Multi-source information fusion detection method for low detectable target

Abstract

The invention belongs to the technical field of detection, and particularly relates to a multi-source information fusion detection method for a low detectable target. Comprising the following steps: S1, establishing a platform construction mechanism and a performance measurement mechanism; s2, a task effect-oriented architecture optimization and reconstruction strategy; and S3: perception and understanding of target characteristic data. The method has the beneficial effects that according to the performance and characteristics of a detection platform, a plurality of multi-type detection nodes are optimally arranged, multi-platform multi-azimuth cluster collaborative detection is performed on a low detectable target, target multi-source feature information is extracted for deep fusion, a cluster collaborative search path is reasonably planned, and the detection efficiency is improved. According to the method, the coverage area for searching the low detectable target is expanded, and the technologies of identifying and tracking the target, alternately detecting time-frequency characteristics, checking credibility and the like under the condition of extremely weak signal-to-noise ratio are utilized, so that the searching efficiency is improved, the discovery of the detected target is enhanced, and the distributed integrated detecting and tracking capability for the low detectable target is formed.

Description

Technical Field

[0001] The present invention belongs to the field of detection technology, and in particular relates to a multi-source information fusion detection method for low-detectable targets. Background Art

[0002] Low detectable technology is a technology that makes the target body difficult to be discovered by certain remote sensing devices, or difficult to track and aim at after being discovered. Platforms that use low detectable technology can not only greatly compress the target discovery distance of the detection system, but also effectively reduce the detection platform's effective processing time for the target, making it difficult for search and detection platforms to discover, identify, track and deal with the target.

[0003] For low-detectable targets, especially low, slow and small targets in ocean, foggy and rainy conditions, the existing detection technology cannot achieve reliable detection and tracking of targets, and at the same time reduces the effective distance of the detection platform to detect and track targets, which has caused great safety hazards. In recent years, with the development of high-tech and the technological progress of various types of sensors, countries around the world are actively exploring new detection technologies, such as over-the-horizon radar, carrier-free radar, dual (multi) base radar, dual-band or multi-band radar, harmonic radar, laser radar, new synthetic aperture radar and other radar detection technologies, as well as other detection methods such as optical detection, acoustic detection, magnetic field detection and electric field detection. However, the single detection means and detection platform currently used are limited by device characteristics and capabilities, and cannot fully meet the detection needs of low-detectable targets. It is also easy to cause misjudgment, false alarm and missed alarm. Summary of the invention

[0004] The purpose of the present invention is to provide a multi-source information fusion detection method for low-detectable targets, deeply fuse the multi-source data obtained by the detector, and optimize the collaboration strategy between multiple platforms, so as to achieve rapid and high-precision capture and tracking of low-detectable targets.

[0005] The technical solution of the present invention is as follows: A multi-source information fusion detection method for low-detectable targets comprises the following steps;

[0006] S1: Establish platform construction mechanism and performance measurement mechanism;

[0007] S2: Architecture optimization and reconstruction strategy for task effectiveness;

[0008] S3: Perception and understanding of target characteristic data.

[0009] The S1 includes the following:

[0010] S11: Establish platform structure theory;

[0011] S12: Establish performance measurement mechanism.

[0012] The S11 includes that the platform structure theory includes the component configuration principle, the components include components and connectors, all of which are described by algebraic complexes, the components are characterized as covering elements of the topological space, described as cell complexes, and the connectors are characterized as function spaces between complex mappings, described as morphisms; various types of algebraic complexes have geometric matching characteristics in the topological space during the assembly process, and through the homology and homology principles of topology, the algebraic relationship between the homology group, homology group and their derivatives of the complex in the assembly process is established.

[0013] The S12 includes that the platform behavior is manifested as a homeomorphic transformation in the topological space, and the behavior can be described as a transformation group, which triggers the transition from one state to another on its differential manifold. According to the relationship between the transformation group and its transition, the behavior utility calculation formula is used to establish the transmission equation of the utility on the behavior path and establish the performance measurement principle.

[0014] The S2 includes the following:

[0015] Based on the algebraic topology theory, conventional algebraic operators are expanded to define the "inspiration", "use", "feedback", "collaboration", "parallelism" and "mutual exclusion" operations between components. Component combination is achieved through component connection operations to represent the interactive cooperation relationship between different components. Functional modules are divided, necessary basic components and algebraic operators are extracted, and on the basis of the task metamodel definition, a library of commonly used basic components and component "operation" relationship models is defined, and the corresponding ontology library is directly mapped and generated.

