Fishery data fusion method for data with different resolutions

Abstract

The invention relates to the field of fishery data fusion, and discloses a fishery data fusion method for data with different resolutions, which comprises the following steps of: collecting various fishery data in a fishery operation area, and classifying the fishery data based on the resolutions of the fishery data, after classification, interpolation processing and time sequence resampling processing are carried out on fishery data to realize data resolution unification; after fishery data resolution is unified, weight distribution and space-time alignment are carried out on the data, and finally data features are extracted to construct a fish distribution prediction model. According to the method, high-resolution and low-resolution data processing can be supported at the same time, the problem that a traditional method is poor in adaptability to multi-source data is solved, the fusion precision is remarkably improved, and deep fishery information can be mined.

Description

Technical Field

[0001] The invention relates to the field of fishery data fusion, in particular to a fishery data fusion method for data with different resolutions. Background Art

[0002] The fusion of fishery acoustic data and climate and environmental data provides key support for modern marine research. However, due to the diversity of data sources, such as acoustic data, satellite remote sensing data, ship-borne sensor data, etc., as well as the inconsistency of resolution and the difference in data quality, traditional data processing methods are difficult to effectively integrate this information. To this end, this paper proposes a practical and innovative data fusion method, which is suitable for the comprehensive application of fishery acoustic data and multi-source climate and environmental data. The data fusion method is a fishery data fusion method for data of different resolutions, which can effectively handle the differences in spatial resolution, temporal resolution and quality level of acoustic data, climate data and environmental data, and realize efficient data fusion and comprehensive analysis. Summary of the invention

[0003] The invention overcomes the shortcomings of the prior art and provides a fishery data fusion method for data with different resolutions.

[0004] In order to achieve the above object, the technical solution adopted by the present invention is: A first aspect of the present invention provides a fishery data fusion method for data of different resolutions, comprising the following steps: Conduct source analysis and resolution analysis of climate data, environmental data, and acoustic data in the fishery operation area to achieve classification and processing of different data in the fishery operation area; Clean the climate data, environmental data and acoustic data in the fishery operation area, and use interpolation technology and time series resampling algorithms to unify the resolution of various data in the fishery operation area; Based on the quality grade classification rules, weights are assigned to fishery data of various target resolutions, and combined with the spatiotemporal alignment algorithm, all target resolution fishery data are spatially aligned and time synchronized to obtain spatiotemporal synchronized fishery data. Based on the Bayesian inference network, multi-source fusion of all spatiotemporal synchronized fishery data is carried out, and feature extraction of the multi-source fused spatiotemporal synchronized fishery data is carried out based on the deep learning model to construct a fish distribution prediction model.

[0005] Furthermore, in a preferred embodiment of the present invention, the source analysis and resolution analysis of climate data, environmental data and acoustic data are performed in the fishery operation area to achieve classification processing of different data in the fishery operation area, specifically: Determine the fishery operation area and obtain the environmental data monitoring platform connected to the fishery operation area, and mark it as the fishery environmental data monitoring platform; The fishery environment data monitoring platform records the climate data and environmental data in the fishery operation area, wherein the climate data includes the wind speed and sea surface temperature in the fishery operation area, and the environmental data includes the salinity content and dissolved oxygen content in the fishery operation area; In the fishery environmental data monitoring platform, the data sources of climate data and environmental data in the fishery operation area are marked and named as fishery climate data source and fishery environmental data source. At the same time, the acoustic data of the fishery operation area is determined in the big data network, and the data source of the acoustic data is marked and named as fishery acoustic data source. For the acoustic data of the fishery operation area, audio editing software is obtained, and the acoustic data of the fishery operation area is converted into an audio file through the audio editing software, and marked as a fishery acoustic audio file, and the fishery acoustic audio file is parsed in the audio editing software to generate the sampling rate and bit depth information of the acoustic data of the fishery operation area; The sampling rate describes the time resolution of the acoustic data of the fishery operation area, and the bit depth information describes the amplitude resolution of the acoustic data of the fishery operation area. Based on the sampling rate and the bit depth information of the acoustic data of the fishery operation area, the acoustic resolution of the acoustic data of the fishery operation area is generated and calibrated as the first resolution. For the climate data and environmental data of the fishery operation area, the resolutions of the climate data and environmental data of the fishery operation area are determined through the corresponding fishery climate data sources and fishery environmental data sources, and calibrated as the second resolution and the third resolution respectively.

