Maritime communication method based on Beidou short message

Abstract

The embodiment of the invention provides a maritime communication method based on a Beidou short message, which is applied to the technical field of communication, and comprises the following steps: S1, collecting maritime navigation original data of a ship in real time, and carrying out space-time alignment and normalization processing on the maritime navigation original data to generate a multi-modal data set; s2, performing data preprocessing on the multi-modal data set, extracting feature vectors of single-modal data, and combining the feature vectors to generate a feature vector data set; s3, performing data fusion on the feature vector data set to generate a fused feature vector data set; s4, encoding, compressing and packaging the fusion feature vector data set to generate a Beidou short message data packet, and sending the Beidou short message data packet to a shore-based monitoring center; and S5, the shore-based monitoring center generates a decision suggestion based on the received Beidou short message data packet. In this way, the technical problem of multi-modal data fusion is solved, the integrity and accuracy of data are improved, and the efficiency of data transmission is improved.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and particularly to a maritime communication method based on Beidou short messages. Background Art

[0002] In the field of maritime communication, the information interaction between ships and shore-based monitoring centers is crucial, which is related to the safety, efficiency of maritime navigation, and the rationality of resource management. Traditional maritime communication methods have many limitations. For example, satellite communication is costly, and the signal coverage is unstable in some remote sea areas, which is prone to cause communication interruptions; while traditional methods such as high-frequency communication are greatly affected by factors such as distance and weather, and it is difficult to guarantee the reliability and real-time nature of data transmission.

[0003] With the gradual improvement of the Beidou satellite navigation system, the Beidou short message communication technology, as a new maritime communication means, has unique advantages. In particular, it is not restricted by the ground network and can achieve reliable short message communication globally, especially suitable for communication requirements in remote areas such as the sea.

[0004] However, the current maritime communication method based on Beidou short messages still has the following disadvantages in data transmission and processing:

[0005] (1) Maritime communication involves ship operation data (such as positioning, speed, equipment status), environmental perception data (such as meteorology, ocean parameters), and multi-modal data such as vision and acoustics. There are problems such as heterogeneous formats, large differences in sampling frequencies, and huge data volumes, making it difficult to effectively integrate and process multi-modal data, resulting in insufficient information acquisition.

[0006] (2) The communication capacity of Beidou short messages is limited, and the amount of data transmitted each time is small, usually only a few hundred bytes, and it is impossible to directly transmit a large amount of data, which limits the transmission and acquisition of data.

[0007] Therefore, the present invention proposes a maritime communication method based on Beidou short messages to solve the above technical problems. Summary of the Invention

[0008] The present invention provides a maritime communication method based on Beidou short messages, which solves the technical problem of multi-modal data fusion, improves the integrity and accuracy of data, and improves the efficiency of data transmission.

[0009] According to the provided maritime communication method based on Beidou short messages, it includes the following steps:

[0010] S1. Through multiple sensors deployed on the ship, the original maritime navigation data of the ship is collected in real time, and the original maritime navigation data is subjected to spatio-temporal alignment and normalization processing to generate a multi-modal data set;

[0011] S2. Perform data preprocessing on the unimodal data in the multimodal dataset respectively, extract the feature vectors of the unimodal data, and combine the feature vectors to generate a feature vector dataset corresponding to the multimodal dataset;

[0012] S3. Perform data fusion on the feature vector dataset through a fusion algorithm to generate a fused feature vector dataset;

[0013] S4. Encode, compress, and encapsulate the fused feature vector dataset to generate a Beidou short message data packet, and send the Beidou short message data packet to the shore-based monitoring center;

[0014] S5. The shore-based monitoring center receives the Beidou short message data packet sent by the ship and generates decision-making suggestions based on the Beidou short message data packet.

[0015] Further, the original maritime navigation data includes ship operation data and environmental perception data;

[0016] The ship operation data includes ship positioning data and ship status data;

[0017] The environmental perception data includes meteorological data, marine environmental data, visual data, and acoustic data.

