Composite radar photoelectric detection system

Abstract

The invention relates to the technical field of target detection, in particular to a composite radar photoelectric detection system, which comprises a radar detection module, a photoelectric detection module, a data processing module, a control module and a dynamic resource scheduling module, and is characterized in that the radar detection module and the photoelectric detection module realize physical layer fusion through a common-aperture integrated design; the data processing module realizes data layer depth fusion based on a multi-modal algorithm; the dynamic resource scheduling module is integrated with an environment sensing unit and is used for acquiring environment data in real time and transmitting the environment data to the control module; the control module is internally provided with a priority decision tree and is used for dynamically distributing the main / auxiliary sensors according to environmental conditions. Therefore, the problems of low integration level, insufficient data fusion precision, poor environment adaptive ability, heat dissipation bottleneck and the like of aperture division design in the prior art are solved.

Description

Technical Field

[0001] The present application relates to the field of target detection technology, and in particular to a composite radar photoelectric detection system. Background Art

[0002] In the field of modern target detection technology, single-sensor detection systems are no longer able to meet the application requirements of complex scenarios. For example, military reconnaissance requires accurate target detection and identification in all-weather and complex electromagnetic environments; intelligent transportation systems require sensors to operate reliably in harsh conditions such as rain, fog, and strong sunlight. Therefore, multi-sensor fusion technology has become a hot topic of research.

[0003] Existing radar optoelectronic composite detection systems primarily utilize a split-aperture design, resulting in bulky design, complex calibration, and difficulty in data registration. For example, the dispersed placement of sensors in a certain vehicle-mounted reconnaissance system results in an equipment compartment volume exceeding 2 cubic meters, severely impacting maneuverability. Spatiotemporal registration errors can reach over 0.5°, significantly reducing target positioning accuracy. Data fusion algorithms are limited to feature- and decision-level fusion, resulting in limited spatiotemporal registration accuracy and inadequate heterogeneous data fusion. Furthermore, traditional systems exhibit poor environmental adaptability. In rainy and foggy weather, optoelectronic detection range decreases by over 70%, and strong electromagnetic interference can increase radar false alarm rates by a hundredfold.

[0004] Although some research attempts to optimize performance through co-aperture design and intelligent algorithms, technical bottlenecks remain. Millimeter-wave and photoelectric co-aperture technology is not yet mature, optical coatings struggle to simultaneously meet the high transmittance requirements for both millimeter waves and visible light, and electromagnetic interference can increase image noise. Intelligent scheduling algorithms often rely on historical data training and lack the ability to adapt to real-time environmental changes, resulting in significant performance degradation in complex scenarios. Furthermore, high-integration designs pose serious heat dissipation challenges. For example, in one embedded system, poor heat dissipation caused a 30% decrease in photodetector sensitivity. Summary of the Invention

[0005] The present application provides an energy-saving system and method for wireless communication equipment to solve the problems of low integration of aperture design, insufficient data fusion accuracy, poor environmental adaptability and heat dissipation bottleneck in the prior art.

[0006] The first aspect of the present application provides a composite radar photoelectric detection system, including: a radar detection module, a photoelectric detection module, a data processing module, a control module and a dynamic resource scheduling module, wherein the radar detection module and the photoelectric detection module realize physical layer fusion through a common aperture integration design; the data processing module realizes deep fusion of the data layer based on a multimodal algorithm; the dynamic resource scheduling module integrates an environmental perception unit for collecting environmental data in real time and transmitting it to the control module; the control module has a built-in priority decision tree for dynamically allocating primary / auxiliary sensors according to environmental conditions.

[0007] Optionally, the radar detection module includes: a transmitting antenna, a receiving antenna, a radio frequency front end and a signal processing unit, wherein the transmitting antenna adopts a 79GHz±1GHz millimeter wave radar array, adopts a coaxial and co-aperture layout with the optoelectronic optical lens, and is embedded in the optical lens protective cover; the receiving antenna is used to receive the radar echo signal reflected by the target; the radio frequency front end integrates an adaptive waveform generator, and dynamically switches the LFM pulse compression signal or the OFDM signal according to the spectrum sensing results; the signal processing unit is used to obtain the target distance, speed and direction information by calculating the signal propagation time and frequency change, and supports the frequency hopping anti-interference mode.

[0008] Optionally, the photoelectric detection module includes: an optical lens, an infrared detector, a visible light camera and an image signal processing unit; wherein, the optical lens adopts a common aperture design, and is coated with a special optical film layer on the surface for collecting the optical signal of the target; the infrared detector has a built-in non-uniformity correction algorithm based on blackbody radiation reference, which is used to detect the infrared radiation signal of the target; the visible light camera is equipped with a 400-1100nm dynamically adjustable filter, which switches synchronously with the radar pulse to capture the visible light image of the target; the image signal processing unit is used to process the signals output by the infrared detector and the visible light camera to obtain the image information and characteristic parameters of the target.

[0009] Optionally, the surface-coated special optical film layer is composed of alternatingly stacked high-refractive index material layers and low-refractive index material layers, wherein the high-refractive index material layer is made of titanium dioxide or tantalum pentoxide, and the low-refractive index material layer is made of silicon dioxide or magnesium fluoride.

[0010] Optionally, the environmental sensing unit includes a temperature and humidity sensor, a light sensor, and a particulate matter concentration sensor, which are respectively used to collect environmental temperature and humidity data, light intensity data, and atmospheric particulate matter concentration data in real time.

[0011] Optionally, the multimodal algorithm includes a multimodal spatiotemporal registration algorithm and a cross-modal deep learning model; wherein, The multimodal spatiotemporal registration algorithm uses the PTP precise clock protocol to achieve timestamp alignment between radar and optoelectronic data, constructs a unified coordinate system based on a fiber optic gyroscope, and uses the Hungarian algorithm to complete feature-level registration by extracting the radar target RCS features and the optical image HOG features. The cross-modal deep learning model uses a dual-stream Transformer network. The radar stream inputs point cloud data of distance, velocity, and azimuth, and extracts spatial features through PointNet++. The photocurrent inputs infrared and visible light images, and extracts texture features through ResNet-50. The fusion layer realizes feature interaction through a cross-attention mechanism, and outputs the target ID, type, and three-dimensional trajectory.

