A Smart Dynamic Evaluation Method and System for Water Ecological Health Based on Multi-Source Data Fusion
By fusing multi-source data and analyzing the spatiotemporal relationships of multi-dimensional monitoring data of water bodies, multi-dimensional health assessment data is constructed, which solves the problem that the spatiotemporal relationships of water body ecological health assessment have not been considered in existing technologies, and realizes a dynamic and comprehensive assessment of water body ecological health.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- YANGTZE RIVER FISHERIES RES INST CHINESE ACAD OF FISHERY SCI
- Filing Date
- 2026-02-14
- Publication Date
- 2026-05-26
Smart Images
Figure CN121706038B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, specifically to a method and system for intelligent dynamic evaluation of water ecological health based on multi-source data fusion. Background Technology
[0002] Multi-source data for aquatic ecological health assessment involves multi-faceted and multi-temporal data monitoring, including satellite data, ground sensor monitoring data, and underwater monitoring data. Simultaneously, multi-dimensional data monitoring is conducted at several monitoring points in the same water area. However, aquatic ecological health assessment usually involves multi-faceted data analysis. Single data sources are limited by time and space, resulting in limited monitoring scope and low timeliness, and cannot fully reflect the health performance of aquatic ecosystems. Multi-source data fusion has become a key development trend for intelligent dynamic assessment of aquatic ecological health.
[0003] Because multi-source data involves many data dimensions and large data volume, existing data fusion methods directly fuse raw monitoring data when the data volume is low, and use principal component analysis for dimensionality reduction fusion when the data dimensions are large, while retaining key information. However, neither of these methods can well adapt to multi-source data for aquatic ecological health monitoring. The relationship characteristics exhibited by the spatiotemporal correlation of multi-source data design cannot be reflected by simply fusing the features of data from various dimensions. At the same time, water flow and water body renewal mean that aquatic ecological health needs to be dynamically evaluated. Therefore, it is necessary to supplement and fuse multi-source data at the spatiotemporal level, so as to realize the spatiotemporal analysis and dynamic evaluation of aquatic ecological health under multi-source data fusion. Summary of the Invention
[0004] This invention provides a method and system for intelligent dynamic evaluation of water ecological health based on multi-source data fusion, to solve the problem that existing dynamic evaluation processes for water ecological health do not consider the spatiotemporal relationship influence of multi-source data correlation. The specific technical solution adopted is as follows:
[0005] This invention proposes a smart dynamic evaluation method for water ecological health based on multi-source data fusion, which includes the following steps:
[0006] Collect monitoring data from multiple monitoring locations and multiple sources in the water area, and obtain historical monitoring data from multiple sources and multiple dimensions for several water areas;
[0007] Based on the time-series change trend of monitoring data for a single evaluation indicator, and combined with the overall change of monitoring data from multiple sources and multiple evaluation indicators, several monitoring periods for the water area are obtained. By removing anomalies, the indicator data for each evaluation indicator in each monitoring period are obtained. By combining historical monitoring data, the overall change trend of the same evaluation indicator under different monitoring periods is analyzed, and the scenario reference coefficients of the same evaluation indicator in each historical monitoring period are obtained for the current monitoring period. Based on the monitoring location distribution corresponding to the monitoring data of the same evaluation indicator and the historical monitoring data, and combined with the scenario reference coefficients, the spatiotemporal reference factors of each historical monitoring data are obtained.
[0008] The correlation between the change data of different evaluation indicators and historical monitoring data is analyzed to obtain the correlation of changes in different evaluation indicators. The feedback adjustment coefficient of each evaluation indicator is obtained by combining the correlation of changes in all evaluation indicators. Simulation data of each evaluation indicator in the current water area is obtained by Gaussian distribution. Similarity analysis is performed between the simulation data and the spatiotemporal reference factor of historical monitoring data, and the data is adjusted by feedback adjustment coefficient. Finally, multi-source and multi-dimensional health evaluation data of the current water area are obtained.
[0009] Based on the multi-source, multi-dimensional health assessment data of the current water area, an indicator-time data matrix is constructed for each monitoring location, and a collaborative filtering algorithm is used to achieve dynamic health assessment of the current water area.
[0010] Optionally, the specific method for obtaining the monitoring periods of the water area is as follows:
[0011] The monitoring data of any monitoring location for any evaluation indicator in the current water area and several historical monitoring data together constitute the monitoring sequence of the evaluation indicator in the current water area. The monitoring sequence is decomposed by STL time series to obtain the trend term, periodic term and residual term. The periodic term is Fourier transformed to obtain the spectrum of the periodic term of the monitoring sequence. The amplitude of each frequency in the spectrum is obtained. The reciprocal of the frequency corresponding to the maximum amplitude is taken as the monitoring period of the evaluation indicator at that monitoring location.
[0012] Obtain the monitoring period for each evaluation indicator of the current water area at the current monitoring location, obtain the least common multiple of the monitoring location for all monitoring periods, and take the average of the least common multiples corresponding to all monitoring locations as a monitoring period for the current water area.
[0013] Optionally, the specific methods for obtaining the indicator data for each evaluation indicator in each monitoring period include:
[0014] For any evaluation indicator of the current water area and any monitoring data for any monitoring period, based on the normal distribution and The criterion is to obtain the mean and standard deviation of all monitoring data during the monitoring period, remove outliers from the monitoring data during the monitoring period, and use the remaining monitoring data after screening as the screened data for the monitoring period.
[0015] Obtain the screening data for each monitoring period of the current water area for this evaluation indicator, as well as the screening data for each monitoring period from all historical monitoring data. Standardize all the screening data for this evaluation indicator, and use the results as the indicator data for each monitoring period of this evaluation indicator.
[0016] Optionally, the specific method for obtaining the scenario reference coefficients of the same evaluation indicators for each historical monitoring period to the current monitoring period is as follows:
[0017] For any evaluation indicator in the current water area, several indicator data for the current monitoring period are collected and arranged in chronological order to form an indicator sequence for the current monitoring period of the current evaluation indicator in the current water area; several indicator data for any evaluation indicator in any historical water area are collected and arranged in chronological order to form a historical indicator sequence for the evaluation indicator in that monitoring period of the current water area; DTW matching is performed between the indicator sequence and the historical indicator sequence to obtain the DTW distance, and the inversely proportionally normalized result of the obtained DTW distance is used as a reference factor for the change of the evaluation indicator in that monitoring period of the current water area to the current monitoring period of the current water area;
[0018] Based on the trend changes of current and historical water monitoring data under the same evaluation indicators, and the time distribution of the corresponding monitoring period, time reference weights are obtained.
