Spaceflight multi-source heterogeneous information intelligent fusion method

Abstract

The application discloses a kind of aerospace multi-source heterogeneous information intelligent fusion method, comprising the following steps: step 1: if L (t1) =L (t2), target shutdown period does not move in situ;Step 2: if L (t1) ≠L (t2), determine whether there is remote sensing satellite during signal loss to the target;Step 3: when not shooting target remote sensing image, the starting point L (t1) of missing path, the end point is L (t2), the missing duration gives m recommended paths p1, p2, p3,..., p m ;Step 4: when shooting target remote sensing image, path is divided into 2 sections, first section, second section missing path is respectively executed step 3, and the missing path is estimated.It has the advantages that: fusion aerospace electronic signal sensing data, aerospace optical remote sensing data and map data target, time, space information, solve the existing data time, spatial rate low problem, can provide accurate input for aerospace multi-source information correlation analysis, also can grasp event importance, event scale, event urgency and other key sensitive information in time.

Description

Technical Field

[0001] This invention belongs to information science, specifically relating to a data mining and path planning method. Background Technology

[0002] Aerospace electronic sensing and aerospace optical remote sensing belong to different disciplines and have developed in siloed fashion over the years, resulting in their own technological development and application systems. Their data generation, processing, output data attributes, standards, and service targets are all different, making direct integration impossible. The main method for obtaining the position of non-cooperative targets is using aerospace electronic sensing satellites to calculate the target's position based on its electromagnetic signals. However, once the target implements electromagnetic silence, its position becomes completely unknown. Therefore, it is urgent to study the position and trajectory of targets during electromagnetic silence.

[0003] Remote sensing fusion technology can be divided into three different levels: pixel / data level, feature level, and decision level. One existing deep learning-based multi-source heterogeneous data fusion method, based on the basic architecture of data fusion, conceives a highly generalizable deep learning model for multi-source heterogeneous image data fusion. It performs pixel-level fusion, feature-level fusion, and decision-level fusion of optical, SAR, and infrared images respectively, applying deep learning technology to the extraction and mining of information from multi-source heterogeneous data, thus achieving the fusion of multi-source heterogeneous images.

[0004] Existing pixel-level fusion methods mainly focus on optical data fusion. Advanced fusion includes feature-level and decision-level fusion of multi-source data, such as synthetic aperture radar, optical images, LiDAR, and other types of data.

[0005] For multi-user decision-making problems based on multi-source data fusion, some existing technologies have established multi-source heterogeneous data fusion models. This paper studies a unified quantization representation method for heterogeneous data based on triangular fuzzy numbers, and incorporates decision-makers' preferences using an ordered weighted average operator. A multi-source heterogeneous data fusion algorithm supporting multi-user decision-making is designed. This addresses the fuzziness, differences, and heterogeneity of multi-source heterogeneous data in terms of structure and semantics. By considering decision-makers' preferences during the data fusion process, the reliability of multi-user decision results is improved. The algorithm achieves the fusion processing of semantic data with different descriptive methods.

[0006] To address the processing challenges of multi-source heterogeneous sensing information in the Internet of Things (IoT), some existing technologies propose multi-level multi-source heterogeneous data fusion methods. Taking target localization and tracking applications based on wireless signals, video, and depth sensing data as a starting point, these technologies focus on the processing, feature representation, and data fusion methods for multi-source heterogeneous data. Different data fusion methods are employed based on the characteristics of different data types. By mining the inherent correlations among multi-source heterogeneous data such as wireless signals, video, and depth sensing data, valuable information from the multi-source heterogeneous data can be effectively utilized.

[0007] In the prior art, there is a data model based on the principle of Exchangeable Image Files (EXIF) that uses digital images as a carrier to fuse spatial location information and general formal attributes. This model embeds the associated spatial location and general attributes into the physical structure of the digital image, achieving a high degree of integration of associated data, thereby realizing the fusion between spatial location data and digital image data.

[0008] Whether it is the fusion of targets or the fusion of location information, existing work fuses the spatiotemporal information obtained by electronic signal sensing with the spatiotemporal information extracted by optical remote sensing images to obtain more accurate information on the basis of complete data. However, there is a lack of research on fusing electronic signal sensing and optical remote sensing information to obtain missing information. At the same time, there is a lack of research on spatiotemporal information fusion methods for targets and the events they participate in. Summary of the Invention

[0009] The purpose of this invention is to provide a method for intelligent fusion of multi-source heterogeneous information in aerospace, which can acquire more complete time, location, target and event information.