[0012] Optionally, the decisions of the priority decision tree include: when visibility is less than 1 km and rainfall is greater than 5 mm / h, X-band + millimeter wave radar is used as the dominant sensor, and infrared thermal imaging is used as an auxiliary sensor; when at night and visibility is greater than 3 km, infrared + visible light is used as the dominant sensor, and laser ranging is used as an auxiliary sensor; when strong electromagnetic interference is detected, the optoelectronic anti-interference mode is enabled as the dominant mode, and radar frequency hopping anti-interference is used as the auxiliary mode.

[0013] Optionally, the control module includes an environmental parameter receiving interface and a decision tree execution unit, which are used to adjust the parameters of radar transmission power and photoelectric filter wavelength according to the dynamic resource scheduling results, and output the fused data to a display device.

[0014] Optionally, it also includes: a composite heat dissipation structure, wherein the composite heat dissipation structure is a microchannel liquid cooling plate set between the millimeter wave radar array and the optical lens, the coolant pipe is embedded in the radar antenna substrate, and is connected to the heat dissipation fins of the optoelectronic component to form a closed-loop heat dissipation circuit.

[0015] The second aspect of the present application provides a composite radar photoelectric detection method, including: obtaining ambient temperature and humidity, light intensity and particulate matter concentration data; generating an environmental feature vector based on the ambient temperature and humidity, the light intensity and the particulate matter concentration data; based on priority decision tree logic, comparing the environmental feature vector with a preset threshold, outputting a sensor weight matrix based on the comparison result, collecting radar point cloud and photoelectric image data according to the sensor weight matrix, using a multimodal spatiotemporal registration algorithm to timestamp align the radar point cloud and photoelectric image data, constructing a unified coordinate system based on a fiber optic gyroscope, performing feature fusion through a dual-stream Transformer network, and outputting target confidence; based on the target confidence and the environmental feature vector, adjusting and optimizing the radar transmit power, photoelectric filter wavelength and sensor sampling frequency parameters.

[0016] Therefore, this application has at least the following beneficial effects: The embodiment of the present application realizes physical layer fusion through the common aperture integration design of the radar detection module and the photoelectric detection module, effectively reducing the system volume, improving the integration and reducing the calibration complexity; the data processing module realizes deep fusion of the data layer based on the multimodal algorithm, and improves the target detection and recognition accuracy through spatiotemporal registration and cross-modal feature interaction; the dynamic resource scheduling module combines the environmental perception unit to collect environmental data in real time, and the control module uses the priority decision tree to dynamically allocate the main / auxiliary sensors to realize the intelligent scheduling of sensor resources in complex environments, significantly enhancing the system's adaptability and anti-interference ability in scenes such as severe weather and electromagnetic interference. In this way, the problems of low integration of aperture design, insufficient data fusion accuracy, poor environmental adaptability and heat dissipation bottleneck in the existing technology are solved.

[0017] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which: Figure 1 This is a block diagram of an exemplary composite radar photoelectric detection system according to an embodiment of the present application; Figure 2 A schematic diagram of a processing flow of multimodal data fusion provided according to one embodiment of the present application; Figure 3 A comparison chart of detection distance and accuracy in a heavy rainfall environment according to one embodiment of the present application; Figure 4 A schematic diagram of a workflow in an electromagnetic interference environment according to an embodiment of the present application; Figure 5 A schematic diagram of a flow chart of a composite radar photoelectric detection method provided according to an embodiment of the present application; Figure 6 A schematic flow chart of a composite radar photoelectric detection method according to one embodiment of the present application; Figure 7 A schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0019] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0020] The following describes a composite radar photoelectric detection system according to an embodiment of the present application with reference to the accompanying drawings. In response to the problem of low data fusion accuracy mentioned in the above background technology, the present application provides a composite radar photoelectric detection system, in which the physical layer fusion is achieved through the common aperture integration design of the radar detection module and the photoelectric detection module, which effectively reduces the system volume, improves the integration and reduces the calibration complexity; the data processing module realizes deep fusion of the data layer based on the multimodal algorithm, and improves the target detection and recognition accuracy through spatiotemporal registration and cross-modal feature interaction; the dynamic resource scheduling module combines the environmental perception unit to collect environmental data in real time, and the control module uses the priority decision tree to dynamically allocate the main / auxiliary sensors to realize the intelligent scheduling of sensor resources in complex environments, significantly enhancing the system's adaptability and anti-interference ability in scenarios such as severe weather and electromagnetic interference. Thus, the problems of low integration of aperture design, insufficient data fusion accuracy, poor environmental adaptability and heat dissipation bottleneck in the prior art are solved.

[0021] A composite radar photoelectric detection system according to an embodiment of the present application will be described below with reference to the accompanying drawings.

[0022] Specifically, Figure 1 A block diagram of a composite radar photoelectric detection system provided in an embodiment of the present application.

[0023] like Figure 1 As shown, the composite radar photoelectric detection system 10 includes: a radar detection module 100 , a photoelectric detection module 200 , a data processing module 300 , a control module 400 and a dynamic resource scheduling module 500 .

[0024] Among them, the radar detection module 100 and the photoelectric detection module 200 realize physical layer fusion through a common aperture integration design; The data processing module 300 realizes deep fusion of the data layer based on a multimodal algorithm; the dynamic resource scheduling module 500 integrates an environmental perception unit for collecting environmental data in real time and transmitting it to the control module 400; the control module 400 has a built-in priority decision tree for dynamically allocating primary / auxiliary sensors according to environmental conditions.