[0019] The product of the time reference weight and the change reference factor is used as the scenario reference coefficient for the current monitoring period of the current monitoring period of the current water area and the evaluation index.
[0020] Optionally, the method for obtaining the time reference weight based on the trend changes of current and historical water monitoring data under the same evaluation index, and the time distribution of the corresponding monitoring period, includes the following specific methods:
[0021] Based on the current monitoring period of any evaluation indicator in the current water area, extract the trend sequence of the current monitoring period of the evaluation indicator and the trend sequence of the adjacent previous monitoring period from the trend item of the monitoring sequence of the evaluation indicator. Obtain the difference sequence obtained by subtracting the trend sequence of the adjacent previous monitoring period from the trend sequence of the current monitoring period, and use it as the trend change sequence of the current monitoring period.
[0022] Obtain any monitoring period of the evaluation index for any water body and use it as a reference monitoring period for the evaluation index for a reference water body; obtain the trend sequence of the reference monitoring period and the corresponding trend change sequence; obtain the cosine similarity between the trend change sequence of the current monitoring period and the trend change sequence of the evaluation index for the reference water body during the reference monitoring period.
[0023] Calculate the time distance between the first moment in the current monitoring period and the first moment in the reference monitoring period. Divide the time distance by the length of one year and use the decimal part of the quotient as the time difference factor between the current monitoring period and the reference monitoring period. Subtract the time difference factor from 1 and multiply it by the cosine similarity as the time reference weight of the reference monitoring period for the current monitoring period of the current water body.
[0024] Optionally, the spatiotemporal reference factors for each historical monitoring data are obtained using the following method:
[0025] For any evaluation indicator in the current water area, obtain the local monitoring sequence of any two monitoring locations during the current monitoring period, and obtain the cosine similarity of the two local monitoring sequences as the local similarity of the two monitoring locations for the current evaluation indicator in the current monitoring period. For the historical monitoring data of the current water area, obtain the corresponding local similarity of the two monitoring locations for each evaluation indicator in each monitoring period, and take the mean of all local similarities as the monitoring similarity of the two monitoring locations for the current water area.
[0026] For two monitoring locations in different water areas, the local similarity is obtained from the local monitoring sequences of the same evaluation index and the same monitoring period. Then, the monitoring similarity of the two monitoring locations in different water areas is obtained by the mean of the local similarity of the two monitoring locations under the same monitoring period for all evaluation indexes.
[0027] Obtain the actual distance between any two monitoring locations, and then obtain the water system basins to which the two monitoring locations belong. If the two monitoring locations belong to the same water system basin, obtain the ratio of the actual distance between the two monitoring locations to the average span of the water system basin. Subtract the ratio from 1, add 1 to the difference, and use the sum as the spatial reference factor for the two monitoring locations. If the two monitoring locations belong to different water system basins, set the spatial reference factor for the two monitoring locations to 1.
[0028] The product of the spatial reference factor of the two monitoring locations and the monitoring similarity is linearly normalized to obtain the spatial similarity factor of the two monitoring locations. Then, for any evaluation index of the current water area and the current monitoring location in the current monitoring period, the spatial similarity factor of the two monitoring locations is multiplied by the scenario reference coefficient of the two locations to obtain the spatiotemporal reference factor of the current monitoring location of the current evaluation index in the current monitoring period for the current water area.
[0029] The spatiotemporal reference factor is assigned to the local monitoring sequence of the monitoring location for the evaluation index in the water area during the monitoring period, and serves as the spatiotemporal reference factor for each historical monitoring data point for the current monitoring location of the evaluation index in the current water area during the current monitoring period.
[0030] Optionally, the specific method for obtaining the correlation of changes in different evaluation indicators and synthesizing the correlation of changes in all evaluation indicators to obtain the feedback adjustment coefficient of each evaluation indicator includes:
[0031] For any two evaluation indicators in the current water area during the same monitoring period, calculate the DTW distance between the two indicator sequences and perform inverse proportional normalization. The result is used as the correlation of changes in the two evaluation indicators in the current water area during the same monitoring period. Calculate the mean of the correlation of changes in the two evaluation indicators in all water areas during the same monitoring period, and use it as the correlation of changes in the two evaluation indicators.
[0032] Using any evaluation indicator as the current evaluation indicator, obtain the monitoring period corresponding to the current evaluation indicator at any monitoring location in the current water area. The ratio of this monitoring period to the least common multiple of all monitoring periods corresponding to this monitoring location is used as the period factor of the current evaluation indicator at this monitoring location in the current water area. Obtain the period factor of any other evaluation indicator besides the current evaluation indicator at this monitoring location in the current water area. The absolute value of the difference between the two period factors is used as the period difference factor between the current evaluation indicator and the other evaluation indicator at this monitoring location in the current water area. Obtain the mean of the period difference factors between the current evaluation indicator and the other evaluation indicator at all monitoring locations in the current water area. Subtract the mean from 1 to obtain the difference, which is used as the period reference weight of the other evaluation indicator relative to the current evaluation indicator.
[0033] Based on the aforementioned periodic reference weights, the correlation between the changes of all evaluation indicators other than the current evaluation indicator and the current evaluation indicator is weighted and averaged, and the result is used as the feedback adjustment coefficient of the current evaluation indicator.
[0034] Optionally, the specific methods for obtaining the multi-source, multi-dimensional health assessment data of the current water area include:
[0035] For any simulated data, the corresponding indicator data of the simulated data at the same location in each monitoring period of the current evaluation indicator of each water area in the historical monitoring data are weighted and averaged with the corresponding spatiotemporal reference factor. The result is used as the reference data of the simulated data.
[0036] The absolute value of the difference between the simulated data and its reference data is obtained as the initial simulated difference. The initial simulated differences of each simulated data in the current monitoring period are obtained, and a difference threshold is preset. If there are simulated data with an initial simulated difference greater than the difference threshold, the simulated data is multiplied by the feedback adjustment coefficient, and the simulated difference is recalculated until there are no simulated data with a difference greater than the difference threshold. The final simulated data is used as the final simulated data of the current evaluation indicator in the current monitoring period of the current water area. Combined with the indicator data, the health data of the current evaluation indicator in the current monitoring period of the current water area is constituted. The health data of all evaluation indicators in the current water area in all monitoring periods constitute the multi-source and multi-dimensional health evaluation data of the current water area.