[0006] Furthermore, in a preferred embodiment of the present invention, the climate data, environmental data and acoustic data in the fishery operation area are cleaned, and the various data in the fishery operation area are processed in a unified resolution through interpolation technology and time series resampling algorithm, specifically: The climate data, environmental data and acoustic data in the fishery operation area are collectively referred to as fishery data, wherein the fishery data includes the data source and data resolution; The Min-max standardization algorithm is introduced to clean the fishery data, and the fishery data after data cleaning is obtained and calibrated as fishery cleaned data; Among them, the steps of the Min-max standardization algorithm for data cleaning of fishery data are: presetting a standard numerical interval, performing a linear transformation on the fishery data, and mapping the fishery data after the linear transformation within the standard numerical interval to achieve data cleaning processing; Performing resolution analysis on the first resolution, the second resolution, and the third resolution, and determining the highest resolution, which is calibrated as the target resolution; All fishery cleaned data with resolutions not equal to the target resolution are calibrated as fishery data with resolutions to be adjusted, and fishery clear data at the target resolution are calibrated as fishery data at the target resolution; Introducing a KNN interpolation algorithm, based on the target resolution fishery data, determining the K value in the KNN interpolation algorithm, calibrating it as the target K value, and calculating the Manhattan distance between different data points in the fishery data with a resolution to be adjusted and the target K value, and performing a weighted average calculation on different data points in the fishery data with a resolution to be adjusted according to the Manhattan distance, so that the fishery data with a resolution to be adjusted generates new data points, and the resolution of the fishery data with a resolution to be adjusted is achieved for the target resolution, thereby obtaining the fishery data with a target resolution; The target resolution fishery data, i.e., the resolution of climate data, environmental data and acoustic data in the fishery operation area, are all equal to the target resolution.

[0007] Furthermore, in a preferred embodiment of the present invention, the weights of various target resolution fishery data are allocated based on the quality grade classification rules, and all target resolution fishery data are spatially registered and time synchronized in combination with the spatiotemporal alignment algorithm to obtain spatiotemporal synchronized fishery data, specifically: Constructing data integrity thresholds and conducting data integrity analysis on target resolution fishery data, and combining the data integrity thresholds, marking the target resolution fishery data that is maintained within the data integrity threshold as the first-level data quality level, and marking the target resolution fishery data that is not maintained within the data integrity threshold as the second-level data quality level; Constructing a quality grade classification rule, wherein the quality grade classification rule is constructed according to the data integrity of the target resolution fishery data, and the data acquisition accuracy of all the target resolution fishery data is calculated, and the quality grade classification rule is optimized in combination with the data acquisition accuracy of all the target resolution fishery data to obtain an optimized quality grade classification rule; Wherein, the data sampling accuracy of the target resolution fishery data of the first-level data quality level and the second-level data quality level is marked in the optimized quality level classification rule; In the optimized quality grade classification rules, data weights are allocated to all target resolution fishery data according to the data quality level, so that target resolution fishery data of different quality levels are marked with weight factors; A spatiotemporal alignment algorithm is introduced to perform spatial registration and temporal synchronization on the target resolution fishery data and the corresponding marked weight factors to obtain spatiotemporal synchronized fishery data.

[0008] Furthermore, in a preferred embodiment of the present invention, the temporal and spatial alignment algorithm is introduced to perform spatial registration and temporal synchronization on the target resolution fishery data and the corresponding marked weight factors to obtain temporal and spatial synchronized fishery data, specifically: Introducing a spatiotemporal alignment algorithm, wherein the spatiotemporal alignment algorithm includes a spatial alignment algorithm and a temporal alignment algorithm, wherein the spatial alignment algorithm is a grid-based alignment algorithm, and the temporal alignment algorithm is a sliding window algorithm; A spatial alignment grid is preset, wherein the spatial alignment grid is a spatial coordinate system, and all target resolution fishery data are imported into the spatial alignment grid, so that the target resolution fishery data are aligned in the spatial alignment grid to obtain spatially aligned target resolution fishery data; Wherein, registering the target resolution fishery data in the spatial alignment grid means controlling the spatial information of the target resolution fishery data to be equal, and the spatial information of the target resolution fishery data is the geographical location coordinates of the target resolution fishery data; For all spatially registered target resolution fishery data, extracting time information of the spatially registered target resolution fishery data, wherein the time information of the spatially registered target resolution fishery data is the collection time of fishery data with different target resolutions; Based on the time alignment algorithm, the time information of all spatially registered target resolution fishery data is synchronized to obtain the target resolution fishery data after spatial registration and time information synchronization, which is calibrated as spatiotemporal synchronized fishery data.