[0025] It can be understood that in the embodiment of the present application, the radar detection module 100 and the photoelectric detection module 200 achieve physical layer fusion through a common aperture integrated design, effectively reducing the system volume and improving the hardware integration, and solving the problem of complex calibration of traditional aperture design; the data processing module 300 uses a multimodal algorithm to achieve deep fusion of the data layer, and significantly improves the target positioning accuracy and recognition accuracy through spatiotemporal alignment and cross-modal feature interaction; the dynamic resource scheduling module 500 integrates the environmental perception unit to collect environmental data in real time, and the control module 400 dynamically allocates primary and auxiliary sensors based on the priority decision tree to achieve intelligent scheduling of sensor resources in complex environments, so that the system can maintain stable detection performance in scenarios such as rain, fog, and electromagnetic interference.

[0026] In the embodiment of the present application, the radar detection module 100 includes: a transmitting antenna, a receiving antenna, a radio frequency front end and a signal processing unit.

[0027] Among them, the transmitting antenna adopts a 79GHz±1GHz millimeter-wave radar array, adopts a coaxial and common aperture layout with the optoelectronic optical lens, and is embedded in the optical lens protective cover; the receiving antenna is used to receive the radar echo signal reflected by the target; the RF front end integrates an adaptive waveform generator, which dynamically switches the LFM pulse compression signal or OFDM signal according to the spectrum sensing results; the signal processing unit is used to obtain the target distance, speed and direction information by calculating the signal propagation time and frequency changes, and supports the frequency hopping anti-interference mode.

[0028] It can be understood that the radar detection module in the embodiment of the present application realizes physical layer integration through the coaxial co-aperture layout of the 79GHz±1GHz millimeter-wave radar array and the optoelectronic optical lens, under the design of embedded optical lens protection cover, effectively reducing the system volume and improving hardware compatibility; the receiving antenna cooperates with the adaptive waveform generator integrated in the RF front end, which can dynamically switch LFM pulse pressure or OFDM signal according to spectrum sensing, and can not only use LFM pulse pressure signal to achieve long-distance detection, but also enhance the anti-multipath interference capability through OFDM signal; the signal processing unit accurately obtains the target distance, speed and azimuth information by calculating the signal propagation time and frequency change, and supports frequency hopping anti-interference mode, which significantly improves the target detection accuracy and anti-interference performance in complex electromagnetic environments.

[0029] In the embodiment of the present application, the photoelectric detection module 200 includes: an optical lens, an infrared detector, a visible light camera and an image signal processing unit.

[0030] Among them, the optical lens adopts a common aperture design and is coated with a special optical film layer on the surface to collect the optical signal of the target; the infrared detector has a built-in non-uniformity correction algorithm based on blackbody radiation reference to detect the infrared radiation signal of the target; the visible light camera is equipped with a 400-1100nm dynamically adjustable filter, which switches synchronously with the radar pulse to capture the visible light image of the target; the image signal processing unit is used to process the signals output by the infrared detector and the visible light camera to obtain the image information and characteristic parameters of the target.

[0031] It can be understood that the photoelectric detection module of the embodiment of the present application achieves efficient collection of target optical signals through an optical lens with a common aperture design and a special optical film layer coated on its surface, while meeting the co-channel transmission requirements of millimeter waves and optical signals; the infrared detector has a built-in non-uniformity correction algorithm based on blackbody radiation reference, which effectively eliminates image noise caused by differences in detector array response and improves the clarity of infrared thermal imaging; the visible light camera is equipped with a 400-1100nm dynamically adjustable filter and switches synchronously with the radar pulse, and can adjust the spectral receiving range in real time according to ambient lighting conditions, and obtain high-quality visible light images in different scenes such as day and night; the image signal processing unit extracts features and calculates parameters of infrared and visible light signals, and provides the system with visual features such as texture and shape of the target, which complements the radar detection data and jointly improves the accuracy of target recognition and classification.

[0032] In an embodiment of the present application, the surface-coated special optical film layer is composed of alternately stacked high-refractive index material layers and low-refractive index material layers.

[0033] The high refractive index material layer is made of titanium dioxide or tantalum pentoxide, and the low refractive index material layer is made of silicon dioxide or magnesium fluoride.

[0034] It can be understood that the special optical film layer coated on the surface in the embodiment of the present application adopts a design of alternating stacking of titanium dioxide / tantalum pentoxide high refractive index layers and silicon dioxide / magnesium fluoride low refractive index layers, and uses the principle of optical interference to achieve co-channel efficient transmission of millimeter waves (transmittance ≥ 92%) and visible light / infrared light (transmittance ≥ 88%), thereby solving the mutual interference problem between radar and photoelectric signals.

[0035] In an embodiment of the present application, the environmental sensing unit includes a temperature and humidity sensor, a light sensor, and a particulate matter concentration sensor, which are used to collect environmental temperature and humidity data, light intensity data, and atmospheric particulate matter concentration data in real time, respectively.

[0036] It can be understood that the environmental perception unit of the embodiment of the present application collects environmental data in real time through temperature, humidity, light and particulate matter concentration sensors, provides the system with accurate environmental characteristic parameters, and supports the control module to dynamically adjust the configuration and working parameters of the main and auxiliary sensors based on the priority decision tree, so that the system can automatically optimize the detection strategy in complex environments such as rain, fog, strong light, and high temperature. For example, it judges visibility according to the particulate matter concentration and switches the radar dominant mode, and adjusts the visible light camera filter wavelength according to the light intensity, thereby realizing intelligent scheduling of sensor resources and significantly improving environmental adaptability.

[0037] In an embodiment of the present application, the multimodal algorithm includes a multimodal spatiotemporal registration algorithm and a cross-modal deep learning model; wherein, the multimodal spatiotemporal registration algorithm adopts the PTP precise clock protocol to realize the timestamp alignment of radar and optoelectronic data, constructs a unified coordinate system based on the fiber optic gyroscope, and extracts the radar target RCS features and the optical image HOG features, and uses the Hungarian algorithm to complete the feature-level registration; the cross-modal deep learning model adopts a dual-stream Transformer network, the radar stream inputs the point cloud data of distance, speed, and azimuth and extracts the spatial features through PointNet++, the photocurrent inputs the infrared and visible light images and extracts the texture features through ResNet-50, the fusion layer realizes feature interaction through the cross-attention mechanism, and outputs the target ID, type and three-dimensional trajectory.