[0037] Optionally, the specific methods for constructing the indicator-time data matrix for each monitoring location and using a collaborative filtering algorithm to achieve dynamic health assessment of the current water area include:
[0038] For any monitoring location in the current water area, the health data of each evaluation indicator at each time point is obtained from the multi-source multi-dimensional health evaluation data. A matrix is constructed with the evaluation indicators as columns and time as rows, and the health data is filled into the matrix to obtain the indicator-time data matrix of the monitoring location.
[0039] Data fusion is performed on the indicator-time data matrix of the monitoring location to obtain the fusion results of each monitoring location in the current water area, and prediction is made based on the indicator-time data matrix; a comprehensive health score is then calculated based on the indicator-time data matrix.
[0040] After obtaining the fusion results and health scores, the fusion results and comprehensive health scores of each monitoring location in the current water area are analyzed and recommended based on the collaborative filtering algorithm. The algorithm output results serve as a predictive health assessment of the current water area.
[0041] This invention also proposes an intelligent dynamic evaluation system for water ecological health based on multi-source data fusion. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the above method.
[0042] The beneficial effects of this invention are as follows: This invention involves deploying monitoring units at multiple locations in a water area to collect monitoring data on multiple ecological health evaluation indicators over a long period. It also retrieves a large amount of historical monitoring data from the current water area and other water areas. By analyzing the overall periodic changes of the same evaluation indicators, the invention first divides the monitoring period to reflect the overall periodic changes in water health data. Based on the monitoring period, it analyzes the temporal similarity reference relationship between monitoring data, examining both the data itself and the overall trend changes in different monitoring periods. Then, combining the differences in data changes at different monitoring locations under the same evaluation indicators, it conducts a spatial similarity analysis to obtain historical monitoring data. Based on the spatiotemporal reference factors, a foundation is provided for subsequent simulation data interpolation, supplementation, and adjustment. Further consideration is given to the correlation of data changes among different evaluation indicators of the water area, providing a basis for iterative correction of the simulation data, thus obtaining feedback adjustment coefficients. Interpolation is performed using Gaussian distribution to initially obtain simulation data. Then, combined with historical monitoring data and spatiotemporal reference factors, the data is continuously adjusted using feedback adjustment coefficients, ultimately achieving multi-dimensional data alignment. Multi-source, multi-dimensional health evaluation data is then obtained, and a data matrix is constructed based on this. Health scoring is performed using the data matrix from multiple monitoring locations, and recommendations are implemented using collaborative filtering algorithms, thereby achieving a comprehensive dynamic health evaluation of the current water area. Attached Figure Description
[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0044] Figure 1 This is a schematic diagram of the process of a smart dynamic evaluation method for water ecological health based on multi-source data fusion, provided in one embodiment of the present invention. Detailed Implementation
[0045] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0046] Please see Figure 1 The diagram illustrates a flowchart of a method for intelligent dynamic evaluation of water ecological health based on multi-source data fusion, according to an embodiment of the present invention. The method includes the following steps:
[0047] Step S001: Collect monitoring data of multiple evaluation indicators at multiple monitoring locations in the water area, and obtain historical monitoring data of multiple sources and dimensions for several water areas.
[0048] The purpose of this embodiment is to dynamically evaluate the ecological health of the water area by deploying monitoring units at multiple locations within the water area, including multi-source data monitoring from the air, ground, and underwater, and by acquiring monitoring data from multiple evaluation indicators. At the same time, to facilitate the comparison of monitoring data, a large amount of historical monitoring data is retrieved from the database for subsequent analysis.
[0049] Specifically, the water area to be monitored is designated as the current water area. Multiple data acquisition devices are deployed at various monitoring locations within this water area. These devices include, but are not limited to, underwater robots, buoy arrays, eDNA sampling equipment, and various physical and chemical sensors for measuring dissolved oxygen, nitrite content, ammonia nitrogen content, and pH. The collected evaluation indicators, in addition to the aforementioned physical and chemical indicators, also include fish distribution (the number of fish at the corresponding monitoring location), fish species diversity, plankton content, and the accumulation of heavy metals and harmful substances in fish. The sampling frequency is set to once every half day. Long-term monitoring of the current water area is conducted for at least one year, yielding monitoring data for several evaluation indicators at several monitoring locations. Simultaneously, multi-source... The data also includes aerial and ground monitoring. Water spectra and temperature are obtained through satellite remote sensing (such as Sentinel-2), and water density is sampled through drone patrols. The sampling frequencies are the same, and drone patrols correspond to monitoring locations. Satellite remote sensing is not affected by the monitoring location, and the collected monitoring data directly corresponds to each monitoring location. Ground monitoring combines video surveillance with underwater monitoring and biological sampling points to collect data on various evaluation indicators related to fish. In other words, the above-mentioned fish evaluation indicators need to be obtained in conjunction with ground monitoring. At the same time, monitoring data of the same evaluation indicators for a large number of water areas are retrieved in the database, and the corresponding monitoring locations are recorded as historical monitoring data.
[0050] It should be noted that the scenarios of water bodies vary based on dimensions such as hydrology, meteorology, and human activities, including: hydrological scenarios (flood season / dry season / normal water season), meteorological scenarios (heavy rain / high temperature / normal), human activity scenarios (irrigation season / peak shipping / low disturbance), ecological status scenarios (algal bloom / peak season of aquatic plant growth), and spatial scenarios (upstream / middle reaches / downstream / estuary).
[0051] Step S002: Based on the time-series change trend of monitoring data of a single evaluation indicator, and combined with the overall change of monitoring data of multiple sources and multiple evaluation indicators, several monitoring periods of the water area are obtained. Through anomaly removal, the indicator data of each evaluation indicator in each monitoring period are obtained. Combined with historical monitoring data, the overall change trend of the same evaluation indicator under different monitoring periods is analyzed, and the scenario reference coefficient of the same evaluation indicator in each historical monitoring period is obtained for the current monitoring period. Based on the monitoring data of the same evaluation indicator and the monitoring location distribution corresponding to the historical monitoring data, and combined with the scenario reference coefficient, the spatiotemporal reference factor of each historical monitoring data is obtained.