[0009] Furthermore, in a preferred embodiment of the present invention, the multi-source fusion of all spatiotemporal synchronized fishery data is performed based on the Bayesian inference network, and feature extraction of the multi-source fused spatiotemporal synchronized fishery data is performed based on the deep learning model, so as to construct a fish distribution prediction model, specifically: Introduce the Bayesian inference network, and import all the spatiotemporal synchronized fishery data into the Bayesian inference network, wherein all the spatiotemporal synchronized fishery data need to be imported according to the conditional probability model; Based on the Bayesian inference network, a maximum a posteriori probability estimation is performed on all spatiotemporal synchronized fishery data, and conflicting fishery data and normal fishery data are generated after the maximum a posteriori probability estimation, wherein the conflicting fishery data are spatiotemporal synchronized fishery data whose maximum a posteriori probability is not within a preset range, and the normal fishery data are spatiotemporal synchronized fishery data whose maximum a posteriori probability is within a preset range; Performing weighted average processing on the conflicting fishery data until the conflicting fishery data is converted into normal fishery data, outputting the normal fishery data and performing multi-source fusion of the data to obtain spatiotemporal synchronized fishery data after multi-source fusion processing, and marking it as multi-source fused spatiotemporal synchronized fishery data; The CNN convolutional neural network is introduced, and data convolution analysis is performed on the multi-source fused spatiotemporal synchronized fishery data in the convolution layer of the CNN convolutional neural network. After the data convolution analysis, the time characteristics and spatial characteristics of the multi-source fused spatiotemporal synchronized fishery data are extracted in the output layer and calibrated as target time characteristics and target spatial characteristics. The principal component analysis method is used to eliminate the redundancy of target time features and target space features to obtain redundant processing time features and redundant processing space features. Based on the CNN convolutional neural network, a convolutional neural network model that can predict fish distribution is constructed by combining the redundant processing time features and the redundant processing space features. It is calibrated as a fish distribution prediction model.

[0010] The second aspect of the present invention further provides a fishery data fusion system for data of different resolutions, the fishery data fusion system comprising a memory and a processor, the memory storing a fishery data fusion method, and when the fishery data fusion method is executed by the processor, the following steps are implemented: Conduct source analysis and resolution analysis of climate data, environmental data, and acoustic data in the fishery operation area to achieve classification and processing of different data in the fishery operation area; Clean the climate data, environmental data and acoustic data in the fishery operation area, and use interpolation technology and time series resampling algorithms to unify the resolution of various data in the fishery operation area; Based on the quality grade classification rules, weights are assigned to fishery data of various target resolutions, and combined with the spatiotemporal alignment algorithm, all target resolution fishery data are spatially aligned and time synchronized to obtain spatiotemporal synchronized fishery data. Based on the Bayesian inference network, multi-source fusion of all spatiotemporal synchronized fishery data is carried out, and feature extraction of the multi-source fused spatiotemporal synchronized fishery data is carried out based on the deep learning model to construct a fish distribution prediction model.

[0011] The present invention solves the technical defects existing in the background technology, and has the following beneficial effects: various fishery data in the fishery operation area are collected, and the fishery data are classified based on the resolution of the fishery data, and after classification, the fishery data are interpolated and time series resampled to achieve unified data resolution; after the fishery data resolution is unified, the data are weighted and aligned in time and space, and finally the data features are extracted to construct a fish distribution prediction model. The present invention can support high-resolution and low-resolution data processing at the same time, solve the problem that traditional methods have poor adaptability to multi-source data, significantly improve the fusion accuracy, and is conducive to mining deep-level fishery information. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, drawings of other embodiments can be obtained based on these drawings without paying creative work.

[0013] Figure 1 A flow chart of a fishery data fusion method for data of different resolutions is shown; Figure 2 A flow chart of a method for obtaining spatiotemporally aligned fishery data is shown; Figure 3 A program view of a fishery data fusion system for data with different resolutions is shown. DETAILED DESCRIPTION

[0014] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0015] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0016] Figure 1 A flow chart of a fishery data fusion method for data of different resolutions is shown, comprising the following steps: S102: performing source analysis and resolution analysis of climate data, environmental data and acoustic data in the fishery operation area, and realizing classification processing of different data in the fishery operation area; S104: Clean the climate data, environmental data and acoustic data in the fishery operation area, and unify the resolution of various data in the fishery operation area through interpolation technology and time series resampling algorithm; S106: Based on the quality grade classification rules, weights are allocated to various target resolution fishery data, and combined with the spatiotemporal alignment algorithm, all target resolution fishery data are spatially aligned and time synchronized to obtain spatiotemporal synchronized fishery data; S108: Based on the Bayesian inference network, all spatiotemporal synchronized fishery data are fused from multiple sources, and features are extracted from the spatiotemporal synchronized fishery data after multi-source fusion based on the deep learning model, so as to construct a fish distribution prediction model.

