counterweight block 13 is pushed by the end cover 14 to drive the sliding cylinder 401 to rise. The counterweight block 13 moves upward, pushing the sliding cylinder 401 to move upward, and then driving the pressure rod 402 to move upward. With the cooperation of the limit groove 404, the starting column 403 rotates counterclockwise, and then drives the driving gear 2025 to rotate counterclockwise. The sliding cylinder 401 can be driven to move downward through the spring reset, and then the pressure column is driven to move downward, and then with the cooperation of the limit groove 404, the starting column 403 is rotated clockwise, and then the driving gear 2025 is rotated, and then the driven gear 2026 is driven counterclockwise, and then the rotating shaft 2027 is driven to rotate, and then the rotating leaf 2022 is driven to rotate and fit on the inner wall of the installation groove, and the rotating leaf 2022 is folded to facilitate carrying the entire device.

[0114] An embedding groove is provided on the lower surface of the sliding cylinder 401 , and a ball 3 is movably arranged in the embedding groove. By arranging the ball 3 , the friction between the sliding groove and the counterweight block 13 is reduced, and tightening the external threaded tube 15 is convenient.

[0115] The air supply component 1 includes a fixed cylinder 101 fixedly connected to the lower surface of the air supply box 6, the fixed cylinder 101 is communicated with the air supply box 6, the top end of the transmission shaft 2023 extends to the interior of the air supply box 6 and is connected to the fan blades 102 through the gear box 104, an air inlet is opened on the surface of the fixed cylinder 101, and the inner wall of the air inlet is fixedly connected to the filter screen 103, an electric heating wire is arranged in the fixed cylinder 101, and the fan blades 102 are driven to rotate rapidly through the transmission shaft 2023 and the gear box 104, so that the outside air enters the fixed cylinder 101 after passing through the filter screen 103, and the air is heated by the electric heating wire to make it dry and then enter the air supply box 6, and the outside wind force is used as the driving force, which has the effect of energy saving.

[0116] A connecting port is provided on the inner bottom wall of the air supply box 6, and the fixed tube 101 is connected to the air supply box 6 through the connecting port. A one-way valve 32 is fixedly connected to the inner wall of the connecting port. The one-way valve 32 is provided to restrict the flow of airflow so that it can only flow from bottom to top.

[0117] Working principle: When the navigation mark inspection device is used, the user first uses the tripod bracket to place the device, then uses the clamping block 16 and the clamping slot to unscrew the external threaded tube 15, and then puts the counterweight 13 on the end cover 14, and tightens the external threaded tube 15 and the lower connecting tube 203 in the same way. During this process, hold the lower connecting tube 203 with one hand to prevent it from rotating. During the tightening process, the external threaded tube 15 drives the end cover 14 to move upward, and the end cover 14 drives the counterweight 13 to move upward until the counterweight 13 contacts the ball 3 and pushes the ball 3 and the sliding tube 401 to move upward. The sliding tube 401 drives the pressure rod 402 to move upward, and with the cooperation of the limit groove 404, the starting column 403 rotates counterclockwise, and the starting column 403 drives the driving gear 2025 to rotate counterclockwise, and the driving gear 2025 drives the driven gear 2026 to rotate clockwise. The wheel 2026 drives the rotating shaft 2027 to rotate clockwise, and the rotating shaft 2027 drives the rotating blades 2022 to rotate clockwise and then unfold so as to be pushed by the wind. Under the action of external wind, the rotating blades 2022 drive the rotating column 2021 to rotate, and the rotating column 2021 drives the transmission shaft 2023 to rotate. The transmission shaft 2023 cooperates with the gear box 104 to drive the fan blades 102 to rotate rapidly, so that the outside air passes through the filter 103 to reduce the salinity and then enters the fixed cylinder 101. The air is heated by the electric heating wire to make it dry and then enters the air supply box 6. As the pressure in the air supply box 6 increases, the first air outlet 8 and the second air outlet 10 are used to spray outward to form a wind curtain to surround the entire inspection equipment body 5, thereby reducing the corrosion of the inspection equipment body 5 by marine environmental factors such as humid air and salt spray, so that the navigation mark inspection device has the effect of extending the service life of the equipment.