[0052] It should be noted that the current monitoring data of multiple evaluation indicators in the water area, as well as the historical monitoring data, are big data collected over a long period of time from the current water area and multiple water areas. Water health-related data usually exhibits periodic changes with seasonal, hydrological, and meteorological scenarios, with seasonality and hydrology being the main factors. Therefore, trend terms, periodic terms, and residual terms are obtained through time series decomposition to eliminate the influence of trend changes and occasional data fluctuations. The monitoring period of each evaluation indicator is obtained based on the periodic term, thereby obtaining the monitoring period. At the same time, abnormal data is eliminated, and the remaining data is standardized to meet the prerequisite of eliminating dimensions in subsequent multi-source data fusion.
[0053] Preferably, in one embodiment of the present invention, based on the time-series change trend of monitoring data of a single evaluation indicator, combined with the overall change of monitoring data of multiple sources and multiple evaluation indicators, several monitoring periods of the water area are obtained, and the indicator data of each evaluation indicator for each monitoring period are obtained through anomaly removal. The specific method includes:
[0054] For any evaluation indicator at any monitoring location in the current water area, the monitoring data and several historical monitoring data together constitute the monitoring sequence of that evaluation indicator in the current water area. STL time series decomposition is performed on this monitoring sequence to obtain a trend term, a periodic term, and a residual term. A Fourier transform is performed on the periodic term to obtain the spectrum of the periodic term of the monitoring sequence. The amplitude of each frequency in the spectrum is obtained, and the reciprocal of the frequency corresponding to the maximum amplitude is taken as the monitoring period of that monitoring location for that evaluation indicator. STL time series decomposition and Fourier transform are well-known techniques and will not be elaborated upon in this embodiment. The current water area is obtained according to the above method. For each evaluation indicator of a multi-source water body, the monitoring period at the monitoring location is determined. The least common multiple of the monitoring locations for all monitoring periods is obtained, and the average of the least common multiples corresponding to all monitoring locations is taken as a monitoring period for the current water body. It is worth noting that if historical monitoring data or monitoring data is insufficient to constitute a monitoring period (i.e., the data volume and time length are inadequate), but most of the data is present in the current monitoring period, it will be directly used as a monitoring period in subsequent calculations. At the same time, the monitoring periods of different water bodies in the historical monitoring data are all processed based on the monitoring period of the current water body to reduce the impact of different monitoring periods.
[0055] It should be noted that the monitoring data of a single evaluation indicator in the water area are affected by changes in seasonality, environment, and temperature, and exhibit periodic variations. These variations can range from days and months to years. Furthermore, the trends of monitoring data across different periods tend to be similar. Therefore, trend terms, periodic terms, and residual terms are obtained through STL decomposition. Based on the periodic term, the initial period of a single evaluation indicator is extracted. Then, considering the comprehensive performance of the water area from multiple sources and multiple evaluation indicators, the least common multiple of multiple monitoring periods is used for subsequent monitoring and analysis.
[0056] Furthermore, for any evaluation indicator of the current water area and any monitoring data (or historical monitoring data) for any monitoring period, based on the normal distribution and The criterion is to obtain the mean and standard deviation of all monitoring data for that monitoring period, and then remove outliers from the data, i.e., data that deviates from a normal distribution. Outliers in the monitoring data within the specified range need to be removed, and the remaining monitoring data after screening are used as the screening data for that monitoring period. The screening data for each monitoring period of the current water area for this evaluation indicator are obtained according to the above method, as well as the screening data for each monitoring period from all historical monitoring data (not necessarily all from the current water area). All screening data for this evaluation indicator are standardized (linearly normalized), and the results are used as the indicator data for each monitoring period of this evaluation indicator. The indicator data for all evaluation indicators and historical monitoring data are obtained according to the above method.
[0057] It should be further explained that after obtaining the indicator data, it is necessary to conduct a similarity analysis on the data change trends of the same evaluation indicator in different monitoring periods, that is, a similarity reference analysis in the time dimension. First, a similarity analysis is performed on the sequence of indicator data, and then the differences in trend changes are considered. On the basis of similarity, the similar performance between the local trend items corresponding to different monitoring periods is introduced. That is, the changes in indicator data in different monitoring periods are also affected by the overall trend. At the same time, a time distribution difference analysis is performed on an annual basis. The smaller the time difference within a year, the greater the reference value.
[0058] Preferably, in one embodiment of the present invention, the method of analyzing the overall changing trend of the same evaluation index under different monitoring periods by combining historical monitoring data to obtain the scenario reference coefficient of the same evaluation index in each historical monitoring period for the current monitoring period includes:
[0059] For any evaluation indicator in the current water area, several indicator data points for the current monitoring period are used to construct an indicator sequence for the current monitoring period of that evaluation indicator in the current water area, arranged chronologically. Several indicator data points for any historical evaluation indicator in any historical water area are obtained for any monitoring period, and a historical indicator sequence for that evaluation indicator in that monitoring period is constructed chronologically. DTW matching is performed between the indicator sequence and the historical indicator sequence to obtain the DTW distance. The inversely proportionally normalized result of the obtained DTW distance is used as a reference factor for the change of that evaluation indicator in that monitoring period in that water area relative to the current monitoring period in the current water area. This embodiment adopts... The model is used to represent the inverse proportional relationship and for normalization processing. As input to the model, This represents an exponential function with the natural constant as the base. Implementers can set inverse proportional functions and normalization functions according to the actual situation.
[0060] It should be noted that, for the indicator data corresponding to the historical monitoring data, although the monitoring period is the same, the sequence length is not equal due to outlier screening. Therefore, similarity analysis is carried out by matching the indicator sequence with the historical indicator sequence using DTW, which is used as a preliminary reference factor for the changes of the current monitoring period in each monitoring period of each historical water area.