[0018] Further, the step S1 specifically includes the following steps:

[0019] S11. All the sensors calibrate the local clock through the Beidou second pulse signal, and control the hardware time deviation of all the sensors at the microsecond level;

[0020] S12. The sensors collect the original maritime navigation data in real time according to the acquisition frequency preset by the system. The formula for the original maritime navigation data is:

[0021] D(t) = [D1(t), D2(t), …, D i (t), …, D n (t)]

[0022] where i represents the number of the sensor, 1 ≤ i ≤ n, t represents the timestamp after GNSS calibration, D i (t) represents the original data collected by the i-th sensor at time t, and n represents the total number of the sensors;

[0023] S13. Align the original maritime navigation data in time. First, define the alignment time sequence:

[0024] t k = t0 + k·Δt, k = 0, 1, 2…

[0025] Among them, t0 represents the whole second moment when the first data packet arrives, Δt represents the alignment time interval, and t k represents the k-th alignment timestamp;

[0026] Interpolate and calculate the data of each of the sensors. For each sensor i, traverse the original data D i (t) it collects, and find each t k corresponding interpolation point, and use the linear interpolation algorithm to perform interpolation calculation. Its calculation formula is:

[0027]

[0028] where, t k represents the timestamp in the alignment time series, t j and t j+1 represent the timestamps when the sensor i actually collects the original data, and t j and t j+1 are the two nearest timestamps before and after t k , D i (t j ) and D i (t j+1 ) respectively represent the original data collected by the sensor i at the timestamps t j and t j+1 , and D i (t k ) represents the interpolated data of the sensor i obtained by interpolation calculation at the timestamp t k ;

[0029] Combine the interpolation results of all the sensors at the timestamp t k to form a time-aligned data set:

[0030] D 对齐 (t k ) = [D1(t k ), D2(t k ), …, D i (t k ), …, D n (t k )]

[0031] where, D 对齐 (t k ) represents the time-aligned data set;

[0032] S14. For the time-aligned data set D 对齐 (t k ), perform data Min-Max normalization processing, and normalize the time-aligned data set D 对齐 (tk ) The data is transformed into a unified numerical range [0, 1], and its calculation formula is:

[0033]

[0034] Among them, D min and D max respectively represent the minimum value and the maximum value in the time-aligned dataset D 对齐 (t k ), and D 归一化 (t k ) represents the normalized dataset;

[0035] S15. Further integrate the normalized dataset D 归一化 (t k ) into a multi-modal dataset. For each timestamp t k , generate the multi-modal dataset M(t k ) of the timestamp t k , and its calculation formula is:

[0036]

[0037] And store the multi-modal dataset M(t k ) in the JSON data format.

[0038] Further, the step S2 specifically includes the following steps:

[0039] S21. Perform data preprocessing on the multi-modal dataset M(t k ). Use statistical methods to identify and remove the noise and outliers in the multi-modal dataset M(t k ), and fill in the missing values to obtain the multi-modal high-quality dataset M 干净 (t k );

[0040] S22. For the multi-modal high-quality dataset M 干净 (t k ), extract the key features for the data of different modalities respectively, and generate the feature vector of the timestamp t k , and its calculation formula is:

[0041] F(t k ) = [F 定位 (t k ), F 状态 (t k ), F 气象 (t k ), F 环境 (t k ), F视觉 (t k ),F 声学 (t k )]

[0042] Among them, F 定位 (t k ), F 状态 (t k ), F 气象 (t k ), F 环境 (t k ), F 视觉 (t k ), F 声学 (t k ) respectively represent the ship positioning feature vector, ship state feature vector, meteorological feature vector, ocean environment feature vector, visual feature vector, and acoustic feature vector;

[0043] Combine the feature vectors of the time stamp t k in the aligned time series to generate a feature vector dataset F, and its calculation formula is:

[0044] F = {F(t1), F(t2), …, F(t k )}

[0045] Among them, F represents the feature vector dataset.

[0046] Furthermore, the step S3 specifically includes the following steps:

[0047] S31. Fuse the feature vectors F(t k ) of all modalities by the weighted average method to obtain a fused feature vector F 融合 (t k ), and its calculation formula is:

[0048]

[0049] Among them, m represents the number of modalities, w i represents the weight of the i-th modality, F i (t k ) represents the feature vector of the i-th modality at the time stamp t k , and F 融合 (t k ) represents the fused feature vector;

[0050] S32. Combine the fused feature vectors of the time stamp t k in the aligned time series to generate a fused feature vector dataset F 融合 , and its calculation formula is

[0051] F融合 = {F 融合 (t1), F 融合 (t2), …, F 融合 (t k )}

[0052] Among them, F 融合 represents the fused feature vector data set.

[0053] Furthermore, the step S4 specifically includes the following steps:

[0054] S41. Compress the fused feature vector data set F 融合 using Huffman coding and encapsulate it into a Beidou short message data packet according to the communication protocol of Beidou short message;

[0055] S42. Send the Beidou short message data packet to the shore-based monitoring center.