[0118] Example 2

[0119] A multi-type navigation mark inspection device based on artificial intelligence includes an inspection device body 5, the inspection device body 5 includes a detection platform 1, a visual detection module 17, an audio detection module 18, a radar detection module 19, and a processor 20 are fixedly installed inside the detection platform 1, an LCD display screen 21 is fixedly installed on the front surface of the detection platform 1, a sound pickup area 22, a Beidou communication module 23, a radar transmitting device 24 and a radar receiving device 25 are respectively provided on the upper surface of the detection platform 1, and a photosensitive area 26, a temperature sensor 27 and a laser rangefinder 28 are provided on the rear surface of the detection platform 1.

[0120] Specifically, by setting up the visual detection module 2, the audio detection module 3 and the radar detection module 4, the visual navigation mark, the audio navigation mark and the radar navigation mark can be inspected respectively, which can meet the inspection needs of various navigation marks, so that the portable multi-type navigation mark inspection device has the effect of adapting to the inspection of various types of navigation marks and improving the inspection efficiency.

[0121] The front surface of the detection platform 1 is fixedly installed with a visual detection switch button 14, an audio detection switch button 15 and a radar detection switch button 16. The visual detection switch button 14, the audio detection switch button 15 and the radar detection switch button 16 are set to start the visual detection module 2, the audio detection module 3 and the radar detection module 4 respectively.

[0122] The visual detection module 2 includes a CMOS image sensor, a lens and a light intensity sensor, and a filter film is arranged in front of the lens.

[0123] The key frames are extracted from the captured video through the camera 204, and the image frame data of the navigation light video is enhanced based on the artificial intelligence model Mask R-CNN during the model training stage to improve the adaptability to lighting changes. The lighting correction technology is applied to reduce the impact of uneven lighting. The feature fusion and attention mechanism are used to improve the model's recognition ability of the navigation object features. A loss function that is insensitive to lighting is designed, and image processing is performed after segmentation to remove the noise caused by lighting, segment different objects in the image, and then remove background ambient light and sunlight. The color image is used for light intensity detection. Specifically, the CMOS image sensor converts the acquired light signal into an electrical signal, and outputs it into a corresponding digital signal data through A / D conversion to facilitate subsequent analysis and calculation. Then, the processor calculates and processes the light source illumination, chromaticity, and the interval time of the light flash, and then transmits the result to the LCD display 21.

[0124] The sound detection module 3 includes a sound pressure sensor 301 and a detection unit 302. The sound pressure sensor 301 is connected to the detection unit 302. The output end of the detection unit 302 is connected to the input end of the LCD display screen 21. First, an appropriate amplifier and filter are used to enhance the sound signal and remove possible interference and noise to improve the accuracy and reliability of the detection. The collected sound signal is processed by the detection unit 302 using a digital signal (DSP) algorithm. The process includes sampling, quantization, time domain processing (filtering), frequency domain processing (Fourier transform), noise elimination and feature extraction. The hearing range detection result is displayed on the LCD display screen 21, and the hearing range range is displayed or an alarm is issued to indicate whether the navigation sound is within the expected range.

[0125] The radar detection module 4 includes a transmitting unit 401, a receiving unit 402, an intermediate frequency unit 403, a digital signal processing unit 404 and a main control unit 405. The transmitting unit 401 is used to generate an analog radar signal to trigger a transponder. The output end of the receiving unit 402 is connected to the signal input end of the intermediate frequency unit 403, the output end of the intermediate frequency unit 403 is connected to the signal input end of the digital signal processing unit 404, the output end of the digital signal processing unit 404 is connected to the signal input end of the main control unit 405, and the main control unit 405 is connected to the LCD display screen 21.