[0061] Furthermore, based on the current monitoring period of the evaluation indicator in the current water area, the trend sequence of the current monitoring period of the evaluation indicator and the trend sequence of the adjacent previous monitoring period are extracted from the trend term of the monitoring sequence of the evaluation indicator (the trend sequence is the local trend term corresponding to the monitoring period). The difference sequence obtained by subtracting the trend sequence of the adjacent previous monitoring period from the trend sequence of the current monitoring period is used as the trend change sequence of the current monitoring period. It should be noted that if the trend terms of different monitoring periods are of equal length, the trend terms with the same order value are subtracted to obtain the difference. If the current monitoring period is the most recent monitoring period, resulting in insufficient trend sequence length, the trend sequence of the current monitoring period is supplemented to be of equal length with other trend sequences by using data prediction through an LSTM model. The above method is used to obtain any monitoring period of the evaluation indicator for any water area and use it as the reference monitoring period of the evaluation indicator for the reference water area. The process involves: 1) obtaining a trend sequence and corresponding trend change sequence for the reference monitoring period; 2) calculating a cosine similarity between the trend change sequence of the current monitoring period and the trend change sequence of the reference monitoring period for the same evaluation index in the reference water area; 3) calculating the time distance between the first moment in the current monitoring period and the first moment in the reference monitoring period, and using the fractional part of the quotient obtained by dividing the time distance by the length of one year (with the same unit for both time length and time distance) as the time difference factor between the current monitoring period and the reference monitoring period; 4) multiplying the difference obtained by subtracting the time difference factor from 1 with the cosine similarity result as the time reference weight of the reference monitoring period for the same evaluation index in the reference water area relative to the current monitoring period in the current water area; and 5) multiplying the time reference weight by the change reference factor as the scenario reference coefficient of the same evaluation index for the same monitoring period in the current water area relative to the current monitoring period in the current water area.
[0062] It should be noted that, based on the similarity analysis of the changes in the indicator sequence by the change reference factor, the influence of the trend changes of the same evaluation indicator in different periods and seasons is considered. The trend changes of adjacent monitoring periods are analyzed. The more similar the trend change sequences of the current monitoring period and the historical monitoring period are, and the smaller the spatiotemporal distance in years, the greater the time reference between the two monitoring periods. The change reference factor is then corrected to obtain the scenario reference coefficient.
[0063] It should be further noted that the scenario reference coefficient is obtained by analyzing the changes in the monitoring data and indicator data from the data itself and the trend changes during the monitoring period. Its corresponding time dimension data performance needs to be further combined with the differences in the distribution of monitoring locations corresponding to different monitoring data to conduct similarity analysis from the spatial dimension.
[0064] Preferably, in one embodiment of the present invention, the spatiotemporal reference factor for each historical monitoring data is obtained based on the monitoring location distribution corresponding to the index data and historical monitoring data of the same evaluation index, combined with the scenario reference coefficient. The specific method includes:
[0065] It should be noted that the obtained scenario reference coefficients are based on any monitoring location of any evaluation indicator in the current monitoring period of any water area, and any monitoring location of any evaluation indicator in any water area during any monitoring period (historical monitoring period). The scenario reference coefficients are obtained based on monitoring data and indicator data and are not limited by monitoring location. The spatiotemporal reference factors need to take into account the differences in the distribution of monitoring locations for analysis.
[0066] Specifically, for any two monitoring locations of any evaluation indicator in the current water area during the current monitoring period, local monitoring sequences are obtained (the monitoring sequences of the corresponding locations are truncated for the current monitoring period). The cosine similarity between the two local monitoring sequences is obtained as the local similarity between the two monitoring locations of the current evaluation indicator in the current water area during the current monitoring period. Following the above method, for the monitoring data of the current water area in the historical monitoring data, the corresponding local similarities for each evaluation indicator in each monitoring period for the two monitoring locations are obtained. The average of all local similarities is used as the monitoring similarity between the two monitoring locations in the current water area. For two monitoring locations in different water areas, the local similarity of the local monitoring sequences of the same evaluation indicator in the same monitoring period is obtained. Then, the monitoring similarity between the two monitoring locations in different water areas is obtained by using the average of the local similarities of the two monitoring locations in the same monitoring period for all evaluation indicators.
[0067] Furthermore, the actual distance between any two monitoring locations is obtained (calculated based on geospatial data). The river basins to which these two locations belong (upstream basins of the same river, etc.) are also obtained. If the two monitoring locations belong to the same river basin, the ratio of the actual distance between the two locations to the average span of the river basin (obtained directly from the river basin; existing data will not be elaborated upon in this embodiment) is calculated. The difference obtained by subtracting this ratio from 1, plus the sum obtained by adding 1, is used as the spatial reference factor for the two monitoring locations. If the two monitoring locations belong to different river basins, the spatial reference factor for the two monitoring locations is set to 1. The product of the spatial reference factor of the two monitoring locations and the monitoring similarity is then linearly normalized. The result obtained by unification is used as the spatial similarity factor of the two monitoring locations. Then, for any evaluation index of the current water area and the current monitoring location in the current monitoring period, and any monitoring location of the same evaluation index in any water area in any monitoring period, the spatial similarity factor corresponding to the two monitoring locations is multiplied based on the scenario reference coefficients of the two locations to obtain the spatiotemporal reference factor of the monitoring location of the same evaluation index in the current water area for the current monitoring location of the same evaluation index in the current monitoring period. The spatiotemporal reference factor is then assigned to the local monitoring sequence of the monitoring location of the same evaluation index in the current water area for the current monitoring location in the current monitoring period, as the spatiotemporal reference factor of each historical monitoring data for the current monitoring location of the same evaluation index in the current water area.
[0068] It should be noted that, firstly, the similarity of monitoring units at different monitoring locations under the same evaluation index during the same monitoring period is analyzed. The more similar the changes in monitoring data in the local monitoring sequences, the more similar the water body performance in the monitoring units of the two monitoring locations under the same evaluation index. Then, multiple monitoring periods and multiple evaluation indicators are combined to quantify the similarity of monitoring units. Next, the actual geographical distribution is analyzed, divided by river basins. For those in the same river basin, the spatial weight of actual distance is further quantified based on the basin span, while for different river basins, it is directly set to 1. That is, since different river basins are different, geographical differences are not considered. When set to 1, the spatial reference factor of the same river basin is greater than 1. Combining the monitoring similarity, the spatial similarity factor is obtained, and the scenario reference coefficient is further adjusted to obtain the spatiotemporal reference factor.
[0069] Thus, the spatiotemporal reference factors for each historical monitoring data are obtained.
[0070] Step S003: Analyze the correlation between the indicator data of different evaluation indicators and historical monitoring data to obtain the correlation of changes in different evaluation indicators. Combine the correlation of changes in all evaluation indicators to obtain the feedback adjustment coefficient of each evaluation indicator. Obtain simulated data for the indicator data of each evaluation indicator in the current water area in each monitoring period through Gaussian distribution. Perform similarity analysis with the simulated data based on the spatiotemporal reference factor of historical monitoring data and adjust it through the feedback adjustment coefficient to finally obtain the multi-source and multi-dimensional health evaluation data of the current water area.