[0016] The S2 includes the perception and understanding of the following target characteristic data, including intelligent guidance, management and access control of different security levels, data integration and information fusion, sharing and distribution, etc., involving technical mechanisms and concepts such as understanding, association, insight, and prediction. The implementation methods include:

[0017] Understand, extract the geometric and physical features of the target, and model the subject, scene, behavior, and emotion of the target;

[0018] Association, linking related objects based on the relationships between geometric and physical features;

[0019] Insight, revealing the complete characteristic portrait, cause and effect, relationship, and characteristic laws of the target or event.

[0020] The beneficial effects of the present invention are as follows: according to the performance and characteristics of the detection platform, multiple multi-type detection nodes are optimized and arranged respectively, multi-platform and multi-directional cluster collaborative detection of low-detectable targets is carried out, multi-source feature information of the target is extracted for deep fusion, and cluster collaborative search paths are reasonably planned, so as to expand the coverage area of ​​the search for low-detectable targets, utilize technologies such as target identification and tracking under extremely weak signal-to-noise ratio conditions, alternating detection of time-frequency features and credibility verification, improve search efficiency, strengthen the discovery of detection targets, and form a distributed integrated detection and tracking capability for low-detectable targets. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 A multi-information fusion detection system architecture for low-detectable targets;

[0022] Figure 2 The construction mechanism of the digital support platform for target characteristics;

[0023] Figure 3 for description of structure, behavior, and energy;

[0024] Figure 4 Component computing optimized for architecture;

[0025] Figure 5 To establish a flexible restructuring strategy. DETAILED DESCRIPTION

[0026] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] In view of the fact that the current single detection means and detection platform are limited by device characteristics and capabilities, they cannot fully meet the detection needs of low-detectable targets and are prone to misjudgment, false alarm and missed alarm. It is necessary to adopt the information fusion technology of acoustic, optical and electrical multi-source composite detection system to give full play to the advantages of different platform information sources in target color characteristics, thermal radiation intensity, acoustic characteristics, specific band radar monitoring, distance measurement, significant differences relative to the background and noise characteristics, so as to ensure the requirements of composite detection system for low-detectable target discovery, tracking and disposal.

[0028] The core idea is to optimize the layout of multiple multi-type detection nodes in the main detection direction according to the performance and characteristics of the detection platform, conduct multi-platform and multi-directional cluster collaborative detection of low-detectable targets, extract multi-source feature information of the target for deep fusion, and reasonably plan cluster collaborative search paths, expand the coverage area of ​​the search for low-detectable targets, use technologies such as target identification and tracking under extremely weak signal-to-noise ratio conditions, alternating detection of time-frequency features and credibility verification, to improve search efficiency, strengthen the discovery of detection targets, and form a distributed integrated detection and tracking capability for low-detectable targets.

[0029] The present invention provides a comprehensive detection method for multi-platform collaborative detection of low-detectable targets using deep fusion of multi-source information, and particularly involves arranging detectors on multiple platforms on land, sea, air and space, deeply fusing multi-source data obtained by the detectors, and optimizing the collaboration strategy between multiple platforms, in order to achieve rapid and high-precision capture and tracking of low-detectable targets.

[0030] The multi-source information fusion detection of low-detectable targets is an integrated distributed optimized target discovery and tracking system, which can realize the unified, safe, reliable, real-time, effective and flexible combination of system platform architecture, and provide target characteristic data perception and shared understanding services for task-driven low-detectable target collaborative perception, so as to realize target discovery and tracking. The overall system architecture established by the multi-source information fusion detection method for low-detectable targets provided by the present invention is as follows: Figure 1 shown.

[0031] A multi-source information fusion detection method for low-detectable targets specifically comprises the following steps:

[0032] S1: Establish platform construction mechanism and performance measurement mechanism

[0033] The integrated target characteristic digital support platform involved is abstracted as an algebraic engineering system. Various systems, components, and modules are described as algebraic devices. The platform is assembled from various algebraic devices, and its construction problem is the assembly problem of algebraic devices under a certain assembly strategy. If the algebraic device is described as an algebraic topological complex, then the platform's "component configuration, configuration coordination, energy transmission" and other models can be established in the topological space, thereby establishing the platform's construction mechanism and performance measurement, such as Figure 2 shown.

[0034] Therefore, assembly is transformed into a complex connection combination under homomorphic mapping and a complex assembly problem under topological geometry. Combination occurs between components at the same level, and the hierarchical system is combined into a non-closed category through the combination operation of algebraic complexes. Assembly occurs between systems at different levels, and the non-closed category of the hierarchical system is assembled into a larger non-closed category through morphism functors, which is essentially an algebraic combination operation of complexes. Specifically, it includes the following:

[0035] S11: Establishing platform structure theory

[0036] The platform structure theory includes the principle of component configuration. Components include members and connectors, which can all be described by algebraic complexes.