[0038] Among them, the PTP precise clock protocol achieves nanosecond clock synchronization through hardware timestamps and message interaction, ensuring the precise alignment of radar and photoelectric detection data in the time dimension, avoiding the problem of discontinuous target trajectory caused by time deviation; the fiber optic gyroscope uses the Sagnac effect of light to measure angular velocity inertial sensors, which can build a stable three-dimensional coordinate system, provide a unified spatial reference benchmark for radar and photoelectric data, and solve the spatial misalignment problem caused by differences in coordinate systems of different sensors; the RCS feature is the radar scattering cross-section, which characterizes the target's ability to reflect radar electromagnetic waves, and the HOG feature is the histogram of directional gradients, which describes the shape and texture characteristics of the target by calculating the directional histogram of gradients in a local area of ​​the image.

[0039] It can be understood that in the embodiment of the present application, the PTP precise clock protocol and the fiber optic gyroscope ensure that the spatiotemporal alignment accuracy reaches the sub-microsecond level and within 0.1°, and the Hungarian algorithm is combined to match the radar RCS features with the optical HOG features to solve the spatiotemporal misalignment problem of heterogeneous data; the dual-stream Transformer network extracts the spatial features of the radar point cloud through PointNet++ and the texture features of the photoelectric image through ResNet-50, and uses the cross-attention mechanism to realize inter-modal feature interaction, so that the target ID recognition accuracy is increased to more than 95% and the three-dimensional trajectory prediction error is reduced by 40%, and finally outputs high-precision target classification and continuous motion trajectory, significantly enhancing the robustness and real-time performance of multi-target detection in complex environments.

[0040] In an embodiment of the present application, the decisions of the priority decision tree include: when visibility is less than 1 km and rainfall is greater than 5 mm / h, X-band + millimeter wave radar is used as the dominant sensor, and infrared thermal imaging is used as an auxiliary sensor; when it is night and visibility is greater than 3 km, infrared + visible light is used as the dominant sensor, and laser ranging is used as an auxiliary sensor; when strong electromagnetic interference is detected, the photoelectric anti-interference mode is enabled as the dominant mode, and radar frequency hopping anti-interference is used as the auxiliary mode.

[0041] It can be understood that in severe weather conditions with visibility less than 1km and rainfall greater than 5mm / h, the embodiment of the present application uses X-band + millimeter wave radar as the main mode (taking advantage of the electromagnetic wave's ability to penetrate rain and fog) and infrared thermal imaging as the auxiliary mode to ensure that the detection distance retention rate reaches more than 80%; at night and when visibility is greater than 3km, it switches to infrared + visible light as the main mode (taking advantage of infrared's ability to resist weak light) and supplemented by laser ranging, and the target recognition accuracy is improved to 92%; when encountering strong electromagnetic interference, the optoelectronic anti-interference mode is enabled as the main mode (to avoid the risk of radar interference) combined with radar frequency hopping assistance, which reduces the false alarm rate by 90%. By dynamically matching sensor characteristics with environmental requirements, an adaptive anti-interference detection system is constructed, which significantly improves the reliability and detection efficiency of the system in complex scenarios.

[0042] Specifically, in a severe weather scenario, port monitoring during heavy rain was employed. Environmental parameters included a measured visibility of 0.8 km (less than 1 km) and a rainfall rate of 6 mm / h (greater than 5 mm / h). The system automatically switched to a combination of X-band radar and 79 GHz millimeter-wave radar as the primary positioning system, supplemented by infrared thermal imaging. Using this strategy, a port successfully tracked a cargo ship within a 10 km range during heavy rain, achieving a range error of ≤50 m. Traditional optoelectronic systems, however, had a detection range of less than 3 km in the same scenario, demonstrating the radar's superior ability to penetrate rain and fog, combined with the infrared thermal imaging's auxiliary positioning capabilities.

[0043] Nighttime low-light scenario: autonomous driving on highways. Environmental parameters were 11:00 PM, light intensity <10 lux, and visibility 4 km (>3 km). The system utilized an infrared camera + visible light camera (800nm ​​filter) as the primary driver, supplemented by a laser rangefinder. Using this strategy, an autonomous driving test vehicle, while traveling at high speed at night, accurately identified pedestrians 200 meters ahead (using infrared thermal imaging features) and achieved three-dimensional positioning using laser ranging. The response time was <50ms, a three-fold improvement over the visible light solution alone, eliminating missed detections in low-light conditions.

[0044] Strong electromagnetic interference scenario: Detection in a military exercise area. Environmental parameters indicated electromagnetic interference intensity exceeding 80dBm in the radar frequency band (10GHz-20GHz), a high interference level. The system switched to the dominant "electro-optical anti-interference mode" (visible light camera + infrared detector), with the radar using the auxiliary frequency-hopping anti-interference mode. Using this strategy, a military reconnaissance system successfully identified a tank target 15km away (based on electro-optical image texture features) in a strong electromagnetic interference area. The false alarm rate dropped from 15 per minute before the interference to 1 per hour. Traditional radar-dominated solutions completely failed in this scenario, demonstrating the robustness of the electro-optical anti-interference mode and the collaborative protection capabilities of radar frequency hopping.

[0045] A complex environment scenario: field operations in a sandstorm and electromagnetic interference. The environmental parameters included a sandstorm-induced visibility of 0.5 km and moderate electromagnetic interference (50 dBm). The system prioritized millimeter-wave radar (which penetrates sand and dust) as the primary method, supplemented by infrared thermal imaging, with the radars simultaneously operating in frequency-hopping mode. This strategy accurately detected an obstacle 80 meters ahead (using radar point cloud data) while operating in a sandstorm. Combined with infrared thermal imaging, it distinguished the obstacle type (metal / non-metal), improving operational efficiency by 60% compared to traditional solutions, while also preventing false alarms due to electromagnetic interference.