[0071] It should be noted that there are often correlations between different evaluation indicators in aquatic health. For example, dissolved oxygen content in water is closely related to the distribution of fish in the water. The species content and bioaccumulation of fish also involve multiple evaluation indicators, such as nitrite, ammonia nitrogen and pH value. Therefore, it is necessary to conduct correlation analysis on the changes of different evaluation indicators to provide a basis for data alignment of each evaluation indicator before subsequent simulation data and final multi-source data fusion.
[0072] Preferably, in one embodiment of the present invention, the method of analyzing the correlation between changes in the indicator data and historical monitoring data of different evaluation indicators to obtain the correlation of changes in different evaluation indicators, and synthesizing the correlation of changes in all evaluation indicators to obtain the feedback adjustment coefficient of each evaluation indicator includes:
[0073] For any two evaluation indicators in the current water area during the same monitoring period, the DTW distance between the two indicator sequences is calculated and inversely normalized. The result is used as the correlation of changes in the two evaluation indicators in the current water area during the same monitoring period. This embodiment adopts... The model is used to represent the inverse proportional relationship and for normalization processing. As input to the model, It represents an exponential function with a natural constant as the base. Implementers can set an inverse proportional function and a normalization function according to the actual situation. The mean value of the changes in the two evaluation indicators for the same monitoring period in all water areas is used as the correlation of the changes in the two evaluation indicators.
[0074] Furthermore, taking any evaluation indicator as the current evaluation indicator, the monitoring period corresponding to the current evaluation indicator at any monitoring location in the current water area is obtained. The ratio of this monitoring period to the least common multiple of all monitoring periods corresponding to this monitoring location is used as the period factor of the current evaluation indicator at this monitoring location in the current water area. Similarly, the period factor of any other evaluation indicator besides the current evaluation indicator at this monitoring location in the current water area is obtained. The absolute value of the difference between the two period factors is used as the period difference factor between the current evaluation indicator and the other evaluation indicator at this monitoring location in the current water area. The mean of the period difference factors of the current evaluation indicator and the other evaluation indicator at all monitoring locations in the current water area is obtained. The difference obtained by subtracting the mean from 1 is used as the period reference weight of the other evaluation indicator on the current evaluation indicator. Based on the period reference weight, the weighted average of the changes in the correlation between all other evaluation indicators besides the current evaluation indicator and the current evaluation indicator is calculated, and the result is used as the feedback adjustment coefficient of the current evaluation indicator.
[0075] It should be noted that, firstly, the correlation between the evaluation indicators is analyzed through the indicator sequence to quantify the correlation of changes and reflect the impact of changes in other evaluation indicators on the evaluation indicators. Then, the difference in the monitoring period between the evaluation indicators is considered. The greater the difference in the monitoring period, the more the local trend affects the overall trend of another evaluation indicator, the smaller its overall reference value, and the smaller the reference weight of the corresponding period. Finally, the feedback adjustment coefficient of the current evaluation indicator is obtained by comprehensive analysis.
[0076] It should be further explained that the feedback adjustment coefficient reflects the correlation between changes in different evaluation indicators. For the current evaluation indicators of the water area, interpolation is required to obtain simulated data to ensure data alignment between different evaluation indicators, thereby enabling multi-source data fusion. Therefore, simulated data is first obtained through Gaussian distribution. Based on the spatiotemporal reference factor, other monitoring periods of historical monitoring data are used as references and weighted fusion is performed to obtain the corresponding reference data. By analyzing the differences between simulated data and reference data, the feedback adjustment coefficient is introduced for iterative adjustment. Under the condition that the difference between simulated data and reference data becomes smaller and smaller, simulated data is generated in combination with the influence of changes in different evaluation indicators, and finally, multi-source and multi-dimensional health evaluation data is obtained.
[0077] Preferably, in one embodiment of the present invention, simulated data of the current water body's various evaluation indicators for each monitoring period are obtained through Gaussian distribution. Similarity analysis is performed between the simulated data and the spatiotemporal reference factor of historical monitoring data, and adjustments are made using a feedback adjustment coefficient. Finally, multi-source, multi-dimensional health evaluation data of the current water body is obtained. The specific method includes:
[0078] For the current evaluation indicator data of the current water area during the current monitoring period, based on the time corresponding to the monitoring data of all other evaluation indicators, the current evaluation indicator during the current monitoring period is interpolated using a Gaussian distribution to obtain simulated data, ensuring that there are corresponding indicator data or simulated data for all other evaluation indicators at the corresponding time.
[0079] Furthermore, for any simulated data, the corresponding indicator data at the same location for each current evaluation indicator in each monitoring period of the historical monitoring data (if there is no corresponding indicator data at the same location under the corresponding monitoring, it is not included in the calculation; and the time length of each monitoring period is the same, so the corresponding time at the same location can be obtained), is weighted and averaged using the corresponding spatiotemporal reference factor as the weight, and the result is used as the reference data for the simulated data; the absolute value of the difference between the simulated data and the reference data is obtained as the initial simulation difference of the simulated data, the initial simulation difference of each simulated data in the current monitoring period is obtained, and a difference threshold is preset. In this embodiment, the difference threshold is described as 0.2. If there is simulated data with an initial simulation difference greater than the difference threshold, it is multiplied by the simulated data through the feedback adjustment coefficient, and the simulation difference is recalculated until there is no simulated data greater than the difference threshold. The final simulated data is used as the final simulated data of the current evaluation indicator in the current monitoring period of the current water area. Combined with the indicator data, the health data of the current evaluation indicator in the current monitoring period of the current water area is constituted. The health data of all evaluation indicators in the current water area in all monitoring periods constitute the multi-source and multi-dimensional health evaluation data of the current water area.
[0080] Thus, we have obtained multi-source, multi-dimensional health assessment data for the current water area.
[0081] Step S004: Based on the multi-source and multi-dimensional health assessment data of the current water area, construct an indicator-time data matrix for each monitoring location, and realize the dynamic health assessment of the current water area through a collaborative filtering algorithm.
[0082] It should be noted that for multi-source, multi-dimensional health assessment data, a matrix is constructed based on monitoring locations (monitoring units). This matrix is built using multi-source, multi-dimensional health assessment data at the same time. It can reflect the overall health performance of each monitoring location in the current water area. By obtaining scores through collaborative filtering algorithms and data fusion through principal component analysis, the fusion results of monitoring locations and a comprehensive health score are obtained. The recommendation function of the collaborative filtering algorithm can predict the overall health status of each monitoring location in the current water area. That is, the recommendation effect achieves comprehensive analysis of multiple monitoring locations, while also having a certain predictive effect, thereby realizing dynamic health assessment.