[0017] Furthermore, in a preferred embodiment of the present invention, the source analysis and resolution analysis of climate data, environmental data and acoustic data are performed in the fishery operation area to achieve classification processing of different data in the fishery operation area, specifically: Determine the fishery operation area and obtain the environmental data monitoring platform connected to the fishery operation area, and mark it as the fishery environmental data monitoring platform; The fishery environment data monitoring platform records the climate data and environmental data in the fishery operation area, wherein the climate data includes the wind speed and sea surface temperature in the fishery operation area, and the environmental data includes the salinity content and dissolved oxygen content in the fishery operation area; In the fishery environmental data monitoring platform, the data sources of climate data and environmental data in the fishery operation area are marked and named as fishery climate data source and fishery environmental data source. At the same time, the acoustic data of the fishery operation area is determined in the big data network, and the data source of the acoustic data is marked and named as fishery acoustic data source. For the acoustic data of the fishery operation area, audio editing software is obtained, and the acoustic data of the fishery operation area is converted into an audio file through the audio editing software, and marked as a fishery acoustic audio file, and the fishery acoustic audio file is parsed in the audio editing software to generate the sampling rate and bit depth information of the acoustic data of the fishery operation area; The sampling rate describes the time resolution of the acoustic data of the fishery operation area, and the bit depth information describes the amplitude resolution of the acoustic data of the fishery operation area. Based on the sampling rate and the bit depth information of the acoustic data of the fishery operation area, the acoustic resolution of the acoustic data of the fishery operation area is generated and calibrated as the first resolution. For the climate data and environmental data of the fishery operation area, the resolutions of the climate data and environmental data of the fishery operation area are determined through the corresponding fishery climate data sources and fishery environmental data sources, and calibrated as the second resolution and the third resolution respectively.

[0018] It should be noted that the fishery operation area is the living and breeding area of ​​aquatic fish. In the fishery operation area, it is necessary to locate and distribute the fish, so it is necessary to build a model of fish location distribution. To build the model, it is necessary to first obtain the support of a lot of data, such as climate data, environmental data and acoustic data. According to these data, the purpose of providing data support for the fish positioning model can be achieved. Due to the diversity of data sources and inconsistent data resolution, it is impossible to effectively integrate useful data for data support, so it is necessary to perform data fusion processing on fishery data of different resolutions. After collecting the fishery data, it is necessary to determine its source. Among them, the fishery environmental data monitoring platform can directly retrieve the data sources of climate data and environmental data in the fishery operation area, and combine the acoustic data source to obtain the source of the fishery data. At the same time, based on the audio editing software, the acoustic data of the fishery operation area is converted into an audio file, calibrated as a fishery acoustic audio file, which is used to generate the sampling rate and bit depth information of the acoustic data. The audio editing software includes but is not limited to Audacity and other software. The sampling rate and bit depth information of the audio file can be viewed in the software interface, because the sampling rate determines the "resolution" of the audio signal in time, and the bit depth determines the "resolution" of the audio signal in amplitude, so the first resolution can be obtained. The resolution determination method for other types of data can be obtained by consulting data documents, etc., that is, the second resolution and the third resolution can be obtained.

[0019] Furthermore, in a preferred embodiment of the present invention, the climate data, environmental data and acoustic data in the fishery operation area are cleaned, and the various data in the fishery operation area are processed in a unified resolution through interpolation technology and time series resampling algorithm, specifically: The climate data, environmental data and acoustic data in the fishery operation area are collectively referred to as fishery data, wherein the fishery data includes the data source and data resolution; The Min-max standardization algorithm is introduced to clean the fishery data, and the fishery data after data cleaning is obtained and calibrated as fishery cleaned data; Among them, the steps of the Min-max standardization algorithm for data cleaning of fishery data are: presetting a standard numerical interval, performing a linear transformation on the fishery data, and mapping the fishery data after the linear transformation within the standard numerical interval to achieve data cleaning processing; Performing resolution analysis on the first resolution, the second resolution, and the third resolution, and determining the highest resolution, which is calibrated as the target resolution; All fishery cleaned data with resolutions not equal to the target resolution are calibrated as fishery data with resolutions to be adjusted, and fishery clear data at the target resolution are calibrated as fishery data at the target resolution; Introducing a KNN interpolation algorithm, based on the target resolution fishery data, determining the K value in the KNN interpolation algorithm, calibrating it as the target K value, and calculating the Manhattan distance between different data points in the fishery data with a resolution to be adjusted and the target K value, and performing a weighted average calculation on different data points in the fishery data with a resolution to be adjusted according to the Manhattan distance, so that the fishery data with a resolution to be adjusted generates new data points, and the resolution of the fishery data with a resolution to be adjusted is achieved for the target resolution, thereby obtaining the fishery data with a target resolution; The target resolution fishery data, i.e., the resolution of climate data, environmental data and acoustic data in the fishery operation area, are all equal to the target resolution.