[0056] Furthermore, the step S5 specifically includes the following steps:

[0057] S51. The shore-based monitoring center receives the Beidou short message data packet, unpacks and decompresses the Beidou short message data packet to obtain the fused feature vector data set F 融合 ;

[0058] S52. The shore-based monitoring center deeply analyzes the fused feature vector data set F 融合 and generates decision-making suggestions.

[0059] The technical effects of the present invention are as follows:

[0060] (1) Through time alignment and normalization processing, this solution integrates ship positioning data, ship status data, meteorological data, ocean environment data, visual data, and acoustic data into a unified multi-modal data set, ensuring the temporal synchronization and numerical range consistency of different modal data;

[0061] (2) By extracting key features from the data of each modality and fusing the key features, this solution not only reduces the amount of data to be transmitted, but also highlights the important information of each modality data, improving the usability and analysis efficiency of each modality data;

[0062] (3) By compressing the fused feature vector data set through Huffman coding, this solution significantly reduces the amount of data, and the compressed data is segmented into multiple data packets and encapsulated into data packets according to the communication protocol of Beidou short message, ensuring the efficient transmission of data under limited communication capacity.

[0063] It should be understood that the content described in the Summary of the Invention section is not intended to limit the key or important features of the embodiments of the present invention, nor to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In conjunction with the accompanying drawings and with reference to the following detailed description, the above and other features, advantages, and aspects of the embodiments of the present invention will become more apparent. The drawings are used to better understand the solution and do not constitute a limitation to the present invention. In the drawings, the same or similar reference numerals represent the same or similar elements, where:

[0065] Figure 1 is a flowchart of the maritime communication method based on Beidou short message of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0067] In addition, the term "and / or" in this article is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this article generally represents an "or" relationship between the associated objects before and after.

[0068] It should be understood that various forms of the processes shown above can be used, reordering, adding, or deleting steps. For example, the steps described in the present invention can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in the present invention can be achieved. This is not limited herein.

[0069] As Figure 1 shown, the present invention proposes a maritime communication method based on Beidou short message, including the following steps: S1. Through a plurality of sensors deployed on a ship, the original data of the ship's maritime navigation is collected in real time, and the original data of the maritime navigation is subjected to spatio-temporal alignment and normalization processing to generate a multi-modal data set;

[0070] S2. Data preprocessing is respectively performed on the single-modal data in the multi-modal data set, the feature vectors of the single-modal data are extracted, and the feature vectors are combined to generate a feature vector data set corresponding to the multi-modal data set;

[0071] S3. Perform data fusion on the feature vector data set through a fusion algorithm to generate a fused feature vector data set;

[0072] S4. Encode, compress, and encapsulate the fused feature vector data set to generate a Beidou short message data packet, and send the Beidou short message data packet to the shore-based monitoring center;

[0073] S5. The shore-based monitoring center receives the Beidou short message data packet sent by the ship and generates decision-making suggestions based on the Beidou short message data packet.

[0074] Furthermore, the original maritime navigation data includes ship operation data and environmental perception data;

[0075] The ship operation data includes ship positioning data and ship status data;

[0076] The environmental perception data includes meteorological data, marine environmental data, visual data, and acoustic data.

[0077] Preferably, the sensors include ship operation data sensors and environmental perception data sensors. The ship operation data sensors are used to collect ship operation data, and the environmental perception data sensors are used to collect environmental perception data;

[0078] The ship operation data sensors include ship positioning sensors and ship status sensors;

[0079] The environmental perception data sensors include meteorological sensors, marine environmental sensors, visual sensors, and acoustic sensors;

[0080] The types of various sensors and the monitored data are shown in the following table:

[0081]

[0082]

[0083] Furthermore, step S1 specifically includes the following steps:

[0084] S11. All the sensors calibrate their local clocks through the Beidou second pulse signal, and control the hardware time deviation of all the sensors at the microsecond level;

[0085] S12. The sensors collect the original maritime navigation data in real time according to the acquisition frequency preset by the system. The formula for the original maritime navigation data is:

[0086] D(t) = [D1(t), D2(t), …, D i (t), …, D n (t)]

[0087] wherein, i represents the number of the sensor, 1 ≤ i ≤ n, t represents the timestamp after GNSS calibration, and D i (t) represents the raw data collected by the i-th sensor at time t, and n represents the total number of the sensors;

[0088] S13. Align the raw data of the sea voyage in time. First, define the alignment time series:

