Deep fusion architecture based on double-helix interleaving directional air-space interference equipment and gallium nitride technology

Abstract

The invention discloses a deep fusion architecture based on double-helix interleaving directional air-space interference equipment and a gallium nitride technology, and belongs to the technical field of low-altitude defense. The architecture comprises a detection module, an identification module, an interception module and a cooperative control module. The detection module adopts a gallium nitride T / R assembly and a carbon fiber lightweight antenna, and realizes high-maneuverability and high-precision detection in combination with an adaptive beam forming algorithm. And the identification module fuses the micro-Doppler features and the multi-modal data, and realizes high-accuracy target identification through an AI algorithm. The interception module integrates three means of radio frequency, electromagnetic pulse and laser, is driven by an AI grading decision algorithm, and realizes low-cost and efficient damage. And the cooperative control module realizes multi-node resource scheduling and cooperative combat through distributed networking and semi-autonomous AI decision. The problems that a traditional low-altitude defense system is poor in maneuverability, low in recognition rate, high in interception cost and weak in cooperative capacity are solved, and the method is suitable for regional protection of unmanned aerial vehicle clusters and low-altitude high-speed targets.

Description

Technical Field

[0001] This invention relates to a deep integration architecture based on double-helix intertwined directional aerospace jamming equipment and gallium nitride technology, and is involved in the fields of low-altitude target defense and electronic countermeasures technology. Specifically, it relates to a lightweight, distributed, collaborative defense system architecture based on gallium nitride technology and artificial intelligence that integrates detection, identification, interception and control. Background Technology

[0002] With the widespread use of drones, cruise missiles, and various low-altitude, slow-moving, small targets (such as small drones), traditional air defense systems face severe challenges. Existing technologies mainly suffer from the following shortcomings: Poor mobility: Traditional low-altitude detection radars are large and heavy (usually ≥25 kg), and take a long time to deploy, making them difficult to adapt to the tactical requirements of rapid mobility.

[0003] Low recognition accuracy: In complex electromagnetic environments and urban terrain, traditional radar has a large blind zone and is difficult to effectively distinguish between low, slow and small targets such as drones and birds, resulting in high false alarm and false miss rates.

[0004] High interception costs and low efficiency: Traditional interception methods (such as missiles) are expensive per attack and are difficult to deal with large-scale drone swarm attacks. Electronic jamming equipment often has limited functionality and lacks intelligent, hierarchical, and coordinated interception capabilities.

[0005] Weak coordination capabilities: Existing systems are mostly "siloed" independent architectures, with no information exchange between nodes, making it impossible to share detection resources and coordinate the scheduling of interception weapons, thus limiting overall defense effectiveness.

[0006] Therefore, there is an urgent need for a new type of defense system that integrates advanced materials, intelligent algorithms, and collaborative architecture to achieve efficient and low-cost interception of low-altitude, clustered, and intelligent threats. Summary of the Invention

[0007] 1. Technical issues This invention aims to solve the following technical problems existing in current aerospace defense systems: The detection module has low power density, large weight, complicated deployment, and weak anti-clutter capability; The identification module has a high false alarm rate, a low hypersonic target recognition rate, and poor resistance to electromagnetic interference. The interception module is costly, has low resource utilization, and lacks multi-target interception capabilities. The system has a closed architecture, poor coordination and scheduling, high labor costs, and a short life cycle. 2. Technical Solution

[0008] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a "deep integration architecture based on double-helix intertwined directional aerospace jamming equipment and gallium nitride technology". This architecture achieves full-process optimization of detection, identification, interception and control through material innovation, algorithm innovation and system architecture innovation, and significantly improves the defense capability against low-altitude and near-space targets.

[0009] To achieve the above objectives, the present invention adopts the following technical solution: A deep integration architecture based on double-helix intertwined directional aerospace jamming equipment and gallium nitride technology includes a detection module, an identification module, an interception module, and a collaborative control module.

[0010] The detection module employs gallium nitride (GaN) solid-state T / R components, whose power density (≥3W / mm²) is more than three times that of traditional silicon-based or gallium arsenide components. Combined with a microchannel liquid cooling design, it can operate stably for extended periods in a wide temperature range of -40℃ to 70℃. The antenna uses carbon fiber honeycomb composite material, reducing the overall radar weight to less than 5 kg, supporting various mobile deployment methods such as backpack, vehicle-mounted, and airborne deployments. Simultaneously, the module integrates an adaptive beamforming algorithm based on minimum mean square error (LMS), which can suppress multipath clutter generated by complex terrain (such as urban buildings) in real time (suppression ratio ≥40dB), achieving high-precision detection.