[0020] It should be noted that the purpose of defining fishery data and cleaning fishery data is to remove abnormal data or fill missing data. The Min-max standardization algorithm is a cleaning algorithm that linearly transforms the data and maps the data to a specified numerical range. The formula is new data = (original data - minimum value) / (maximum value - minimum value). After cleaning the fishery data, the data needs to be interpolated, because the data after cleaning is not complete, and different data resolutions will lead to different consistency in the data analysis process, so the data resolution is unified through interpolation technology. The KNN interpolation algorithm is used to unify the data resolution, in which the K value in the KNN interpolation method is determined, that is, how many neighboring data points are selected to estimate the value of the target point. The selection of the K value usually needs to be determined according to the characteristics and requirements of the data. The distance between data points uses the Manhattan distance. The target resolution fishery data can be obtained based on the KNN algorithm.

[0021] Furthermore, in a preferred embodiment of the present invention, the multi-source fusion of all spatiotemporal synchronized fishery data is performed based on the Bayesian inference network, and feature extraction of the multi-source fused spatiotemporal synchronized fishery data is performed based on the deep learning model, so as to construct a fish distribution prediction model, specifically: Introduce the Bayesian inference network, and import all the spatiotemporal synchronized fishery data into the Bayesian inference network, wherein all the spatiotemporal synchronized fishery data need to be imported according to the conditional probability model; Based on the Bayesian inference network, a maximum a posteriori probability estimation is performed on all spatiotemporal synchronized fishery data, and conflicting fishery data and normal fishery data are generated after the maximum a posteriori probability estimation, wherein the conflicting fishery data are spatiotemporal synchronized fishery data whose maximum a posteriori probability is not within a preset range, and the normal fishery data are spatiotemporal synchronized fishery data whose maximum a posteriori probability is within a preset range; Performing weighted average processing on the conflicting fishery data until the conflicting fishery data is converted into normal fishery data, outputting the normal fishery data and performing multi-source fusion of the data to obtain spatiotemporal synchronized fishery data after multi-source fusion processing, and marking it as multi-source fused spatiotemporal synchronized fishery data; The CNN convolutional neural network is introduced, and data convolution analysis is performed on the multi-source fused spatiotemporal synchronized fishery data in the convolution layer of the CNN convolutional neural network. After the data convolution analysis, the time characteristics and spatial characteristics of the multi-source fused spatiotemporal synchronized fishery data are extracted in the output layer and calibrated as target time characteristics and target spatial characteristics. The principal component analysis method is used to eliminate the redundancy of target time features and target space features to obtain redundant processing time features and redundant processing space features. Based on the CNN convolutional neural network, a convolutional neural network model that can predict fish distribution is constructed by combining the redundant processing time features and the redundant processing space features. It is calibrated as a fish distribution prediction model.

[0022] It should be noted that the effect of the Bayesian inference network is to integrate multi-source data to maximize information gain. The method is to construct a Bayesian inference network, import all spatiotemporal synchronized fishery data into the Bayesian inference network, and regard it as a conditional probability input, and then use the maximum a posteriori probability method to estimate data fusion to determine whether there is conflicting data, that is, to determine whether there is data with poor fusion effect during the fusion process. If so, the weighted average algorithm is used to optimize the conflicting data so that the fused fishery data can achieve the purpose of maximizing information gain. Multi-source fused spatiotemporal synchronized fishery data is obtained, and the CNN convolutional neural network is combined to extract data features. Because a variety of fishery data has been fused at this time, the fish distribution prediction model can be constructed after extracting the features. The purpose of removing redundancy by principal component analysis is to improve feature accuracy. Finally, the CNN convolutional neural network can be combined to construct the fish distribution prediction model.

[0023] Figure 2 A flow chart of a method for obtaining spatiotemporally aligned fishery data is shown, comprising the following steps: S202: Based on the quality grade classification rules, weights are allocated to fishery data of various target resolutions, and combined with the spatiotemporal alignment algorithm, all fishery data of target resolutions are spatially aligned and time synchronized to obtain spatiotemporal synchronized fishery data; S204: Introduce a spatiotemporal alignment algorithm to perform spatial registration and temporal synchronization on the target resolution fishery data and the corresponding marked weight factors to obtain spatiotemporal synchronized fishery data.