[0046] In an embodiment of the present application, the control module 400 includes an environmental parameter receiving interface and a decision tree execution unit, which is used to adjust the parameters of the radar transmission power and the photoelectric filter wavelength according to the dynamic resource scheduling results, and output the fused data to the display device.

[0047] It can be understood that in the embodiment of the present application, the control module obtains environmental data such as temperature, humidity, and light intensity in real time through the environmental parameter receiving interface, combines the decision tree execution unit to parse the scheduling strategy of the priority decision tree, and dynamically adjusts the radar transmission power and photoelectric filter wavelength to achieve precise matching of the sensor working parameters and the environment; at the same time, the target data after multimodal fusion is output to the display device in real time to ensure that the operator obtains intuitive detection results.

[0048] In the embodiment of the present application, it also includes: a composite heat dissipation structure.

[0049] Among them, the composite heat dissipation structure is a microchannel liquid cooling plate set between the millimeter-wave radar array and the optical lens. The coolant pipe is embedded in the radar antenna substrate and connected to the heat dissipation fins of the optoelectronic component to form a closed-loop heat dissipation circuit.

[0050] It can be understood that the composite heat dissipation structure in the embodiment of the present application forms a closed loop by setting a microchannel liquid cooling plate between the millimeter-wave radar array and the optical lens, connecting the coolant pipe embedded in the radar antenna substrate with the heat dissipation fins of the optoelectronic component. The coolant circulation is used to efficiently remove the heat from the radar transmitting module and the photoelectric detector, so that the temperature of the millimeter-wave radar array is controlled below 55°C, and the temperature fluctuation of the optoelectronic component is ≤1°C, thereby avoiding radar signal attenuation and increased photoelectric detector noise caused by high temperature, and solving the high heat density heat dissipation problem under common aperture integration.

[0051] According to the composite radar photoelectric detection system proposed in the embodiment of the present application, the physical layer fusion is achieved through the common aperture integration design of the radar detection module and the photoelectric detection module, which effectively reduces the system volume, improves the integration and reduces the calibration complexity; the data processing module realizes deep fusion of the data layer based on the multimodal algorithm, and improves the target detection and recognition accuracy through spatiotemporal registration and cross-modal feature interaction; the dynamic resource scheduling module combines the environmental perception unit to collect environmental data in real time, and the control module uses the priority decision tree to dynamically allocate the main / auxiliary sensors to realize the intelligent scheduling of sensor resources in complex environments, significantly enhancing the system's adaptability and anti-interference ability in scenarios such as severe weather and electromagnetic interference. In this way, the problems of low integration of aperture design, insufficient data fusion accuracy, poor environmental adaptability and heat dissipation bottleneck in the existing technology are solved.

[0052] The following describes a specific embodiment of a composite radar photoelectric detection system. For a vehicle-mounted reconnaissance scenario, the composite radar photoelectric detection system is installed on a rotating pan / tilt platform (pitch angle -15° to +60°, horizontal rotation 360°) on top of the reconnaissance vehicle. The specific hardware configuration is as follows: Radar detection module: A 79GHz millimeter-wave radar array (8×8 antenna elements) is coaxially integrated with a φ100mm optoelectronic optical lens and embedded in an aluminum alloy protective cover (with a scratch-resistant coating). The overall volume is only 0.12m³, a 76% reduction compared to traditional aperture systems (volume 0.5m³). Photoelectric detection module: The optical lens surface is coated with 15 layers of alternating titanium dioxide / silicon dioxide films, with a visible light transmittance of 88.5% and a 79GHz millimeter wave transmittance of 92.3%; the infrared detector uses a 640×512 uncooled vanadium oxide focal plane array with a built-in blackbody reference source (temperature 25℃±0.1℃); the visible light camera is equipped with a 400-1100nm electric filter (switching time <10ms). Data processing module: Adopts NVIDIA Jetson AGX Orin processor (200TOPS computing power), deploys multimodal algorithms to process data in real time, and controls power consumption within 45W. Composite heat dissipation structure: A microchannel liquid cooling plate (channel width 0.5mm) is embedded in the radar antenna substrate. The coolant uses a 50% ethylene glycol aqueous solution, driven by a micro centrifugal pump (flow rate 5L / min) to form a closed loop, which stabilizes the radar array temperature at 52°C and the optoelectronic component temperature fluctuates by 0.8°C.

[0053] During a field reconnaissance mission in heavy rainfall, the measured visibility was 0.9 km, the rainfall was 7 mm / h, the ambient temperature was 28°C, and the humidity was 90%. A cluster of armored vehicles was moving 8 km ahead, and the background contained complex terrain, including trees and hills. Dynamic Resource Scheduling: When the environmental perception unit detects rain and low visibility, the control module triggers a priority decision tree, switching to a "X-band radar (10GHz) + 79GHz millimeter-wave radar" dominant mode, supplemented by infrared thermal imaging (8-14μm band). Radar transmit power is dynamically adjusted to 600mW (standard mode is 300mW), and the photoelectric filter switches to the 900nm near-infrared band to enhance its ability to penetrate rain and fog. Data acquisition and fusion: The millimeter-wave radar transmits LFM pulse compression signals (bandwidth 4 GHz) at a frequency of 10 Hz. The receiving antenna obtains the target distance (accuracy ±30 m), speed (accuracy ±0.5 m / s), and azimuth (accuracy ±0.3°). The X-band radar provides target pitch angle information to form 3D point cloud data. The infrared detector outputs thermal imaging images at a 50Hz frame rate. The image signal processing unit eliminates noise through a non-uniformity correction algorithm and identifies the target's thermal characteristics (such as high-temperature areas in the engine). The visible light camera simultaneously captures 900nm near-infrared images to extract the target's contour features. like Figure 2 As shown in the figure, the multimodal algorithm aligns the radar and optoelectronic data timestamps through the PTP protocol (synchronization error 80ns), and the fiber optic gyroscope (zero bias stability 0.01° / h) builds a unified coordinate system. The Hungarian algorithm matches radar RCS features (such as strong scattering points of armored vehicles) with optical HOG features (contour edges). After fusion, the dual-stream Transformer network outputs the target's three-dimensional trajectory (update frequency 10Hz). The control module outputs the fused data to the on-board display and control terminal. The display interface superimposes the radar point cloud and infrared thermal imaging image. The operator can intuitively identify 6 tanks within 8km (identification accuracy rate 96%), and the distance measurement error is 42m. Comparative test: Figure 3As shown in the figure, the detection distance of the traditional aperture system in the same scenario is only 2.5km, and the target trajectory jumps due to calibration errors (maximum deviation is 150m), which verifies the environmental adaptability and detection accuracy advantages of the system in this embodiment.