[0011] The identification module employs a multimodal fusion intelligent identification mechanism. First, it extracts the target's micro-Doppler features using Short Time Fourier Transform (STFT) and classifies the time-frequency map using an improved ResNet-18 convolutional neural network, effectively distinguishing the motion characteristics of UAV propellers from bird wings, achieving an accuracy rate ≥99.2%. Second, it uses a fusion model based on DS evidence theory to dynamically fuse data from radar, radio frequency detection, and photoelectric sensors, adaptively adjusting the weights of each sensor according to environmental factors (such as weather and electromagnetic interference intensity) to ensure reliable identification even when a single sensor fails. For hypersonic targets, a Transformer encoder is used to process high-speed time-series data, significantly improving the identification rate compared to traditional methods.

[0012] The interception module employs a tiered interception strategy combining "soft kill" and "hard kill," consisting of a radio frequency jammer, an electromagnetic pulse (EMP) transmitter, and a laser weapon. At its core is a built-in AI-based tiered decision-making algorithm. This algorithm takes the target threat level, distance, speed, and interception resource status as input, and automatically selects the optimal interception method (such as long-range radio frequency jamming, mid-range EMP coverage, and close-range laser precision strike) through a multi-factor weighted model and decision tree logic. The EMP transmitter uses an L-band gallium nitride solid-state source and a Marx generator, capable of producing a wide-area fan-shaped high-energy pulse, covering an area of ​​≥1 square kilometer in a single strike, effectively countering drone swarms. The laser weapon uses a high-beam-quality solid-state laser, achieving rapid switching between multiple targets through a galvanometer.

[0013] The collaborative control module is crucial for enabling distributed operations within the system. It employs COFDM modulation technology and a distributed aperture interconnect protocol to network multiple physically dispersed nodes. Through high-precision time synchronization (≤1μs) and Kalman filtering data fusion, a unified global situational awareness map is generated. The semi-autonomous AI decision-making unit can dynamically optimize resource allocation and interception commands based on the global threat situation and the resource status of each node (such as remaining EMP attempts and laser energy) using reinforcement learning (Q-Learning) algorithms, achieving cross-node collaborative detection and fire support. The module adopts an open architecture (such as the ROS framework and STD-1553B bus), supporting continuous iterative upgrades of software and hardware functions. Beneficial effects

[0014] Compared with the prior art, the present invention has the following significant advantages: High mobility and high detection performance: The lightweight design based on gallium nitride and carbon fiber materials (≤5 kg) enables the system to have excellent tactical mobility while maintaining high power and excellent anti-clutter capability.

[0015] Extremely high recognition accuracy and robustness: Combining micro-Doppler and multimodal fusion AI algorithms, the false alarm rate of targets is reduced to ≤5%, and stable recognition can be maintained in complex electromagnetic environments and harsh weather conditions.

[0016] Low-cost, high-efficiency interception: Through an AI-driven hierarchical interception strategy, the combination of low-cost radio frequency interference, medium-cost electromagnetic pulse and high-precision laser weapons has greatly optimized the cost-effectiveness ratio. The cost of cluster interception is only one-thousandth of that of traditional missiles.

[0017] Distributed collaboration and intelligent decision-making: Through distributed networking and semi-autonomous AI decision-making, "1+N" joint prevention and control is achieved, which expands the system's detection range several times, increases the interception success rate to over 95%, and significantly reduces the manpower requirements of operators. Detailed Implementation

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0019] One specific embodiment of the present invention is a portable low-altitude cluster target defense system. Each node of the system weighs 4.8 kg and can be carried by a single soldier or mounted on a light vehicle.

[0020] Detection Module: Its radar antenna array consists of 256 gallium nitride T / R units, mounted within a flat panel antenna made of carbon fiber composite material. The signal processing unit utilizes an FPGA+GPU heterogeneous computing chip, with a volume of only 0.5L. After power-on, the module completes deployment and begins scanning within 10 minutes. When a group of suspicious targets is detected at a distance of 3 kilometers, the LMS algorithm dynamically adjusts the beam pattern based on the real-time environmental clutter spectrum, creating nulls in the direction of strong clutter to ensure clear target visibility.

[0021] Identification Module: Radar echo data is fed into the identification module. The STFT processing unit generates a time-frequency distribution map of the target, and the AI ​​chip (20 TOPS computing power) runs a trained neural network model, initially identifying it as a "multi-rotor drone swarm, with a confidence level of 98%". Simultaneously, the radio frequency detection subunit captures a 2.4GHz remote control signal, and the photoelectric turret also captures visible light images when visibility is good. The DS fusion model integrates these three types of information and ultimately outputs the identification conclusion: "Consumer-grade multi-rotor drones, 12 units", and assesses the threat level (T) as 5.

