Low-altitude intelligent defense system based on multi-modal fusion and AI decision

Abstract

The invention discloses an anti-unmanned aerial vehicle system which comprises a detection module, an identification module, an interception module and a cooperative control module. The detection module adopts a PLUS solid multi-panel radar and 3D holographic radar double architecture, is equipped with a gallium nitride T / R assembly, is combined with a dynamic beam forming technology, effectively inhibits clutters, eliminates terrain blind areas, realizes wide-area high-precision detection and multi-target tracking, has the weight of only several kilograms, and is convenient for maneuvering deployment. The recognition module fuses multi-source sensor data, analyzes target features through machine learning and an AI model, significantly improves the recognition rate and reduces false alarms. The interception module adopts a grading strategy, integrates a variable frequency band jammer, electromagnetic pulse and a laser weapon, can efficiently melt down or burn down the unmanned aerial vehicle group, and is extremely low in cost. The cooperation module realizes multi-node intelligent decision and closed-loop combat through distributed aperture and semi-autonomous AI, and supports software upgrading and equipment compatibility. The system integrates accurate detection, rapid identification, efficient interception and intelligent cooperation, and is suitable for key infrastructure protection.

Description

TECHNICAL FIELD

[0001] The application is a low-altitude intelligent defense system based on multi-modal fusion and AI decision, relating to the technical field of low-altitude defense, specifically relating to a low-altitude intelligent defense system integrating gallium nitride radar, multi-sensor identification, hierarchical electromagnetic / laser interception, and AI collaborative control, suitable for airport, city, field, and other multi-scene unmanned aerial vehicle and hypersonic target defense. BACKGROUND

[0002] The existing low-altitude defense system has four major problems: detection blind area, high false alarm rate, high interception cost, and inflexible deployment. Traditional radar cannot achieve high-precision detection in urban canyons and hilly terrain due to ground clutter interference; single sensor cannot distinguish unmanned aerial vehicles from birds, with a false alarm rate exceeding 30%; traditional missiles cost over $100,000 to intercept a single target, making it impossible to respond to multiple target clusters; most radars weigh over 20 kg and can only be fixedly deployed, and the data from each device cannot be interconnected, making it difficult to form a defense loop.

[0003] For example, the traditional airport unmanned aerial vehicle defense system relies solely on a single radar, with a false alarm rate exceeding 40%; in field environments, existing radars need to be fixedly deployed by vehicles and cannot adapt to roadless areas such as deserts and hills; electromagnetic jammers can only block a single frequency band and cannot respond to the frequency switching capability of software-defined unmanned aerial vehicles. Therefore, there is an urgent need for a low-altitude defense system that integrates multi-modal detection, AI accurate identification, low-cost hierarchical interception, and flexible collaborative deployment to solve the above technical problems. SUMMARY

[0004] Technical problems to be solved The present application aims to solve the technical problems of existing low-altitude defense systems, including: Radar detection blind area in complex terrain (urban canyons, hills, deserts); High false alarm rate for unmanned aerial vehicles and non-threatening targets such as birds; High interception cost when multiple target clusters attack; System rapid deployment in roadless and mobile scenarios; Each device data cannot be interconnected, making it difficult to form a closed-loop defense chain. Technical solutions

[0005] The present application provides a low-altitude intelligent defense system based on multi-modal fusion and AI decision, including a detection module, an identification module, an interception module, and a collaborative control module, each module interconnected through a universal interface to form a closed-loop defense link, with the following specific technical features: The detection module comprises a first radar unit and a second radar unit; the first radar unit is a PLUS system solid-state multi-panel radar, a transmit / receive (T / R) component of which adopts gallium nitride material, cooperates with a dynamic beam forming technology, is used for high-precision detection of a square-kilometer airspace, and suppresses ground clutter; the second radar unit is a 3D holographic radar, adopts gallium nitride technology to improve signal gain, and has a maximum detection distance of greater than or equal to 18 kilometers, supporting multi-target tracking of unmanned aerial vehicles; the detection module further integrates S-band 3D monitoring technology, and has a total weight of less than or equal to 5 kilograms, being suitable for rapid deployment modes such as vehicle, livestock carrying, and helicopter hoisting.