[0037] The components are characterized as covering elements of topological space, described as cell cavity (CW) complexes, and the connectors are characterized as function spaces between complex mappings, described as morphisms. Both CW complexes and Hom complexes are algebraic complexes with algebraic description forms.

[0038] Configuration matching principle: Various algebraic complexes have geometric matching characteristics in topological space during the assembly process. Through the homotopy and homology principles of topology, the algebraic relationship between the homotopy group, homology group and their derivatives of the complexes in the assembly process can be established, thereby establishing the configuration matching principle.

[0039] S11: Performance measurement mechanism

[0040] The platform behavior is manifested as a homeomorphic transformation in the topological space. The behavior can be described as a transformation group, which triggers the transition from one state (cover) to another state (cover) on its differential manifold. According to the relationship between the transformation group and its transition, a behavior utility calculation formula can be proposed, and the transmission equation of utility on the behavior path can be established, thus forming a performance measurement principle.

[0041] In a specific scenario, the effect of behavior produces performance utility. The so-called utility is the expansion of behavior to a specific goal in a specific scenario for a specific object. According to the principle of differential geometry, the scenario is described as a differential manifold, and the behavior is described as an expansion process from one local coverage of the differential manifold to another local coverage. The change in "capability" accompanying the expansion can be defined as "performance utility". Performance utility is closely related to the behavior tensor gradient and the scale of the differential manifold.

[0042] Assume that the scenario is an n-dimensional differential manifold M, whose scale S is defined as the n-order tensor mixed product of M (a scalar), and the differential form of the manifold is ω, then the performance utility E can be defined as:

[0043] E=s∫ M ω,

[0044] Suppose the behavior path from the object to the target is According to Stokes' theorem, we have:

[0045]

[0046] Set Path is a directed chain consisting of m nodes, ρ α is the decomposition unit of the αth node. Since each node on the path is directed connected, the nodes constitute m equivalence classes but:

[0047]

[0048] The differential form ω is a global quantity, and its integral on M is a scalar, so E is a global scalar. It can be proved that in the probability space, the change trend described by the above formula is consistent with the Shannon information entropy formula, and the behavior utility is the semantic information entropy of the behavior. According to the calculation of performance utility, the utility transmission equation can be established on different paths, thus forming the behavior transmission theory. Therefore, based on the theoretical basis of assembly, the platform construction technology can be constructed.

[0049] S2: Architecture optimization and reconstruction strategy for task performance

[0050] System architecture optimization is based on component computing and is carried out towards task effects, such as Figure 4 shown.

[0051] Based on the algebraic topology theory, conventional algebraic operators are expanded to define operations such as "inspiration", "use", "feedback", "collaboration", "parallelism" and "mutual exclusion" between components. Component combination is achieved through component connection operations to represent the interactive cooperation relationship between different components; functional modules are divided to extract necessary basic components and algebraic operators. On the basis of the definition of the task metamodel, a library of commonly used basic components and component "operation" relationship models is defined, and the corresponding ontology library is directly mapped and generated; referring to the service-oriented architecture concept, the required components are modularly assembled based on the principles of reusability and low coupling, interface technology is studied, good interfaces and contracts are defined to connect different functional units, heterogeneous computing and distributed computing technologies are studied so that intelligent manufacturing can be jointly provided by a group of independent computers, and a modular assembly method is established; the generation method of the algebraic model of the system architecture is explored, and the architecture is defined as

[0052] S= <M1,M2…M n >, M i =<C,O> or M i It is the bottom layer M j In the definition process of this algebraic model, the closedness of software components, the hierarchy of component combination and the extensibility of software are guaranteed.

[0053] How to establish a flexible restructuring strategy Figure 5 shown.

[0054] S3: Collaborative perception and understanding of target information based on effects The multi-source information fusion detection system for low-detectable targets based on information systems integrates the capabilities of multi-domain detection platforms such as land, sea, air, space, network, and electricity, unifies planning, control, and coordinates actions in various fields, and is used to achieve the discovery and tracking of low-detectable targets in the entire space. The system can accommodate multi-domain, heterogeneous, and massive target characteristic data and data from various types of sensors, and use artificial intelligence and machine learning for fusion to achieve machine speed analysis and real-time situational awareness. The perception and understanding of target characteristic data includes intelligent guidance, management and access control of different security levels, data integration and information fusion, shared distribution, etc., involving technical mechanisms and concepts such as understanding, association, insight, and prediction. The main implementation methods include:

[0055] Understand. Extract the geometric and physical features of the target, and model the subject, scene, behavior, and emotion of the target. The core technologies involved are: feature modeling and annotation, behavior analysis, deep learning, probability model, knowledge graph, behavior modeling, information propagation theory, etc.