[0126] Radar transponder detection: the transmitting unit 401 generates an analog radar signal to trigger the transponder to work, the receiving unit 402 receives the response signal and mixes and amplifies it, the intermediate frequency unit 403 converts the intermediate frequency signal into an amplitude component and a frequency component, and demodulates the intermediate frequency signal to obtain a signal envelope, the digital signal processing unit 404 receives the frequency component and amplitude component of the intermediate frequency unit 403 and the transmission intensity signal of the transmitting unit 401, and the main control unit 405 obtains the data signal to measure multiple parameters of the radar transponder to be detected; radar reflector detection: the transmitting unit 401 generates an analog radar signal to trigger the reflector to work, the receiving unit 402 receives the response signal, the digital signal processing unit 404 processes the response signal, and the main control unit 405 obtains the data signal to measure the reflectivity parameters of the radar reflector to be detected; radar beacon detection: the receiving unit 402 receives the signal, the digital signal processing unit 404 processes the response signal, and the main control unit 405 obtains the data signal and analyzes the frequency and effective range of the signal.

[0127] A pan-tilt bracket 17 is fixedly installed under the detection platform 1. The power module 5 is a ternary lithium battery. By setting the pan-tilt bracket 17, the detection platform 1 is supported, which makes it convenient to carry and move the inspection device. The rated voltage of the ternary lithium battery is 48v, 10Ah, and it can work continuously for a maximum of 12 hours.

[0128] The material of the pan / tilt bracket 17 is high-strength aluminum alloy, and the surface of the pan / tilt bracket 17 is coated with a corrosion-resistant layer so that the inspection device is suitable for use in corrosive environments such as the seaside.

[0129] Example 3

[0130] An inspection method for multiple types of navigation mark inspection devices based on artificial intelligence comprises the following steps:

[0131] S1: Choose good weather (atmospheric transparency coefficient is greater than or equal to 0.74, ambient temperature is -25 to +55 degrees Celsius, relative humidity is not more than 95%);

[0132] S2: Move the inspection device to the vicinity of the navigation mark, and then start the inspection device for self-inspection;

[0133] S3: Determine the type of navigation mark to be tested and select the appropriate detection method;

[0134] S31: When the measured beacon is a visual beacon, a visual detection module is used for detection;

[0135] S311: Turn on the visual detection switch button 29, open the protective cover of the photosensitive area 26, and use the camera to aim at the direction of the navigation mark to shoot a video. When shooting, take the navigation mark as the center, shoot at six directions at intervals of 60 degrees, and measure twice at each test position; at the same time, the laser rangefinder 28 also measures the distance between the measured navigation mark and the inspection device in real time;

[0136] S312: separating the navigation light in the image from the background;

[0137] S3121: Generate training model

[0138] Based on multiple pre-prepared images of navigation lights as training data sets, the pre-trained convolutional neural network ResNet is used as the backbone network to extract feature maps.

[0139] The results of length and width compression twice, three times, four times, and five times are taken out, which are p1, p2, p3, and p4 respectively, to construct the feature pyramid structure, such as Fig.15 shown.

[0140] The candidate regions are generated through RPN, and classification and bounding box regression are performed to determine whether the captured content contains the target, and the suggestion box is adjusted to achieve pixel-level object segmentation.

[0141] The candidate region is generated using the following formula:

[0142] a x，y =(x+w a ·r·cos(θ),y+h a·r·sin(θ))

[0143] w a and h a are the width and height of the candidate region, r and θ are the scale and rotation angle respectively.

[0144] Use the labelme package in Python to annotate the obtained pixel-level images. After completing the annotation of all images, use the labelme2coco package in Python to convert the labelme format dataset into COCO format to obtain the training model.

[0145] S3122: Based on the captured video, the processor 20 extracts key frames from the video to obtain multiple navigation mark images, and uses the artificial intelligence Mask R-CNN algorithm to perform instance segmentation. Then, based on the generated training model, the processor 20 performs real-time navigation mark light detection on the image extracted from the video and generates a COCO format image, thereby achieving pixel-level object segmentation.

