








摘" 要: 針對傳統軍事要地安防系統智能化程度較低,各自獨立互不關聯,缺少頂層數據綜合治理等問題,選取外圍周界、重要卡口、無人巡更和區域高點四種典型安防業務場景開展建模,通過數據標準化接入、智能研判分析、安防態勢顯示三個處理環節,構建基于AI的多模態融合感知綜合決策系統,實現前端感知多維化、中臺研判智能化、后端處置多樣化,有效支撐了重要軍事目標安全防衛,系統后續也具有良好的可擴展性與可維護性。
關鍵詞: 要地防衛; 多模態數據融合; 場景建模; 規則定制; 感知處置決策; 智能化安防系統
中圖分類號: TN919?34; TP302.1" " " " " " " " " " " " 文獻標識碼: A" " " " " " " "文章編號: 1004?373X(2025)01?0173?06
Design and implementation of multimodal fusion perceived synthetic decision making system based on AI
FENG Xiaohui1, AI Run2, LIU Linqing1, XU Zhen1
(1. Xi’an Longview Electronic Engineering Co., Ltd., Xi’an 710100, China;
2. Unit 32605 of PLA, Chongqing 400060, China)
Abstract: The traditional military security systems have low level of intelligence, and they are independent and unrelated, and lack comprehensive management of top?level data, so four typical security business scenarios, namely peripheral perimeter, important checkpoints, unmanned patrols, and regional high points, are selected for modeling. By the three processing steps of standardized access of data, intelligent judgment analysis and security situation display, an AI?based multimodal fusion perceived synthetic decision making system is constructed to realize the multi?dimensional front?end perception, intelligent judgment in the middle platform and diversified back?end disposal. The constructed system can support the security defense of important military targets effectively. In addition, the system has good scalability and maintainability in the future.
Keywords: important area defense; multimodal data fusion; scene modeling; rule customization; perceived disposal decision; intelligentized security system
0" 引" 言
近年來國際環境不斷變化,重要軍事目標面臨的威脅日益加劇,另一方面人工智能、大數據、云邊計算等技術飛速發展[1],但傳統安防產品從設備層面看,各單體大多獨立工作、互不關聯,從數據層面看,業務信息僅在本系統內部流轉,缺乏面向復雜應用場景的頂層設計與數據綜合治理[2]。因此,設計新型要地防衛系統,通過有效集成與整合各類感知預警信息,形成針對性處置決策,實現入侵及時發現、隱患及時處置、風險及時防范變得十分迫切。
本文所述系統聚焦安防業務場景建模,在平臺層面通過異構(圖像、視頻、結構化信息等)異源(可見光數據、微波數據、定位數據等)多模態數據共享、數據融合與信息互補[3?4]實現“感知?研判?處置”三個環節的智能聯動,構建了威脅可預警、行為可檢測、攻擊可防御、事件可處置的整體安全防御體系。……