[0083] Specifically, for any monitoring location in the current water area, health data for each evaluation indicator at each time point is obtained from the multi-source, multi-dimensional health assessment data. A matrix is constructed with the evaluation indicators as columns and time points as rows, and the health data is filled into the matrix to obtain the indicator-time data matrix for that monitoring location. Data fusion is performed on the indicator-time data matrix for that monitoring location to obtain the fusion results for each monitoring location in the current water area. In this embodiment, principal component analysis is used to reduce the dimensionality of the indicator-time data matrix to obtain the fusion results based on the monitoring location. At the same time, a comprehensive health score is calculated for the indicator-time data matrix. The specific score acquisition process is based on existing technologies in collaborative filtering algorithms, which evaluate each evaluation indicator based on its data and standard range, and then perform weight analysis on each evaluation indicator within the matrix to obtain the health score. This embodiment will not elaborate further. After obtaining the fusion results and health scores, the fusion results and comprehensive health scores for each monitoring location in the current water area are analyzed and recommended based on the collaborative filtering algorithm. The algorithm output is the predictive health assessment for the current water area, thereby realizing the intelligent dynamic assessment of the water ecological health of the current water area.
[0084] This concludes the embodiment.
[0085] Another embodiment of the present invention provides an intelligent dynamic evaluation system for water ecological health based on multi-source data fusion. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the above method steps S001 to S004.
[0086] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for intelligent dynamic evaluation of water ecological health based on multi-source data fusion, characterized in that, The method includes the following steps: Collect monitoring data from multiple monitoring locations and multiple sources in the water area, and obtain historical monitoring data from multiple sources and multiple dimensions for several water areas; Based on the time-series change trend of monitoring data for a single evaluation indicator, and combined with the overall change of monitoring data from multiple sources and multiple evaluation indicators, several monitoring periods for the water area are obtained. By removing anomalies, the indicator data for each evaluation indicator in each monitoring period are obtained. By combining historical monitoring data, the overall change trend of the same evaluation indicator under different monitoring periods is analyzed, and the scenario reference coefficients of the same evaluation indicator in each historical monitoring period are obtained for the current monitoring period. Based on the monitoring location distribution corresponding to the monitoring data of the same evaluation indicator and the historical monitoring data, and combined with the scenario reference coefficients, the spatiotemporal reference factors of each historical monitoring data are obtained. The correlation between the change data of different evaluation indicators and historical monitoring data is analyzed to obtain the correlation of changes in different evaluation indicators. The feedback adjustment coefficient of each evaluation indicator is obtained by combining the correlation of changes in all evaluation indicators. Simulation data of each evaluation indicator in the current water area is obtained by Gaussian distribution. Similarity analysis is performed between the simulation data and the spatiotemporal reference factor of historical monitoring data, and the data is adjusted by feedback adjustment coefficient. Finally, multi-source and multi-dimensional health evaluation data of the current water area are obtained. Based on the multi-source and multi-dimensional health assessment data of the current water area, an indicator-time data matrix is constructed for each monitoring location, and a collaborative filtering algorithm is used to realize the dynamic health assessment of the current water area. The spatiotemporal reference factors for each historical monitoring data are obtained using the following method: For any evaluation indicator in the current water area, obtain the local monitoring sequence of any two monitoring locations during the current monitoring period, and obtain the cosine similarity of the two local monitoring sequences as the local similarity of the two monitoring locations for the current evaluation indicator in the current monitoring period. For the historical monitoring data of the current water area, obtain the corresponding local similarity of the two monitoring locations for each evaluation indicator in each monitoring period, and take the mean of all local similarities as the monitoring similarity of the two monitoring locations for the current water area. Obtain the actual distance between any two monitoring locations, and determine the spatial reference factor for those two monitoring locations based on the actual distance; The product of the spatial reference factor of the two monitoring locations and the monitoring similarity is linearly normalized to obtain the spatial similarity factor of the two monitoring locations. Then, for any evaluation index of the current water area and the current monitoring location in the current monitoring period, the spatial similarity factor of the two monitoring locations is multiplied by the scenario reference coefficient of the two locations and the corresponding spatial similarity factor of the two monitoring locations to obtain the spatiotemporal reference factor of the current monitoring location of the current evaluation index of the current water area in the current monitoring period.
2. The intelligent dynamic evaluation method for water ecological health based on multi-source data fusion according to claim 1, characterized in that, The specific methods for obtaining data on the monitoring periods of the water area are as follows: The monitoring data of any monitoring location for any evaluation indicator in the current water area and several historical monitoring data together constitute the monitoring sequence of the evaluation indicator in the current water area. The monitoring sequence is decomposed by STL time series to obtain the trend term, periodic term and residual term. The periodic term is Fourier transformed to obtain the spectrum of the periodic term of the monitoring sequence. The amplitude of each frequency in the spectrum is obtained. The reciprocal of the frequency corresponding to the maximum amplitude is taken as the monitoring period of the evaluation indicator at that monitoring location. Obtain the monitoring period for each evaluation indicator of the current water area at the current monitoring location, obtain the least common multiple of the monitoring location for all monitoring periods, and take the average of the least common multiples corresponding to all monitoring locations as a monitoring period for the current water area.
3. The intelligent dynamic evaluation method for water ecological health based on multi-source data fusion according to claim 1, characterized in that, The specific methods for obtaining the indicator data for each evaluation indicator in each monitoring period are as follows: For any evaluation indicator of the current water area and any monitoring data for any monitoring period, based on the normal distribution and The criterion is to obtain the mean and standard deviation of all monitoring data during the monitoring period, remove outliers from the monitoring data during the monitoring period, and use the remaining monitoring data after screening as the screened data for the monitoring period. Obtain the screening data for each monitoring period of the current water area for this evaluation indicator, as well as the screening data for each monitoring period from all historical monitoring data. Standardize all the screening data for this evaluation indicator, and use the results as the indicator data for each monitoring period of this evaluation indicator.