[0024] Furthermore, in a preferred embodiment of the present invention, the weights of various target resolution fishery data are allocated based on the quality grade classification rules, and all target resolution fishery data are spatially registered and time synchronized in combination with the spatiotemporal alignment algorithm to obtain spatiotemporal synchronized fishery data, specifically: Constructing data integrity thresholds and conducting data integrity analysis on target resolution fishery data, and combining the data integrity thresholds, marking the target resolution fishery data that is maintained within the data integrity threshold as the first-level data quality level, and marking the target resolution fishery data that is not maintained within the data integrity threshold as the second-level data quality level; Constructing a quality grade classification rule, wherein the quality grade classification rule is constructed according to the data integrity of the target resolution fishery data, and the data acquisition accuracy of all the target resolution fishery data is calculated, and the quality grade classification rule is optimized in combination with the data acquisition accuracy of all the target resolution fishery data to obtain an optimized quality grade classification rule; Wherein, the data sampling accuracy of the target resolution fishery data of the first-level data quality level and the second-level data quality level is marked in the optimized quality level classification rule; Within the optimized quality grade classification rules, data weights are assigned to all target resolution fishery data according to the data quality levels, so that target resolution fishery data of different quality levels are marked with weight factors.

[0025] It should be noted that data integrity reflects the quality of fishery data. The more complete the fishery data is, the higher the quality of the data is. Combined with the data integrity threshold, the target resolution fishery data is divided into different data quality levels, and refined and optimized according to the data collection accuracy. The higher the data collection accuracy, the higher the data quality. Finally, based on the data integrity and data collection accuracy, an optimized quality level classification rule can be obtained to classify the data quality. Different weight factors need to be assigned to different quality levels because the weight factors participate in the subsequent data fusion calculation, and the participation of data with different weights in the subsequent calculation is different.

[0026] Furthermore, in a preferred embodiment of the present invention, the temporal and spatial alignment algorithm is introduced to perform spatial registration and temporal synchronization on the target resolution fishery data and the corresponding marked weight factors to obtain temporal and spatial synchronized fishery data, specifically: Introducing a spatiotemporal alignment algorithm, wherein the spatiotemporal alignment algorithm includes a spatial alignment algorithm and a temporal alignment algorithm, wherein the spatial alignment algorithm is a grid-based alignment algorithm, and the temporal alignment algorithm is a sliding window algorithm; A spatial alignment grid is preset, wherein the spatial alignment grid is a spatial coordinate system, and all target resolution fishery data are imported into the spatial alignment grid, so that the target resolution fishery data are aligned in the spatial alignment grid to obtain spatially aligned target resolution fishery data; Wherein, registering the target resolution fishery data in the spatial alignment grid means controlling the spatial information of the target resolution fishery data to be equal, and the spatial information of the target resolution fishery data is the geographical location coordinates of the target resolution fishery data; For all spatially registered target resolution fishery data, extracting time information of the spatially registered target resolution fishery data, wherein the time information of the spatially registered target resolution fishery data is the collection time of fishery data with different target resolutions; Based on the time alignment algorithm, the time information of all spatially registered target resolution fishery data is synchronized to obtain the target resolution fishery data after spatial registration and time information synchronization, which is calibrated as spatiotemporal synchronized fishery data.

[0027] It should be noted that the fusion of multi-source data in the same spatiotemporal framework is achieved through the alignment technology of time and space, with the aim of ensuring the consistency of data calculation of multi-source data and improving the calculation accuracy. Therefore, the spatiotemporal alignment algorithm is introduced to integrate fishery data from different sources. The grid-based alignment algorithm spatially aligns data from different sources so that they have a unified spatial reference system, while the sliding window algorithm can synchronize the fishery data collected at different times. Finally, after the spatiotemporal alignment, the spatiotemporal synchronized fishery data can be obtained.

[0028] like Figure 3 As shown, the second aspect of the present invention further provides a fishery data fusion system for data of different resolutions, the fishery data fusion system comprising a memory 31 and a processor 32, the memory 31 stores a fishery data fusion method, and when the fishery data fusion method is executed by the processor 32, the following steps are implemented: Conduct source analysis and resolution analysis of climate data, environmental data, and acoustic data in the fishery operation area to achieve classification and processing of different data in the fishery operation area; Clean the climate data, environmental data and acoustic data in the fishery operation area, and use interpolation technology and time series resampling algorithms to unify the resolution of various data in the fishery operation area; Based on the quality grade classification rules, weights are assigned to fishery data of various target resolutions, and combined with the spatiotemporal alignment algorithm, all target resolution fishery data are spatially aligned and time synchronized to obtain spatiotemporal synchronized fishery data. Based on the Bayesian inference network, multi-source fusion of all spatiotemporal synchronized fishery data is carried out, and feature extraction of the multi-source fused spatiotemporal synchronized fishery data is carried out based on the deep learning model to construct a fish distribution prediction model.