[0010] The technical solution of this invention is as follows: A method for intelligent fusion of multi-source heterogeneous information in aerospace, comprising the following steps:

[0011] Step 1: If L(t1) = L(t2), then the target remains stationary during the shutdown period;

[0012] Step 2: If L(t1)≠L(t2), determine whether a remote sensing satellite captured the target during the signal loss period. If not, proceed to Step 3. If it was captured, the capture time is t, and proceed to Step 4.

[0013] Step 3: The starting point L(t1) and ending point L(t2) of the missing path, and the duration of the missing path. Three recommended paths, p1, p2, and p3, are given.

[0014]

[0015] Where P represents the m recommended paths, p k This represents the recommended k-th path, where H is the time taken for that path. kLet k be the time taken for the k-th path.

[0016] At the same time, historical remote sensing imagery can detect hotspot activity areas of the target.

[0017]

[0018] Where S represents the target's hotspot activity area, and there are a total of n hotspot activity areas, s i Let W be the i-th hotspot region, and W be the probability weight of the target in each hotspot region.

[0019] Recommended path p k The overlap matrix of hotspot activity areas can be represented as C

[0020]

[0021]

[0022] The path with the closest time consumption and signal loss duration and the greatest overlap with the target hotspot activity area is selected as the estimated missing path.

[0023] Step 4: Divide the path into two segments. The first segment has a missing path starting point L(t1) and an ending point L(t), with a missing duration of [missing information]. The second missing path starts at L(t) and ends at L(t2), with a missing duration of [duration missing]. Then, step 3 is performed on the missing paths of the first and second segments respectively to estimate the missing paths.

[0024] In step 1, the target loses its signal at time t1, and the corresponding signal loss point is L(t1). The target recaptures its signal at time t2, and the corresponding signal capture point is L(t2).

[0025] The beneficial effects of this invention are as follows: To solve the problem of intelligent fusion of multi-source heterogeneous information in aerospace, this invention provides an intelligent fusion mechanism for sparse, non-synchronous, and multi-source heterogeneous information. It integrates target, time, and spatial information from aerospace electronic signal sensing data, aerospace optical remote sensing data, and map data, and completes incomplete, sparse, non-synchronous, and multi-source heterogeneous information. This solves the problem of low temporal and spatial resolution of existing aerospace spatiotemporal data. It can not only provide accurate input for the correlation analysis of multi-source information in aerospace, but also timely grasp key sensitive information such as the importance, scale, and urgency of events. Attached Figure Description

[0026] Figure 1 This is a schematic diagram illustrating signal loss and recapture.

[0027] Figure 2A schematic diagram of the aerospace multi-source heterogeneous information intelligent fusion method provided by the present invention;

[0028] Figure 3 This is a schematic diagram of route planning and hotspot areas;

[0029] Figure 4 The image shows the target as captured by an optical remote sensing satellite. Detailed Implementation

[0030] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0031] Although aerospace remote sensing capabilities possess a large amount of aerospace imagery data, they have almost no continuous data for the same target or related targets. Aerospace electronic signal sensing capabilities can continuously detect the target's location, time, velocity, and signal strength information after it is powered on, but they cannot determine the target type, and they cannot obtain any information about the target after it is powered off (electromagnetic silence).

[0032] This invention studies the intelligent fusion mechanism of sparse, non-simultaneous, and multi-source heterogeneous information to obtain more complete time, location, target, and event information. Table 1 shows the aerospace spatiotemporal big data that can support "time, location, and target". The following describes in detail the method of fusing time, location, and target information to complete the spatiotemporal trajectory of the target.

[0033] Table 1: Spacetime Big Data Supporting "Time, Location, and Target"

[0034]

[0035]

[0036] Since electronic signal sensing data can acquire continuous spatiotemporal data of the target after it is powered on, it is necessary to supplement the path information during the target's shutdown period, such as... Figure 1 As shown, the target loses its signal at time t1, and the corresponding signal loss point is L(t1). The target reacquires its signal at time t2, and the corresponding signal reacquisition point is L(t2).