[0022] Interception Module: The AI ​​hierarchical decision-making algorithm receives target information (T=5, distance 2.5km, cluster size 12). According to the decision tree rules, the radio frequency jamming unit is activated first. This unit, based on software-defined radio (SDR), rapidly hops frequencies within the 0.5-6GHz band, successfully suppressing the target's GPS and remote control links. 0.5 seconds later, the detection module reports that two of the drones are unaffected and continue moving forward. The algorithm immediately triggers the electromagnetic pulse (EMP) transmitter. The Marx capacitor array charges to 50kV within milliseconds and emits a high-energy pulse with a 60-degree fan-shaped arc towards the target, burning out the flight control systems of these two drones. The remaining 10 drones hover or make forced landings due to communication interruption.

[0023] Cooperative Control Module: Throughout the process, this node maintains real-time communication with another node 2 kilometers away via a wireless data link (data rate 2Mbps). The cooperative control module fuses the detection data from the two nodes to generate a situational awareness map with a wider coverage area. When the EMP of this node becomes insufficient due to continuous use, the cooperative algorithm can direct the laser weapon of the other node to supplement the interception of individual high-speed targets that are penetrating the defenses.

[0024] This invention is not limited to the above-described embodiments. Those skilled in the art, inspired by this invention, can make equivalent substitutions or improvements to the specific implementation of the module (such as neural network model structure, laser type, and network protocol variant), and all of these fall within the protection scope of this invention.

Claims

1. A deep integration architecture based on double-helix interlaced directional aerospace jamming equipment and gallium nitride technology, characterized in that, include: Detection module, identification module, interception module, and collaborative control module; The detection module employs a solid-state T / R component based on gallium nitride (GaN) technology and a carbon fiber cellular composite antenna, and integrates an adaptive beamforming algorithm based on minimum mean square error (LMS) to achieve high-power, lightweight low-altitude target detection. The identification module employs a multimodal fusion algorithm based on micro-Doppler feature analysis and DS evidence theory, combined with an improved deep neural network model, for the accurate identification and classification of low-altitude, slow-speed, and hypersonic targets. The interception module includes a radio frequency jamming unit, an electromagnetic pulse (EMP) emitting unit, and a laser weapon unit that can work together, and has a built-in AI hierarchical decision-making algorithm based on multi-factor weighting and reinforcement learning, which is used to implement low-cost and high-efficiency hierarchical interception according to the target threat level and resource status. The collaborative control module adopts node networking technology based on COFDM modulation and distributed aperture interconnection protocol, and runs a semi-autonomous AI decision-making unit to realize data fusion, resource scheduling and collaborative operation among multiple nodes.

2. The deep fusion architecture according to claim 1, characterized in that, In the detection module, the power density of the gallium nitride solid-state T / R component is ≥3W / mm², and a microchannel liquid cooling heat dissipation design is adopted; the antenna is made of carbon fiber honeycomb composite material with a density of 1.2g / cm³; the adaptive beamforming algorithm can dynamically form a "null" pointing in the direction of clutter, with a suppression ratio ≥40dB.

3. The deep fusion architecture according to claim 1, characterized in that, In the identification module, the micro-Doppler feature analysis uses short-time Fourier transform (STFT) to extract the target time-frequency map and classifies it through a convolutional neural network; the multimodal fusion algorithm dynamically allocates the weights of radar, radio frequency and photoelectric sensor data.

4. The deep fusion architecture according to claim 1, characterized in that, In the interception module, the AI ​​hierarchical decision-making algorithm takes target type, speed, distance, cluster size, and current interception resource status as inputs, and optimizes through a decision tree model and Q-Learning reinforcement learning to generate and dynamically adjust the interception strategy; the electromagnetic pulse transmitter operates in the L-band, uses a Marx generator, and produces an energy density of up to 10. 6 The laser weapon uses a 1.06μm wavelength solid-state laser with an output power of 10kW and is equipped with a galvanometer scanning system to achieve multi-target time-division irradiation.

5. The deep fusion architecture according to claim 1, characterized in that, The collaborative control module supports cascading networking of ≥8 nodes. Data synchronization between nodes is achieved through a GPS second pulse-based timestamp calibration algorithm with a synchronization accuracy of ≤1μs. It adopts an open architecture, with hardware interfaces conforming to the STD-1553B bus protocol and software framework using the ROS system.