[0029] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A fishery data fusion method for data with different resolutions, characterized in that: The following steps are involved: Conduct source analysis and resolution analysis of climate data, environmental data, and acoustic data in the fishery operation area to achieve classification and processing of different data in the fishery operation area; Clean the climate data, environmental data and acoustic data in the fishery operation area, and use interpolation technology and time series resampling algorithms to unify the resolution of various data in the fishery operation area; Based on the quality grade classification rules, weights are assigned to fishery data of various target resolutions, and combined with the spatiotemporal alignment algorithm, all target resolution fishery data are spatially aligned and time synchronized to obtain spatiotemporal synchronized fishery data. Based on the Bayesian inference network, multi-source fusion of all spatiotemporal synchronized fishery data is carried out, and feature extraction of the multi-source fused spatiotemporal synchronized fishery data is carried out based on the deep learning model to construct a fish distribution prediction model.

2. A fishery data fusion method for data with different resolutions according to claim 1, characterized in that: The source analysis and resolution analysis of climate data, environmental data and acoustic data in the fishery operation area are performed to achieve classification processing of different data in the fishery operation area, specifically: Determine the fishery operation area and obtain the environmental data monitoring platform connected to the fishery operation area, and mark it as the fishery environmental data monitoring platform; The fishery environment data monitoring platform records the climate data and environmental data in the fishery operation area, wherein the climate data includes the wind speed and sea surface temperature in the fishery operation area, and the environmental data includes the salinity content and dissolved oxygen content in the fishery operation area; In the fishery environmental data monitoring platform, the data sources of climate data and environmental data in the fishery operation area are marked and named as fishery climate data source and fishery environmental data source. At the same time, the acoustic data of the fishery operation area is determined in the big data network, and the data source of the acoustic data is marked and named as fishery acoustic data source. For the acoustic data of the fishery operation area, audio editing software is obtained, and the acoustic data of the fishery operation area is converted into an audio file through the audio editing software, and marked as a fishery acoustic audio file, and the fishery acoustic audio file is parsed in the audio editing software to generate the sampling rate and bit depth information of the acoustic data of the fishery operation area; The sampling rate describes the time resolution of the acoustic data of the fishery operation area, and the bit depth information describes the amplitude resolution of the acoustic data of the fishery operation area. Based on the sampling rate and the bit depth information of the acoustic data of the fishery operation area, the acoustic resolution of the acoustic data of the fishery operation area is generated and calibrated as the first resolution. For the climate data and environmental data of the fishery operation area, the resolutions of the climate data and environmental data of the fishery operation area are determined through the corresponding fishery climate data sources and fishery environmental data sources, and calibrated as the second resolution and the third resolution respectively.

3. The fishery data fusion method for data with different resolutions according to claim 1, characterized in that: The climate data, environmental data and acoustic data in the fishery operation area are cleaned, and the resolution of various data in the fishery operation area is unified through interpolation technology and time series resampling algorithm, specifically: The climate data, environmental data and acoustic data in the fishery operation area are collectively referred to as fishery data, wherein the fishery data includes the data source and data resolution; The Min-max standardization algorithm is introduced to clean the fishery data, and the fishery data after data cleaning is obtained and calibrated as fishery cleaned data; Among them, the steps of the Min-max standardization algorithm for data cleaning of fishery data are: presetting a standard numerical interval, performing a linear transformation on the fishery data, and mapping the fishery data after the linear transformation within the standard numerical interval to achieve data cleaning processing; Performing resolution analysis on the first resolution, the second resolution, and the third resolution, and determining the highest resolution, which is calibrated as the target resolution; All fishery cleaned data with resolutions not equal to the target resolution are calibrated as fishery data with resolutions to be adjusted, and fishery clear data at the target resolution are calibrated as fishery data at the target resolution; Introducing a KNN interpolation algorithm, based on the target resolution fishery data, determining the K value in the KNN interpolation algorithm, calibrating it as the target K value, and calculating the Manhattan distance between different data points in the fishery data with a resolution to be adjusted and the target K value, and performing a weighted average calculation on different data points in the fishery data with a resolution to be adjusted according to the Manhattan distance, so that the fishery data with a resolution to be adjusted generates new data points, and the resolution of the fishery data with a resolution to be adjusted is achieved for the target resolution, thereby obtaining the fishery data with a target resolution; The target resolution fishery data, i.e., the resolution of climate data, environmental data and acoustic data in the fishery operation area, are all equal to the target resolution.