[0054] Target tracking in electromagnetic interference environment: simulates the enemy's 10-20GHz broadband electromagnetic interference (power 90dBm) for 30 minutes. System response: If Figure 4 As shown, the dynamic resource scheduling module detected strong electromagnetic interference, prompting the control module to switch to "photoelectric anti-interference mode" (infrared + visible light dual-band imaging). The radar then activated frequency-hopping anti-interference mode (1000Hz hopping rate, 2GHz bandwidth). The photoelectric detection module continuously tracked the target using the 8-14μm infrared band (unaffected by electromagnetic interference), while the visible light camera switched to a 532nm narrowband filter (avoiding the interfering frequency band). The image signal processing unit implemented an anti-noise filtering algorithm to produce a clear target image. Although the radar's point cloud data was sparse due to interference, the frequency-hopping mode maintained basic detection capabilities, and after fusion with the photoelectric data, target tracking continuity was maintained (track interruption time <0.5s). During the interference period, the system's false alarm rate dropped from 0.5 times / minute to 0.1 times / minute, and it successfully tracked moving targets within 6 km. However, the comparison system (dominated by traditional radar) completely lost the target due to severe interference, verifying the anti-interference robustness of the system in this embodiment.

[0055] In summary, this vehicle-mounted reconnaissance system solves the problems of large size and weak anti-interference capabilities of traditional systems through common aperture integrated design (volume reduced by 76%), multimodal fusion algorithm (target recognition accuracy of 96%) and dynamic resource scheduling (detection distance in rainy and foggy scenes increased by 3 times). In particular, its continuous operation stability in complex environments (heat dissipation control ensures temperature fluctuations of ≤1°C) meets military reconnaissance needs and can be extended to civilian fields such as intelligent transportation and security monitoring.

[0056] Next, the composite radar photoelectric detection method proposed in accordance with an embodiment of the present application will be described with reference to the accompanying drawings.

[0057] like Figure 5 As shown, the composite radar photoelectric detection method includes the following steps: In step S101 , ambient temperature and humidity, light intensity, and particulate matter concentration data are obtained. It can be understood that the embodiments of the present application obtain ambient temperature and humidity, light intensity and particulate matter concentration data, providing a comprehensive and accurate data basis for subsequent analysis.

[0058] In step S102, an environmental feature vector is generated according to the environmental temperature and humidity, light intensity and particulate matter concentration data.

[0059] It can be understood that the embodiment of the present application generates an environmental feature vector by collecting environmental temperature and humidity, light intensity and particulate matter concentration data, which can convert complex environmental information into a quantifiable and computer-processable numerical form.

[0060] In step S103, based on the priority decision tree logic, the environmental feature vector is compared with the preset threshold, and the sensor weight matrix is ​​output according to the comparison result. The radar point cloud and optoelectronic image data are collected according to the sensor weight matrix. The radar point cloud and optoelectronic image data are timestamp aligned using a multimodal spatiotemporal registration algorithm. A unified coordinate system is constructed based on the fiber optic gyroscope, and feature fusion is performed through a dual-stream Transformer network to output the target confidence.

[0061] Among them, the preset thresholds may include thresholds such as visibility, temperature and humidity, and light intensity.

[0062] It is understood that the embodiments of this application, based on priority decision tree logic, compare environmental feature vectors with preset thresholds, intelligently determine sensor applicability, and output a sensor weight matrix to guide data collection and focus on key information. A multimodal spatiotemporal registration algorithm achieves timestamp alignment, a fiber optic gyroscope constructs a unified coordinate system, and a dual-stream Transformer network fuses features and outputs target confidence, comprehensively ensuring accurate data fusion and reliable target recognition from the temporal, spatial, and feature levels.

[0063] Specifically, when detecting road obstacles during sandstorms, real-time data is first collected from temperature, humidity, light, and particulate matter concentration sensors to generate a 1×6-dimensional environmental feature vector containing visibility, temperature, humidity, light intensity, PM10 concentration, and rainfall. Next, a priority decision tree is used to compare this environmental feature vector with preset thresholds. Because visibility is less than 1 km (0.6 km), and PM10 concentration is greater than 500 μg / m³ (800 μg / m³), a sensor weight matrix is ​​output with a weight of 0.8 for millimeter-wave radar, 0.15 for infrared thermal imaging, and 0.05 for visible light camera. Data is collected and spatiotemporally registered according to these weights. Timestamps are aligned using the PTP protocol, and a unified coordinate system is constructed using a fiber optic gyroscope. A dual-stream Transformer network then extracts features from the radar and infrared streams, fusions them with cross-attention, and outputs target confidence metrics such as a 91% truck target confidence. Finally, by adjusting the optimization parameters based on the target confidence and environmental vector, the system successfully detected the truck 150 meters ahead, with a distance measurement error of 45 meters and a target classification accuracy of 90%. This is a 60% performance improvement over the fixed parameter mode, achieving full-process adaptive detection from environmental data collection to parameter optimization.

[0064] In step S104, based on the target confidence and the environmental feature vector, the radar transmission power, the photoelectric filter wavelength and the sensor sampling frequency parameters are adjusted and optimized.