[0006] The identification module comprises a multi-sensor data acquisition unit and an AI processing unit; the multi-sensor data acquisition unit is used for acquiring radar data (from the detection module), radio frequency intelligence data, and photoelectric sensor data; the AI processing unit is loaded with a CX6 high-performance computing core, and is internally provided with a machine learning algorithm and an AI target identification model; the machine learning algorithm distinguishes unmanned aerial vehicles and non-threatening targets such as birds by analyzing target micro-Doppler characteristics; the AI target identification model is used for optimizing hypersonic target identification, and the identification rate is improved by greater than or equal to 60%; the false alarm rate of the identification module is reduced by greater than or equal to 70% compared with a traditional radar, and the identification module can stably work in a complex electromagnetic interference environment.

[0007] The interception module comprises a radio frequency interference unit, an electromagnetic pulse unit, and a laser unit; the radio frequency interference unit is a software-defined frequency band jammer, can switch frequency bands, and simultaneously blocks GPS navigation links and 2.4 / 5.8 GHz communication links of unmanned aerial vehicles; the electromagnetic pulse unit adopts a gallium nitride solid-state transmitter, can release an electromagnetic pulse in a 60-degree sector, melts target electronic elements through thermal effects, and can knock down greater than or equal to 49 unmanned aerial vehicles per single pulse, with a cost of less than or equal to 1 / 1000 of a traditional missile; the laser unit is a people-friendly laser weapon, can simultaneously blind or burn greater than or equal to 20 unmanned aerial vehicles within 1 kilometer, and is used for close-range defense; the interception module is internally provided with an AI hierarchical decision algorithm, first starts radio frequency interference, triggers electromagnetic pulses when invalid, and enables laser interception in close range.

[0008] The cooperative control module comprises a distributed interconnection unit and an AI decision unit; the distributed interconnection unit adopts a distributed aperture technology, realizes data interconnection of multiple nodes (greater than or equal to 3), expands the detection range, and reduces the risk of single-point interference; the AI decision unit is a semi-autonomous decision module, supports simultaneous control of greater than or equal to 8 targets by a single operator, and autonomously selects an interception strategy (radio frequency interference, electromagnetic pulse, laser interception, and collision); the cooperative control module adopts an open architecture, supports software iterative upgrading, is compatible with existing low-altitude defense systems, and forms a closed-loop combat link with the detection module, the identification module, and the interception module through a general interface.

[0009] Advantages Compared with the prior art, the application has the following advantages: Eliminate the detection blind area: the combination of dynamic beam forming technology and gallium nitride radar can suppress ground clutter, realize blind area detection in complex terrain such as urban canyons, hills and deserts, and improve the spatial coverage accuracy by more than 50%; Reduce false positives: multi-sensor fusion and micro-Doppler feature analysis, false positive rate reduced by ≥70% compared with traditional radar, hypersonic target recognition rate increased by ≥60%, and invalid interception avoided; Reduce interception cost: the interception cost of electromagnetic pulse is 1 / 1000 of that of traditional missiles, the price of laser weapons is affordable, and multiple targets can be intercepted simultaneously (up to 49 times at a time), with an interception efficiency increase of ≥300%; Flexible and rapid deployment: a few kilograms of radar can be adapted to vehicles, livestock, helicopters and other mobile methods, and single-point deployment can be completed within 10 minutes, suitable for roadless battlefield scenarios; Closed-loop cooperative defense: distributed interconnection + AI decision making, realizing data intercommunication of "detection-identification-interception", a single operator can cope with 8 targets, the defense response time is shortened to seconds, and the open architecture supports continuous upgrading, adapting to new threats. DETAILED DESCRIPTION

[0010] The technical solutions of the application will be described in detail in combination with two specific application scenarios: Scenario 1: Airport Low Altitude Defense

[0011] Deployment: Deploy 3 nodes of the application system within a 5 km range around the airport to form a "hand-in-hand" interconnected network through distributed aperture technology; the detection module (PLUS solid-state multi-panel radar + 3D holographic radar) of each node is installed on the airport tower and runway side, and the radar weighs 3.5 kg and is quickly fixed through a support; the interception module (RF jammer + electromagnetic pulse emitter) is deployed at the airport fence, and the laser unit is installed on the top of the terminal building.