[0146] S313: performing image processing;

[0147] Calculate the stroboscopic period: Based on the captured video, use the CMOS image sensor to convert the light signal of the beacon into an electrical signal; use the A / D converter to convert the electrical signal into a digital signal;

[0148] Calculate the time interval: The digital signal is processed by the image processing algorithm to obtain the illumination and chromaticity of the light source; the frequency and duration of the flash signal are detected to calculate the interval between flashes;

[0149] Calculate the light intensity: The laser rangefinder 28 of the inspection device collects the relative distance information between the inspection device and the navigation light as the distance parameter l, and then calculates the light intensity:

[0150] I0=El 2 cosθ

[0151] Where I0 is the light intensity of the beacon light being measured; E is the illuminance on the receiving surface of the illuminometer; l is the measuring distance; θ is the angle between the light beam and the normal of the receiving surface of the illuminometer.

[0152] S314: Output the result to the display screen 21.

[0153] S32: When the navigation mark to be measured is an acoustic navigation mark, an acoustic detection module is used for detection;

[0154] S321: Turn on the audio detection switch button 30;

[0155] S322: Turn on the test release switch of the acoustic beacon under test;

[0156] S323: The sound pressure sensor 301 collects sound, and the detection unit 302 performs digital signal (DSP) algorithm processing on the collected sound signal. The process includes sampling, quantization, time domain processing (filtering), frequency domain processing (Fourier transform), and feature extraction. The calculation formula for frequency domain processing is:

[0157]

[0158] Among them, F(ω) is a complex function in the frequency domain, f(t) is a function in the time domain, ω represents the frequency, and e -jωt is a complex exponential function. This formula decomposes the time domain signal f(t) into a series of complex amplitude and phase combinations that describe the contribution of the signal at different frequencies.

[0159] SS24: The extracted features are output to the display screen. Based on the displayed results, the staff determines whether they are consistent with the data published when the beacon was originally shipped from the factory, thereby determining whether the beacon is in a valid state.

[0160] S33: When the measured beacon is a radar beacon, a radar detection module is used for detection;

[0161] S331: Turn on the radar detection switch button 31;

[0162] S332: Start detection and output the detection result to the display screen 21;

[0163] a) When the measured beacon is a radar transponder:

[0164] a1): Measure whether the frequency bandwidth range is normal: within the frequency range of the radar transponder, the radar transmitting device 24 sends a radar signal of the corresponding frequency. If the radar receiving device 25 can receive the response signal of the measured navigation mark at any frequency point within the bandwidth, it means that the navigation mark meets the frequency response requirements and is in normal working condition;

[0165] a2): Measure the effective distance: According to the effective distance published by the navigation mark, perform a pull-distance test and move the inspection device to the maximum effective distance; determine whether the radar receiving device 25 can receive the radar signal fed back by the navigation mark;

[0166] a3): Measure the agile frequency: change the transmission frequency. If the received signal frequency also changes and the frequencies are the same, it is determined that the radar transponder under test has the agile frequency characteristic.

[0167] a4): Measure characteristic values: According to the obtained signal packet, obtain the characteristic values ​​contained in the signal;

[0168] a5): Compare the measurement result with the data published by the navigation mark to determine whether the navigation mark is qualified, and output the result to the display screen 21.

[0169] b) When the measured beacon is a radar reflector:

[0170] b1): The radar transmitting device 24 transmits a 9.4Ghz (X-band) radar signal;

[0171] b2): Determine whether the radar receiving device 25 can receive the reflected radar signal;

[0172] b3): If the reflected signal can be received, the radar reflection area RCS is calculated and compared with the data published by the beacon to determine whether the beacon is qualified and output the result to the display; if the reflected signal cannot be received, it is directly judged as unqualified and the result is output to the display.

[0173] The calculation formula for the spherical radar reflector RCS is:

[0174] RCS=4π×r 2 / λ(m 2 )

[0175] r is the radius of the sphere, λ is the radar wavelength

[0176] The calculation formula for the RCS of a square radar reflector is:

[0177] RCS=4πLC / λ 2 (m 2 )

[0178] L, C are the length and width of the rectangle, λ is the radar wavelength;

[0179] The calculation formula for the RCS of a cylindrical radar reflector is:

[0180] RCS=2πrh 2 / λ(m 2 )

[0181] r is the radius of the cylinder, h is the height of the cylinder, and λ is the radar wavelength;

[0182] The calculation formula for the RCS of a triangular radar reflector is:

[0183] RCS=4πL 4 / 3λ 2 (m 2 )

[0184] Where L is the side length of the triangle and λ is the radar wavelength.