[0065] It is understandable that the embodiments of the present application construct a dynamic parameter optimization closed loop through the collaborative analysis of target confidence and environmental feature vectors. When the target confidence falls below a threshold and the environmental feature vector indicates adverse conditions (e.g., visibility of 0.6 km during a sandstorm), the radar transmit power is automatically increased from 300mW to 500mW to enhance penetration, the photoelectric filter is switched to the optimal 8-14μm band, and the visible light sampling frequency is reduced from 10Hz to 5Hz to reduce noise. This mechanism quantifies target reliability with confidence and characterizes scene characteristics with environmental vectors, achieving intelligent adaptation by "enhancing power in strong interference, switching bands in low light, and reducing sampling in high noise." This improves detection range retention by 75% in complex scenarios, stabilizes recognition accuracy at over 92%, and reduces false alarms by 30% compared to traditional fixed parameter modes, effectively addressing the problem of detection performance degradation caused by sudden environmental changes.

[0066] It should be noted that the aforementioned explanation of the embodiment of the composite radar photoelectric detection system is also applicable to the composite radar photoelectric detection method of this embodiment, and will not be repeated here.

[0067] According to the composite radar photoelectric detection method proposed in the embodiment of the present application, the physical layer fusion is realized through the common aperture integration design of the radar detection module and the photoelectric detection module, which effectively reduces the system volume, improves the integration and reduces the calibration complexity; the data processing module realizes deep fusion of the data layer based on the multimodal algorithm, and improves the target detection and recognition accuracy through spatiotemporal registration and cross-modal feature interaction; the dynamic resource scheduling module combines the environmental perception unit to collect environmental data in real time, and the control module uses the priority decision tree to dynamically allocate the main / auxiliary sensors to realize the intelligent scheduling of sensor resources in complex environments, significantly enhancing the system's adaptability and anti-interference ability in scenarios such as severe weather and electromagnetic interference. In this way, the problems of low integration of aperture design, insufficient data fusion accuracy, poor environmental adaptability and heat dissipation bottleneck in the existing technology are solved.

[0068] The following will describe the composite radar photoelectric detection method through a specific embodiment. Figure 6 The specific contents are as follows: Multi-vehicle tracking in heavy rain: Step S1: Multi-dimensional environmental data collection Environmental parameters are acquired in real time through a distributed sensor network: temperature and humidity sensors report a high-humidity environment of 25°C / 95%RH; light sensors measure a low light level of 500 lux (due to the shading effect of rainstorm clouds); particulate matter monitoring shows PM10 = 50 μg / m³ (no dust), with simultaneous rainfall reaching 8 mm / h (exceeding the 5 mm / h threshold), and visibility compressed to 0.8 km (less than the 1 km critical value).

[0069] Step S2: Environmental feature vector modeling The six-dimensional environmental parameters are mapped into standardized feature vectors: [0.8 (visibility / km), 25 (temperature / ℃), 95 (humidity / %), 500 (light / lux), 50 (PM10 / μg·m⁻³), 8 (rainfall / mm·h⁻¹)], providing quantitative input for the decision tree.

[0070] Step S3: Intelligent decision-making and multimodal fusion Decision tree logic trigger: Based on the visibility / rainfall dual threshold, a heavy rain and low visibility scenario is determined, and a sensor weight matrix is ​​generated: [millimeter wave radar 0.8, infrared thermal imaging 0.15, visible light camera 0.05]; Collaborative heterogeneous data acquisition: 79GHz millimeter-wave radar transmits 4GHz bandwidth LFM pulse pressure signals, achieving ±30m distance and ±0.5m / s speed measurements; 8-14μm infrared thermal imaging captures vehicle engine hot spots (temperature difference >15°C) at a 50Hz frame rate; Spatiotemporal consistency processing: The PTP protocol achieves 70ns-level timestamp alignment, and the fiber optic gyroscope builds a unified coordinate system (coordinate conversion error < 0.1°); Cross-modal feature fusion: The radar stream passes through PointNet++ to extract the rectangular point cloud outline of the truck; the infrared stream passes through ResNet-50 to identify the high-temperature area on the car hood. The cross-attention mechanism assigns weights (radar-dominated distance judgment 0.7, infrared-assisted type recognition 0.3), and outputs target confidence: large trucks 93% / small cars 89%, with a false alarm rate of less than 3%.

[0071] Step S4: Dynamic parameter optimization closed loop Adaptive control strategy: Radar transmit power increased to 600mW (enhanced rain and fog penetration); photoelectric filter switched to the 900nm near-infrared band (suppressed mirror reflection of rain); visible light camera sampling frequency reduced to 10Hz (reduced rain noise data redundancy); Performance verification data: 12 motor vehicles were tracked at 8km beyond visual range, with a distance measurement mean square error of 42m. The traditional aperture system had a detection range of only 2.5km in the same scene, and three cars were missed. The target classification accuracy reached 91%, a 55% improvement over the fixed parameter mode, verifying the detection performance advantage in heavy rain environments.

[0072] In summary, in terms of adaptive environmental detection, the detection distance of the embodiment of the present application in heavy rain scenes is 3 times higher than that of traditional solutions, and the target recognition accuracy in strong light environments reaches 90%; in terms of multi-target precise tracking capability, continuous tracking of 12 motor vehicles is achieved with an update frequency of 10Hz, and the trajectory interruption time is controlled within 0.3 seconds; in terms of resource scheduling efficiency, the sensor power consumption is reduced by 25% compared with the fixed mode by dynamically adjusting the sampling frequency; the anti-interference robustness performance is outstanding, and the false alarm rate in complex scenarios is reduced by 80% compared with the traditional solution.

[0073] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include: Memory 701 , processor 702 , and computer programs stored in the memory 701 and executable on the processor 702 .

[0074] When the processor 702 executes the program, the composite radar photoelectric detection method provided in the above embodiment is implemented.

[0075] Furthermore, the electronic device further includes: The communication interface 703 is used for communication between the memory 701 and the processor 702 .

[0076] The memory 701 is used to store computer programs that can be run on the processor 702 .

[0077] The memory 701 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0078] If the memory 701, processor 702, and communication interface 703 are implemented independently, the communication interface 703, memory 701, and processor 702 can be connected to each other via a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, Figure 7 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0079] Optionally, in a specific implementation, if the memory 701, the processor 702 and the communication interface 703 are integrated on a chip, the memory 701, the processor 702 and the communication interface 703 can communicate with each other through an internal interface.