[0037] This invention proposes to utilize Figure 2 The method shown estimates the path between L(t1) and L(t2).

[0038] The intelligent fusion method for multi-source heterogeneous information in aerospace includes the following steps:

[0039] Step 1: If L(t1) = L(t2), then the target remains stationary during the shutdown period.

[0040] Wherein, the target loses its signal at time t1, and the corresponding signal loss point is L(t1). The target recaptures its signal at time t2, and the corresponding signal capture point is L(t2).

[0041] Step 2: If L(t1)≠L(t2), determine whether a remote sensing satellite captured the target during the signal loss period. If not, proceed to Step 3. If it was captured at time t, proceed to Step 4.

[0042] Step 3: The starting point L(t1) and ending point L(t2) of the missing path, and the duration of the missing path. To utilize remote sensing imagery to determine target types and their speeds, and to plan paths for these targets using map information, multiple paths can be recommended, along with the time required for each path to be traversed. For example... Figure 3 The document provides three recommended paths, p1, p2, and p3.

[0043]

[0044] Where P represents the m recommended paths, p k This represents the recommended k-th path, where H is the time taken for that path. k Let be the time taken for the k-th path.

[0045] At the same time, historical remote sensing imagery can detect hotspot activity areas of the target.

[0046]

[0047] Where S represents the target's hotspot activity area, and there are a total of n hotspot activity areas, s i Let be the i-th hotspot region. W represents the probability weight of the target being in each hotspot region. i The larger the value, the more likely the target is to appear in s. i If a certain recommended path and s i If the paths overlap, the probability that the path in question is the estimated path increases. For example... Figure 3 The text presents four hotspot areas, with darker black dots indicating a higher probability weight.

[0048] Recommended path p k The overlap matrix of hotspot activity areas can be represented as C k

[0049]

[0050]

[0051] When s i Compared with the recommended path p kWhen they do not overlap, the corresponding hotspot region s i The overlapping positions are marked as 0, and the opposite positions are marked as 1.

[0052] The path whose latency and signal loss duration are closest and which overlaps most with the target hotspot activity area is selected as the estimated missing path. For example, Figure 3 The recommended path should be p1.

[0053] Step 4: As Figure 4 The path is divided into two segments. The first segment has a missing path starting point L(t1) and an ending point L(t), with a missing duration of [missing information]. The second missing path starts at L(t) and ends at L(t2), with a missing duration of [duration missing]. Then, step 3 is performed on the missing paths of the first and second segments respectively to estimate the missing paths.

Claims

1. A method for intelligent fusion of multi-source heterogeneous information in aerospace, characterized in that, Includes the following steps: Step 1: If The target remains stationary during the shutdown period; The goal is The signal is lost at a certain moment, and the corresponding location of the signal loss point is: ; The goal is The signal is recaptured at a specific time, and the corresponding signal capture point location is... ; Step 2: If Determine whether a remote sensing satellite captured the target during the period of signal loss. If not, proceed to step 3. If it was captured, the capture time is t, and proceed to step 4. Step 3: Starting point of the missing path The destination is Missing duration Three recommended paths are given. ; Step 3 includes, （1） Where P represents the m recommended paths, Let H be the recommended k-th path, and H be the time taken for that path. Let be the time taken for the k-th path; At the same time, historical remote sensing imagery can detect hotspot activity areas of the target. （2） Where S represents the target's hotspot activity areas, and there are a total of n hotspot activity areas. Let W be the i-th hotspot region, and W be the probability weight of the target in each hotspot region. Recommended path The overlap matrix of hotspot activity areas can be represented as C , （3） （4） The path with the closest latency and signal loss duration, and which overlaps most with the target hotspot activity area, is selected as the estimated missing path. ; Step 4: Divide the path into two segments. The first segment is missing the starting point of the path. The destination is Missing duration The starting point of the second missing path The destination is Missing duration Then, step 3 is performed on the missing paths of the first and second segments respectively to estimate the missing paths.

2. The aerospace multi-source heterogeneous information intelligent fusion method as described in claim 1, characterized in that, If no image is captured in step 2, proceed to step 3; if an image is captured at time t, proceed to step 4.

Images