[0089] t k = t0 + k·Δt, k = 0, 1, 2…

[0090] wherein, t0 represents the integer second moment when the first data packet arrives, Δt represents the alignment time interval, and t k represents the k-th alignment timestamp;

[0091] Perform interpolation calculation on the data of each sensor. For each sensor i, traverse the collected raw data D i (t), find the interpolation points corresponding to each t k , and use the linear interpolation algorithm to perform interpolation calculation. The calculation formula is:

[0092]

[0093] wherein, t k represents the timestamp in the alignment time series, t j and t j+1 represent the timestamps when the sensor i actually collects raw data, and t j and t j+1 are the two nearest timestamps before and after t k , D i (t j ) and D i (t j+1 ) respectively represent the raw data collected by the sensor i at the timestamps t j and t j+1 , and D i (t k ) represents the interpolation data of the sensor i obtained by interpolation calculation at the timestamp t k ;

[0094] Combine the interpolation results of all the sensors at the timestamp t k to form a time-aligned data set:

[0095] D 对齐 (t k ) = [D1(t k ), D2(t k ), …, D i (t k ), …, Dn (t k )]

[0096] Among them, D 对齐 (t k ) represents the time-aligned dataset;

[0097] S14. Perform data Min-Max normalization on the time-aligned dataset D 对齐 (t k ), and transform the data of the time-aligned dataset D 对齐 (t k ) into a unified numerical range [0, 1]. The calculation formula is:

[0098]

[0099] Among them, D min and D max respectively represent the minimum and maximum values in the time-aligned dataset D 对齐 (t k ), and D 归一化 (t k ) represents the normalized dataset;

[0100] S15. Further integrate the normalized dataset D 归一化 (t k ) into a multi-modal dataset. For each timestamp t k , generate the multi-modal dataset M(t k ) of the timestamp t k . The calculation formula is:

[0101]

[0102] And store the multi-modal dataset M(t k ) in the JSON data format.

[0103] Furthermore, the step S2 specifically includes the following steps:

[0104] S21. Perform data preprocessing on the multi-modal dataset M(t k ), use statistical methods to identify and remove noise and outliers in the multi-modal dataset M(t k ), and fill in missing values to obtain the multi-modal high-quality dataset M 干净 (t k ) after data cleaning;

[0105] S22. Perform on the multi-modal high-quality dataset M 干净 (t k) Key feature extraction is performed on data of different modalities respectively to generate a timestamp t k The feature vector of is calculated by the formula:

[0106] F(t k ) = [F 定位 (t k ), F 状态 (t k ), F 气象 (t k ), F 环境 (t k ), F 视觉 (t k ), F 声学 (t k )]

[0107] Among them, F 定位 (t k ), F 状态 (t k ), F 气象 (t k ), F 环境 (t k ), F 视觉 (t k ), F 声学 (t k ) respectively represent the ship positioning feature vector, the ship status feature vector, the meteorological feature vector, the ocean environment feature vector, the visual feature vector, and the acoustic feature vector;

[0108] Combine the feature vectors of the timestamp t k in the aligned time series to generate a feature vector dataset F, and its calculation formula is:

[0109] F = {F(t1), F(t2), …, F(t k )}

[0110] Among them, F represents the feature vector dataset.

[0111] Preferably, the formula for the ship positioning feature vector F 定位 (t k ) is:

[0112]

[0113] Preferably, the formula for the ship status feature vector F 状态 (t k ) is:

[0114]

[0115] Preferably, the meteorological feature vector F气象 (t k ) has the formula:

[0116]

[0117] Preferably, the marine environmental feature vector F 环境 (t k ) has the formula: seawater temperature, salinity, depth, flow velocity, flow direction

[0118]

[0119] Preferably, the visual feature vector F 视觉 (t k ) has the formula:

[0120]

[0121] Preferably, the acoustic feature vector F 声学 (t k ) has the formula:

[0122]

[0123] Furthermore, step S3 specifically includes the following steps:

[0124] S31. Fuse the feature vectors F(t k ) of all modalities by the weighted average method to obtain the fused feature vector F 融合 (t k ), and its calculation formula is:

[0125]

[0126] where m represents the number of modalities, w i represents the weight of the i-th modality, F i (t k ) represents the feature vector of the i-th modality at the timestamp t k , and F 融合 (t k ) represents the fused feature vector;

[0127] S32. Combine the fused feature vectors at the timestamp t k in the aligned time series to generate the fused feature vector dataset F 融合 , and its calculation formula is

[0128] F 融合 = {F 融合 (t1), F 融合 (t2), …, F 融合 (t k )}

[0129] Among them, F 融合 represents the fused feature vector data set.