[0185] c) When the measured navigation mark is a radar beacon:

[0186] c1): Measure the effective distance: According to the effective distance published by the beacon, perform a pull-distance test and move the inspection device to the maximum effective distance; determine whether the radar receiving device 25 can receive the radar signal sent by the beacon;

[0187] c2): Measurement frequency range: Check whether the received signal is within the X-band range (9.3Ghz~9.5Ghz). If it is within this band, it proves that the measured beacon is working normally; if not, it proves that it is working abnormally.

[0188] c3): Output C2 result to display screen 21.

[0189] The embodiments of the present invention disclose preferred embodiments, but are not limited thereto. A person skilled in the art can easily understand the spirit of the present invention based on the above embodiments and make different extensions and changes. However, as long as they do not deviate from the spirit of the present invention, they are all within the protection scope of the present invention.

Claims

1. A multi-type navigation mark inspection device based on artificial intelligence, comprising an inspection device body (5), characterized in that: An air supply box (6) is arranged at the bottom of the inspection device body (5); a box cover (7) fixedly connected to the inspection device body (5) is fixedly connected to the inner wall of the air supply box (6); a first air outlet (8) surrounding the inspection device body (5) is provided on the upper surface of the box cover (7); a fixed pipe (9) connected to the air supply box (6) is fixedly connected to the upper surface of the box cover (7); a second air outlet (10) facing the inspection device body (5) is provided on the surface of the fixed pipe (9); the first air outlet (8) and the second air outlet (10) cooperate to form an air curtain to wrap the inspection device body (5); an air supply assembly (1) is fixedly connected to the lower surface of the air supply box (6); the air supply assembly (1) supplies dry low-salinity air into the air supply box (6); The lower surface of the air supply assembly (1) is fixedly connected to a triangular bracket, and the triangular bracket has a tray (11), and the surface of the tray (11) is rotatably connected to three support legs (12), and the bottom ends of the support legs (12) are hemispherical. The lower surface of the tray (11) is fixedly connected to a loading portion (2), and a counterweight (13) is arranged in the loading portion (2); The bottom end of the loading portion (2) is provided with an end cover (14), the upper surface of the end cover (14) is fixedly connected to an externally threaded tube (15), the inner wall of the loading portion (2) is provided with an internal thread that matches the externally threaded tube (15), the externally threaded tube (15) is threadedly connected to the loading portion (2), the surface of the end cover (14) is fixedly connected to a clamping block (16), and the surface of the clamping block (16) is provided with a clamping groove corresponding to the supporting leg (12), and the clamping block (16) is elastic; The loading portion (2) comprises an upper connecting tube (201) fixedly connected to the lower surface of the tray (11); the bottom end of the upper connecting tube (201) is rotatably connected to a driving member (202); the driving member (202) is used to drive the air supply assembly (1) to operate; the bottom end of the driving member (202) is rotatably connected to a lower connecting tube (203); the counterweight (13) is disposed in the lower connecting tube (203); and the external threaded tube (15) is threadedly connected to the lower connecting tube (203); The driving member (202) comprises a rotating column (2021), the two ends of which are rotatably connected to the upper connecting tube (201) and the lower connecting tube (203), respectively; a concave mounting groove is formed on the surface of the rotating column (2021), and a plurality of rotating blades (2022) are rotatably connected to the inner wall of the mounting groove; the rotating blades (2022) can drive the rotating column (2021) to rotate under the influence of external wind force; a transmission shaft (2023) is fixedly connected to the top end of the rotating column (2021); the transmission shaft (2023) is connected to the air supply component (1) and drives the air supply component (1) to operate.

Images