4. The fishery data fusion method for data with different resolutions according to claim 1, characterized in that: Based on the quality grade classification rule, various target resolution fishery data are weighted, and combined with the spatiotemporal alignment algorithm, all target resolution fishery data are spatially aligned and time synchronized to obtain spatiotemporal synchronized fishery data, specifically: Constructing data integrity thresholds and conducting data integrity analysis on target resolution fishery data, and combining the data integrity thresholds, marking the target resolution fishery data that is maintained within the data integrity threshold as the first-level data quality level, and marking the target resolution fishery data that is not maintained within the data integrity threshold as the second-level data quality level; Constructing a quality grade classification rule, wherein the quality grade classification rule is constructed according to the data integrity of the target resolution fishery data, and the data acquisition accuracy of all the target resolution fishery data is calculated, and the quality grade classification rule is optimized in combination with the data acquisition accuracy of all the target resolution fishery data to obtain an optimized quality grade classification rule; Wherein, the data sampling accuracy of the target resolution fishery data of the first-level data quality level and the second-level data quality level is marked in the optimized quality level classification rule; In the optimized quality grade classification rules, data weights are allocated to all target resolution fishery data according to the data quality level, so that target resolution fishery data of different quality levels are marked with weight factors; A spatiotemporal alignment algorithm is introduced to perform spatial registration and temporal synchronization on the target resolution fishery data and the corresponding marked weight factors to obtain spatiotemporal synchronized fishery data.

5. A fishery data fusion method for data with different resolutions according to claim 4, characterized in that: The temporal and spatial alignment algorithm is introduced to perform spatial registration and temporal synchronization on the target resolution fishery data and the corresponding marked weight factors to obtain temporal and spatial synchronized fishery data, specifically: Introducing a spatiotemporal alignment algorithm, wherein the spatiotemporal alignment algorithm includes a spatial alignment algorithm and a temporal alignment algorithm, wherein the spatial alignment algorithm is a grid-based alignment algorithm, and the temporal alignment algorithm is a sliding window algorithm; A spatial alignment grid is preset, wherein the spatial alignment grid is a spatial coordinate system, and all target resolution fishery data are imported into the spatial alignment grid, so that the target resolution fishery data are aligned in the spatial alignment grid to obtain spatially aligned target resolution fishery data; Wherein, registering the target resolution fishery data in the spatial alignment grid means controlling the spatial information of the target resolution fishery data to be equal, and the spatial information of the target resolution fishery data is the geographical location coordinates of the target resolution fishery data; For all spatially registered target resolution fishery data, extracting time information of the spatially registered target resolution fishery data, wherein the time information of the spatially registered target resolution fishery data is the collection time of fishery data with different target resolutions; Based on the time alignment algorithm, the time information of all spatially registered target resolution fishery data is synchronized to obtain the target resolution fishery data after spatial registration and time information synchronization, which is calibrated as spatiotemporal synchronized fishery data.

6. The fishery data fusion method for data with different resolutions according to claim 1, characterized in that: Based on the Bayesian inference network, all spatiotemporal synchronized fishery data are multi-source fused, and the spatiotemporal synchronized fishery data after multi-source fusion are feature extracted based on the deep learning model, so as to construct a fish distribution prediction model, specifically: Introduce the Bayesian inference network, and import all the spatiotemporal synchronized fishery data into the Bayesian inference network, wherein all the spatiotemporal synchronized fishery data need to be imported according to the conditional probability model; Based on the Bayesian inference network, a maximum a posteriori probability estimation is performed on all spatiotemporal synchronized fishery data, and conflicting fishery data and normal fishery data are generated after the maximum a posteriori probability estimation, wherein the conflicting fishery data are spatiotemporal synchronized fishery data whose maximum a posteriori probability is not within a preset range, and the normal fishery data are spatiotemporal synchronized fishery data whose maximum a posteriori probability is within a preset range; Performing weighted average processing on the conflicting fishery data until the conflicting fishery data is converted into normal fishery data, outputting the normal fishery data and performing multi-source fusion of the data to obtain spatiotemporal synchronized fishery data after multi-source fusion processing, and marking it as multi-source fused spatiotemporal synchronized fishery data; The CNN convolutional neural network is introduced, and data convolution analysis is performed on the multi-source fused spatiotemporal synchronized fishery data in the convolution layer of the CNN convolutional neural network. After the data convolution analysis, the time characteristics and spatial characteristics of the multi-source fused spatiotemporal synchronized fishery data are extracted in the output layer and calibrated as target time characteristics and target spatial characteristics. The principal component analysis method is used to eliminate the redundancy of target time features and target space features to obtain redundant processing time features and redundant processing space features. Based on the CNN convolutional neural network, a convolutional neural network model that can predict fish distribution is constructed by combining the redundant processing time features and the redundant processing space features. It is calibrated as a fish distribution prediction model.

7. A fishery data fusion system for data of different resolutions, characterized in that: The fishery data fusion system includes a memory and a processor. The memory stores a fishery data fusion method program. When the fishery data fusion method program is executed by the processor, the fishery data fusion method steps as described in any one of claims 1 to 6 are implemented.

Images