[0130] Further, the step S4 specifically includes the following steps:

[0131] S41. Compress the fused feature vector data set F 融合 using Huffman coding and encapsulate it into a Beidou short message data packet according to the communication protocol of Beidou short message;

[0132] S42. Send the Beidou short message data packet to the shore-based monitoring center.

[0133] Preferably, in step S41, traverse the fused feature vector data set F 融合 , count the occurrence frequency of each feature vector, construct a Huffman tree according to the frequency, the higher the frequency of the feature vector, the shorter the path in the Huffman tree, and then start from the root node of the Huffman tree, generate a unique binary code for each feature vector, and replace each feature vector in the fused feature vector data set F 融合 with its corresponding Huffman code to obtain the fused feature vector data set F 融合 the Huffman-coded data F 编码 after encoding;

[0134] Huffman coding is a prior art and will not be elaborated here.

[0135] Preferably, in step S42, add a Beidou short message protocol header before the data in the Huffman-coded data F 编码 according to the data packet format required by the Beidou short message communication protocol, encapsulate the Huffman-coded data F 编码 . If the size of the encapsulated data exceeds the maximum length limit of the Beidou short message, the encapsulated data needs to be sub-packaged, and a Beidou short message protocol header is added to each sub-package. Finally, the encapsulated Beidou short message data packet is sent to the shore-based monitoring center through the Beidou satellite communication link.

[0136] Further, the step S5 specifically includes the following steps:

[0137] S51. The shore-based monitoring center receives the Beidou short message data packet, unpacks and decompresses the Beidou short message data packet to obtain the fused feature vector data set F 融合 ;

[0138] S52. The shore-based monitoring center deeply analyzes the fused feature vector data set F 融合 and generates decision-making suggestions.

[0139] Preferably, in step S51, the shore-based monitoring center receives the Beidou short message data packet, unpacks it according to the Beidou short message communication protocol, removes the protocol header, merges the sub-packet data, and decompresses the compressed data using the same Huffman coding table as the ship sending end to obtain the fused feature vector dataset F 融合

[0140] Preferably, in step S52, for the decompressed fused feature vector dataset F 融合 perform in-depth analysis, extract key features from the fused feature vector dataset F 融合 remove redundancy, and use deep learning method to perform in-depth analysis on the feature vectors and generate corresponding decision suggestions;

[0141] Preferably, using deep learning method to perform in-depth analysis on the feature vectors and generate corresponding decision suggestions, this is the prior art and will not be elaborated here.

[0142] The above specific embodiments do not constitute a limitation to the protection scope of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A maritime communication method based on Beidou short message, characterized in that, It includes the following steps: S1. Real-time collect the original data of the ship's sea voyage through multiple sensors deployed on the ship, perform spatio-temporal alignment and normalization processing on the original data of the sea voyage, and generate a multi-modal data set; S2. Perform data preprocessing on the single-modal data in the multi-modal data set respectively, extract the feature vectors of the single-modal data, and combine the feature vectors to generate a feature vector data set corresponding to the multi-modal data set; S3. Perform data fusion on the feature vector data set through a fusion algorithm to generate a fused feature vector data set; S4. Encode, compress, and encapsulate the fused feature vector data set to generate a Beidou short message data packet, and send the Beidou short message data packet to the shore-based monitoring center; S5. The shore-based monitoring center receives the Beidou short message data packet sent by the ship and generates decision-making suggestions based on the Beidou short message data packet.

2. The maritime communication method based on Beidou short message according to claim 1, wherein The original data of the sea voyage includes ship operation data and environmental perception data; The ship operation data includes ship positioning data and ship status data; The environmental perception data includes meteorological data, marine environment data, visual data, and acoustic data.