[0080] The processor 702 may be a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0081] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0082] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of this application, "N" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0083] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing a custom logical function or process step, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may be performed in a different order than shown or discussed, including performing functions in a substantially simultaneous manner or in a reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present application pertain.

[0084] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiment, the N steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, it can be implemented using any one or a combination of the following technologies known in the art: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0085] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

Claims

1. A composite radar photoelectric detection system, characterized in that: include: Radar detection module, photoelectric detection module, data processing module, control module and dynamic resource scheduling module, among which, The radar detection module and the photoelectric detection module realize physical layer fusion through a common aperture integration design; The data processing module realizes deep fusion of data layers based on multimodal algorithms; The dynamic resource scheduling module integrates an environment perception unit for collecting environmental data in real time and transmitting it to the control module; The control module has a built-in priority decision tree for dynamically allocating primary / secondary sensors according to environmental conditions.

2. The composite radar photoelectric detection system according to claim 1, characterized in that: The radar detection module includes: a transmitting antenna, a receiving antenna, a radio frequency front end and a signal processing unit, wherein: The transmitting antenna adopts a 79GHz±1GHz millimeter wave radar array, adopts a coaxial and common aperture layout with the photoelectric optical lens, and is embedded in the optical lens protective cover; The receiving antenna is used to receive the radar echo signal reflected by the target; The RF front end integrates an adaptive waveform generator to dynamically switch between LFM pulse compression signal and OFDM signal according to spectrum sensing results; The signal processing unit is used to obtain target distance, speed and direction information by calculating signal propagation time and frequency change, and supports frequency hopping anti-interference mode.

3. The composite radar photoelectric detection system according to claim 1, characterized in that: The photoelectric detection module includes: an optical lens, an infrared detector, a visible light camera and an image signal processing unit; wherein, The optical lens adopts a common aperture design and is coated with a special optical film layer on the surface to collect the optical signal of the target; The infrared detector has a built-in non-uniformity correction algorithm based on blackbody radiation reference, which is used to detect the infrared radiation signal of the target; The visible light camera is equipped with a 400-1100nm dynamically adjustable filter, which switches synchronously with the radar pulse to capture visible light images of the target; The image signal processing unit is used to process the signals output by the infrared detector and the visible light camera to obtain image information and characteristic parameters of the target.

4. The composite radar photoelectric detection system according to claim 3, characterized in that: The surface-coated special optical film layer is composed of alternately stacked high-refractive index material layers and low-refractive index material layers, wherein the high-refractive index material layer is made of titanium dioxide or tantalum pentoxide, and the low-refractive index material layer is made of silicon dioxide or magnesium fluoride.

5. The composite radar photoelectric detection system according to claim 1, characterized in that: The environmental sensing unit includes a temperature and humidity sensor, a light sensor, and a particle concentration sensor, which are respectively used to collect environmental temperature and humidity data, light intensity data, and atmospheric particle concentration data in real time.

6. The composite radar photoelectric detection system according to claim 1, characterized in that: The multimodal algorithm includes a multimodal spatiotemporal registration algorithm and a cross-modal deep learning model; wherein, The multimodal spatiotemporal registration algorithm uses the PTP precise clock protocol to achieve timestamp alignment between radar and optoelectronic data, constructs a unified coordinate system based on fiber optic gyroscopes, and uses the Hungarian algorithm to perform feature-level registration by extracting radar target RCS features and optical image HOG features. The cross-modal deep learning model uses a two-stream Transformer network. The radar stream inputs point cloud data of distance, velocity, and azimuth, and spatial features are extracted using PointNet++. The photocurrent inputs infrared and visible light images, and texture features are extracted using ResNet-50. The fusion layer realizes feature interaction through a cross-attention mechanism and outputs the target ID, type, and three-dimensional trajectory.

7. The composite radar photoelectric detection system according to claim 1, characterized in that: The decisions of the priority decision tree include: when visibility is less than 1 km and rainfall is greater than 5 mm / h, X-band + millimeter wave radar is used as the dominant sensor, and infrared thermal imaging is used as an auxiliary sensor; when at night and visibility is greater than 3 km, infrared + visible light is used as the dominant sensor, and laser ranging is used as an auxiliary sensor; when strong electromagnetic interference is detected, the optoelectronic anti-interference mode is enabled as the dominant mode, and radar frequency hopping anti-interference is used as the auxiliary mode.

8. The composite radar photoelectric detection system according to claim 1, characterized in that: The control module includes an environmental parameter receiving interface and a decision tree execution unit, which is used to adjust the parameters of radar transmission power and photoelectric filter wavelength according to the dynamic resource scheduling results, and output the fused data to the display device.

9. The composite radar photoelectric detection system according to claim 1, characterized in that: Also includes: A composite heat dissipation structure, wherein the composite heat dissipation structure is a microchannel liquid cooling plate set between the millimeter wave radar array and the optical lens, the coolant pipe is embedded in the radar antenna substrate and connected to the heat dissipation fins of the optoelectronic component to form a closed-loop heat dissipation circuit.

10. A composite radar photoelectric detection method, characterized in that: The following steps are involved: Obtain ambient temperature and humidity, light intensity, and particulate matter concentration data; generating an environmental feature vector according to the environmental temperature and humidity, the light intensity, and the particulate matter concentration data; Based on the priority decision tree logic, the environmental feature vector is compared with the preset threshold, and a sensor weight matrix is ​​output according to the comparison result. Radar point cloud and photoelectric image data are collected according to the sensor weight matrix. The radar point cloud and photoelectric image data are timestamp-aligned using a multimodal spatiotemporal registration algorithm. A unified coordinate system is constructed based on the fiber optic gyroscope. Feature fusion is performed through a dual-stream Transformer network to output the target confidence. Based on the target confidence and the environmental feature vector, the radar transmission power, the photoelectric filter wavelength and the sensor sampling frequency parameters are adjusted and optimized.

Images