[0012] Workflow: Detection stage: 3D holographic radar detects drones 18 km away, and PLUS solid-state multi-panel radar suppresses airport surrounding building clutter through dynamic beam forming to accurately capture low-altitude targets within several square kilometers; Identification stage: multi-sensor collects radar data, radio frequency signals (drone communication frequency band), and photoelectric images (drone shape), and AI processing unit analyzes micro-Doppler features (drone propeller speed 1000-3000 rpm, bird wing flapping frequency 5-10 times / s), false positive rate controlled within 5%; Interception stage: If the target is a rule-breaking drone, first start the radio frequency jammer to block its GPS and 2.4GHz communication link, forcing the drone to hover; if the drone continues to fly, trigger the electromagnetic pulse emitter (60-degree fan-shaped coverage over the airport runway) to melt its flight control chip; for drones within 1 km, use the laser unit to blind them to avoid the risk of falling debris; Cooperative control: Real-time data exchange between 3 nodes, AI decision unit automatically assesses threat level (e.g. "low risk - interference" "high risk - pulse"), response time ≤3 seconds, a single operator can monitor 3 nodes and 8 targets. Scenario 2: Low-altitude defense in the field

[0013] Deployment: In desert and hilly areas, deploy 4 nodes of the invention, each node is quickly deployed by soldiers carrying (radar weight 40 kg) or camels carrying (total weight of interception module 55 kg); nodes are interconnected through cofdm devices to form a warning network with a detection radius of 6-7 km.

[0014] Workflow: Detection stage: S-band 3D surveillance radar detects small drones 6-7 km away, PLUS solid-state multi-panel radar eliminates sand dune clutter interference, achieving high-precision tracking; Identification stage: AI processing unit combines radio intelligence (encrypted communication signals of drones) and optoelectronic data to distinguish between drones and desert birds, with an identification rate of ≥98%; Interception stage: For a swarm of drones (≤49), a single electromagnetic pulse release can knock them all down, with a cost of only $500 (traditional missiles cost $4.9 million); for 20 drones within 1 km, use the laser unit to burn them simultaneously, leaving no debris; Cooperative control: If a node is interfered, other nodes can compensate for the blind spot through distributed aperture technology, the AI decision unit automatically selects the "pulse interception + data backhaul" strategy, and a single operator can control 4 nodes through a portable terminal to deal with 8 targets.

Claims

1. A low-altitude intelligent defense system based on multimodal fusion and AI decision-making, characterized in that: It includes a detection module, an identification module, an interception module, and a collaborative control module. These modules are interconnected through a common interface to form a closed-loop defense link. The detection module includes a PLUS solid-state multi-panel radar and a 3D holographic radar using gallium nitride (GaN) material transmit / receive (T / R) components. The identification module includes a multi-sensor data acquisition unit and an AI processing unit equipped with a CX6 high-performance computing core. The interception module includes a software-defined radio frequency jammer, a GaN solid-state electromagnetic pulse transmitter, and a civilian-friendly laser weapon. The collaborative control module includes a distributed aperture interconnect unit and a semi-autonomous AI decision-making unit.

2. The system according to claim 1, characterized in that: The PLUS solid-state multi-panel radar, combined with dynamic beamforming technology, is used for high-precision detection over several square kilometers of airspace, suppressing ground clutter and eliminating detection blind spots in urban canyons and hilly terrain.

3. The system according to claim 1, characterized in that: The 3D holographic radar uses gallium nitride technology to improve signal gain, with a maximum detection range of ≥18 kilometers, and supports multi-target tracking by UAVs.

4. The system according to claim 1, characterized in that: The AI ​​processing unit incorporates machine learning algorithms to distinguish between drones and birds by analyzing the micro-Doppler features of the target, reducing the false alarm rate by ≥70% compared to traditional radar; at the same time, it embeds an AI target recognition model, which improves the recognition rate of hypersonic targets by ≥60%.

5. The system according to claim 1, characterized in that: The detection module integrates S-band 3D surveillance technology, weighs ≤5 kg, and is suitable for rapid deployment by vehicle, livestock carrying, or helicopter transport. It can detect small drones 6-7 km away in desert terrain.

6. The system according to claim 1, characterized in that: The software-defined radio frequency jammer can switch frequency bands and simultaneously block the UAV's GPS navigation link and 2.4 / 5.8GHz communication link; the gallium nitride solid-state electromagnetic pulse transmitter can release electromagnetic pulses in a 60-degree fan-shaped area, and a single pulse can shoot down ≥49 UAVs, with a cost ≤1 / 1000 of that of traditional missiles.

7. The system according to claim 1, characterized in that: The civilian-friendly laser weapon can simultaneously blind or burn ≥20 drones within 1 kilometer, and is used for close-range defense of critical infrastructure.

8. The system according to claim 1, characterized in that: The distributed aperture interconnection unit enables data interconnection of ≥3 nodes, expanding the detection range while reducing the risk of interference at a single point; the semi-autonomous AI decision-making unit supports a single operator to control ≥8 targets simultaneously, and autonomously selects interception strategies such as radio frequency interference, electromagnetic pulse, laser interception or impact.