3. The maritime communication method based on Beidou short message according to claim 2, wherein The specific steps of step S1 include the following steps: S11. All the sensors calibrate the local clock through the Beidou second pulse signal, and control the hardware time deviation of all the sensors at the microsecond level; S12. The sensors collect the original data of the sea voyage in real time according to the acquisition frequency preset by the system. The formula for the original data of the sea voyage is: D(t) = [D1(t), D2(t), …, D i (t), …, D n (t)] where i represents the number of the sensor, 1 ≤ i ≤ n, t represents the timestamp after GNSS calibration, and D i (t) represents the raw data collected by the i-th sensor at time t, and n represents the total number of the sensors; S13. Align the original data of the sea voyage in time. First, define the alignment time series: t k = t0 + k·Δt, k = 0, 1, 2… Among them, t0 represents the whole second moment when the first data packet arrives, Δt represents the alignment time interval, and t k represents the k-th alignment timestamp; Interpolate the data of each of the sensors. For each sensor i, traverse the original data D i (t) it has collected, and find each t k corresponding interpolation point. Using the linear interpolation algorithm, perform interpolation calculation. Its calculation formula is: where t k represents the timestamp in the aligned time series, t j and t j+1 represent the timestamps when sensor i actually acquires the original data, and t j and t j+1 are the two nearest timestamps before and after t k . D i (t j ) and D i (t j+1 ) respectively represent the original data acquired by sensor i at timestamps t j and t j+1 . D i (t k ) represents the interpolated data of sensor i obtained by interpolation calculation at timestamp t k ; Interpolate all the sensors at timestamp t k to form a time-aligned dataset: D 对齐 (t k ) = [D1(t k ), D2(t k ), …, D i (t k ), …, D n (t k )] Among them, D 对齐 (t k ) represents a time-aligned data set; S14. Perform data Min-Max normalization on the time-aligned dataset D 对齐 (t k ), and convert the data of the time-aligned dataset D 对齐 (t k ) to a unified numerical range of [0, 1]. The calculation formula is as follows: Among them, D min and D max respectively represent the minimum value and the maximum value in the time-aligned data set D 对齐 (t k ), and D 归一化 (t k ) represents the normalized data set; S15. Further integrate the normalized dataset D 归一化 (t k ) into a multi-modal dataset. For each timestamp t k , generate the multi-modal dataset M(t k ) of the timestamp t k . Its calculation formula is: and store the multimodal dataset M(t k ) in JSON data format.

4. The maritime communication method based on Beidou short message according to claim 3, characterized in that, The specific steps of step S2 include the following steps: S21. Preprocess the multimodal dataset M(t k ) by using statistical methods to identify and remove noise and outliers in the multimodal dataset M(t k ), fill in missing values, and obtain the cleaned high-quality multimodal dataset M 干净 (t k ); S22. For the multi-modal high-quality dataset M 干净 (t k ), key feature extraction is performed on the data of different modalities respectively to generate a feature vector at time stamp t k . Its calculation formula is as follows: F(t k ) = [F 定位 (t k ), F 状态 (t k ), F 气象 (t k ), F 环境 (t k ), F 视觉 (t k ), F 声学 (t k )] Among them, F 定位 (t k )、F 状态 (t k )、F 气象 (t k )、F 环境 (t k )、F 视觉 (t k )、F 声学 (t k ) respectively represent the ship positioning feature vector, ship state feature vector, meteorological feature vector, marine environment feature vector, visual feature vector, and acoustic feature vector; Combine the feature vectors of the time stamps t in the aligned time series k to generate a feature vector data set F, and its calculation formula is: F = {F(t1), F(t2), …, F(t k )} where F represents the feature vector data set.

5. The maritime communication method based on Beidou short message according to claim 4, characterized in that The specific steps of step S3 include the following steps: S31. Fuse the feature vectors F(t k ) of all modalities by weighted averaging to obtain the fused feature vector F 融合 (t k ), and its calculation formula is as follows: Among them, m represents the number of modes, and w i represents the weight of the i-th mode, and F i (t k ) represents the eigenvector of the i-th mode at the timestamp t k , and F 融合 (t k ) represents the fused eigenvector; S32. Combine the feature vectors of the aligned time series with timestamps t k to generate a fused feature vector dataset F 融合 , and its calculation formula is F 融合 = {F 融合 (t1), F 融合 (t2), …, F 融合 (t k )} Among them, F 融合 represents the fused feature vector data set.

6. The maritime communication method based on Beidou short message according to claim 5, characterized in that, The specific steps of step S4 include the following steps: S41. Compress the fused feature vector dataset F 融合 using Huffman coding, and encapsulate it into a Beidou short message data packet according to the communication protocol of Beidou short messages; S42. Send the Beidou short message data packet to the shore-based monitoring center.

7. The maritime communication method based on Beidou short message according to claim 5, characterized in that, The specific steps of step S5 include the following steps: S51. The shore-based monitoring center receives the Beidou short message data packet, unpacks and decompresses the Beidou short message data packet to obtain the fusion feature vector data set F 融合 ; S52. The shore-based monitoring center performs in-depth analysis on the fused feature vector dataset F 融合 and generates decision-making suggestions.

Images