[0056] Association. Link related targets based on the relationship between geometric and physical features. The core technologies involved are: similarity search, correlation calculation, graph (network) model, multivariate network, link prediction, probabilistic modeling, learning methods, etc.

[0057] Insight. Reveal the complete feature portrait, cause and effect, relationship, and characteristic rules of the target or event. The core technologies involved are: target portrait, event trajectory, causal reasoning, transfer learning, complex network analysis, manifold learning, visualization analysis, etc. Prediction. Predict the development trend of events and task actions. The core technologies involved are: regression analysis, model deduction, Bayesian prediction, abnormal pattern detection, etc.

[0058] The relevant technologies constitute a unified, real-time, secure, reliable and effective system that can be flexibly combined, which can collaboratively integrate multi-domain and multi-source target characteristic big data, achieve rapid understanding of low-detectable targets, and support mission decision-making.

[0059] The overall idea of ​​the multi-source information fusion detection system for low-detectable targets is: to ensure the real-time and high-precision flexible reorganization, a platform construction theory and performance measurement principle are proposed; a formal description and combination operation method of the system, components and modules are established; a precise calculation principle of the platform construction process and performance optimization is proposed, and according to the calculation of the task effect, an overall optimization method and structural reconstruction strategy of the platform architecture are established; based on the collaborative perception and understanding of the target information of the effect, a unified, real-time, safe, reliable and effective system with flexible combination is constructed, which collaboratively integrates the multi-domain and multi-source target characteristic big data to achieve rapid understanding and disposal of low-detectable targets.

Claims

1. A multi-source information fusion detection method for low detectable targets, characterized in that: The steps include: S1: Establish platform construction mechanism and performance measurement mechanism; S2: Architecture optimization and reconstruction strategy for task effectiveness; S3: Perception and understanding of target characteristic data.

2. A multi-source information fusion detection method for low detectable targets as claimed in claim 1, characterized in that: The S1 includes the following: S11: Establish platform structure theory; S12: Establish performance measurement mechanism.

3. A multi-source information fusion detection method for low detectable targets as claimed in claim 2, characterized in that: The S11 includes that the platform structure theory includes the component configuration principle, the components include components and connectors, all of which are described by algebraic complexes, the components are characterized as covering elements of the topological space, described as cell complexes, and the connectors are characterized as function spaces between complex mappings, described as morphisms; various types of algebraic complexes have geometric matching characteristics in the topological space during the assembly process, and through the homology and homology principles of topology, the algebraic relationship between the homology group, homology group and their derivatives of the complex in the assembly process is established.

4. The multi-source information fusion detection method for low detectable targets according to claim 2, characterized in that: The S12 includes that the platform behavior is manifested as a homeomorphic transformation in the topological space, and the behavior can be described as a transformation group, which triggers the transition from one state to another on its differential manifold. According to the relationship between the transformation group and its transition, the behavior utility calculation formula is used to establish the transmission equation of the utility on the behavior path and establish the performance measurement principle.

5. The multi-source information fusion detection method for low detectable targets according to claim 1, characterized in that: The S2 includes the following: Based on the algebraic topology theory, the conventional algebraic operators are extended to define the "inspiration", "use", "feedback", "cooperation", "parallelism" and "mutual exclusion" operations between components. The component combination is realized through the component connection operation to express the interactive cooperation relationship between different components. Divide the functional modules, extract the necessary basic components and algebraic operators, define the commonly used basic components and component "operation" relationship model library based on the task metamodel definition, and directly map and generate the corresponding ontology library.

6. The multi-source information fusion detection method for low detectable targets according to claim 1, characterized in that: The S2 includes the perception and understanding of the following target characteristic data, including intelligent guidance, management and access control of different security levels, data integration and information fusion, sharing and distribution, etc., involving technical mechanisms and concepts such as understanding, association, insight, and prediction. The implementation methods include: Understand, extract the geometric and physical features of the target, and model the subject, scene, behavior, and emotion of the target; Association, linking related objects based on the relationships between geometric and physical features; Insight, revealing the complete characteristic portrait, cause and effect, relationship, and characteristic laws of the target or event.

Images