4. The intelligent dynamic evaluation method for water ecological health based on multi-source data fusion according to claim 2, characterized in that, The specific method for obtaining the scenario reference coefficients of the same evaluation indicators for each historical monitoring period to the current monitoring period is as follows: For any evaluation indicator in the current water area, several indicator data for the current monitoring period are collected and arranged in chronological order to form an indicator sequence for the current monitoring period of the current evaluation indicator in the current water area; several indicator data for any evaluation indicator in any historical water area are collected and arranged in chronological order to form a historical indicator sequence for the evaluation indicator in that monitoring period of the current water area; DTW matching is performed between the indicator sequence and the historical indicator sequence to obtain the DTW distance, and the inversely proportionally normalized result of the obtained DTW distance is used as a reference factor for the change of the evaluation indicator in that monitoring period of the current water area to the current monitoring period of the current water area; Based on the trend changes of current and historical water monitoring data under the same evaluation indicators, and the time distribution of the corresponding monitoring period, time reference weights are obtained. The product of the time reference weight and the change reference factor is used as the scenario reference coefficient for the current monitoring period of the current monitoring period of the current water area and the evaluation index.
5. The intelligent dynamic evaluation method for water ecological health based on multi-source data fusion according to claim 4, characterized in that, The method for obtaining time reference weights based on the trend changes of current and historical water monitoring data under the same evaluation indicators, and the time distribution of the corresponding monitoring periods, includes the following specific methods: Based on the current monitoring period of any evaluation indicator in the current water area, extract the trend sequence of the current monitoring period of the evaluation indicator and the trend sequence of the adjacent previous monitoring period from the trend item of the monitoring sequence of the evaluation indicator. Obtain the difference sequence obtained by subtracting the trend sequence of the adjacent previous monitoring period from the trend sequence of the current monitoring period, and use it as the trend change sequence of the current monitoring period. Obtain any monitoring period for the evaluation index in any water area and use it as a reference monitoring period for the evaluation index in a reference water area; obtain the trend sequence of the reference monitoring period and the corresponding trend change sequence. The cosine similarity is obtained between the trend change sequence of the current monitoring period and the trend change sequence of the reference monitoring period of the evaluation index in the reference water area. Calculate the time distance between the first moment in the current monitoring period and the first moment in the reference monitoring period. Divide the time distance by the length of one year and use the decimal part of the quotient as the time difference factor between the current monitoring period and the reference monitoring period. Subtract the time difference factor from 1 and multiply it by the cosine similarity as the time reference weight of the reference monitoring period for the current monitoring period of the current water body.
6. The intelligent dynamic evaluation method for water ecological health based on multi-source data fusion according to claim 1, characterized in that, The spatiotemporal reference factors for each historical monitoring data are obtained using the following method: For any evaluation indicator in the current water area, obtain the local monitoring sequence of any two monitoring locations during the current monitoring period, and obtain the cosine similarity of the two local monitoring sequences as the local similarity of the two monitoring locations for the current evaluation indicator in the current monitoring period. For the historical monitoring data of the current water area, obtain the corresponding local similarity of the two monitoring locations for each evaluation indicator in each monitoring period, and take the mean of all local similarities as the monitoring similarity of the two monitoring locations for the current water area. For two monitoring locations in different water areas, the local similarity is obtained from the local monitoring sequences of the same evaluation index and the same monitoring period. Then, the monitoring similarity of the two monitoring locations in different water areas is obtained by the mean of the local similarity of the two monitoring locations under the same monitoring period for all evaluation indexes. Obtain the actual distance between any two monitoring locations, and obtain the water system basin to which the two monitoring locations belong. If the two monitoring locations belong to the same water system basin, obtain the ratio of the actual distance between the two monitoring locations to the average span of the water system basin. Subtract the ratio from 1 to obtain the difference, and add 1 to obtain the sum, which is used as the spatial reference factor for the two monitoring locations. If the two monitoring locations belong to different river basins, the spatial reference factor for the two monitoring locations is set to 1. The product of the spatial reference factor of the two monitoring locations and the monitoring similarity is linearly normalized to obtain the spatial similarity factor of the two monitoring locations. Then, for any evaluation index of the current water area and the current monitoring location in the current monitoring period, the spatial similarity factor of the two monitoring locations is multiplied by the scenario reference coefficient of the two locations to obtain the spatiotemporal reference factor of the current monitoring location of the current evaluation index in the current monitoring period for the current water area. The spatiotemporal reference factor is assigned to the local monitoring sequence of the monitoring location for the evaluation index in the water area during the monitoring period, and serves as the spatiotemporal reference factor for each historical monitoring data point for the current monitoring location of the evaluation index in the current water area during the current monitoring period.
7. The intelligent dynamic evaluation method for water ecological health based on multi-source data fusion according to claim 2, characterized in that, The specific methods for obtaining the correlation of changes in different evaluation indicators and synthesizing the correlation of changes in all evaluation indicators to obtain the feedback adjustment coefficient of each evaluation indicator include: For any two evaluation indicators in the current water area during the same monitoring period, calculate the DTW distance between the two indicator sequences and perform inverse proportional normalization. The result is used as the correlation of changes in the two evaluation indicators in the current water area during the same monitoring period. Calculate the mean of the correlation of changes in the two evaluation indicators in all water areas during the same monitoring period, and use it as the correlation of changes in the two evaluation indicators. Take any evaluation index as the current evaluation index, obtain the monitoring period corresponding to the current evaluation index at any monitoring location in the current water area, and take the ratio of the monitoring period to the least common multiple of all monitoring periods corresponding to the monitoring location as the period factor of the current evaluation index at that monitoring location in the current water area. Obtain the periodic factor of any other evaluation indicator besides the current evaluation indicator at the current water area and at the current monitoring location. Use the absolute value of the difference between the two periodic factors as the periodic difference factor between the current evaluation indicator and the other evaluation indicator at the current water area and at the current monitoring location. Obtain the mean of the periodic difference factors between the current evaluation indicator and other evaluation indicators at all monitoring locations in the current water area, and subtract the mean from 1 to obtain the difference, which is used as the periodic reference weight of the other evaluation indicators relative to the current evaluation indicator. Based on the aforementioned periodic reference weights, the correlation between the changes of all evaluation indicators other than the current evaluation indicator and the current evaluation indicator is weighted and averaged, and the result is used as the feedback adjustment coefficient of the current evaluation indicator.
8. The intelligent dynamic evaluation method for water ecological health based on multi-source data fusion according to claim 1, characterized in that, The specific methods for obtaining the multi-source, multi-dimensional health assessment data of the current water area include: For any simulated data, the corresponding indicator data of the simulated data at the same location in each monitoring period of the current evaluation indicator of each water area in the historical monitoring data are weighted and averaged with the corresponding spatiotemporal reference factor. The result is used as the reference data of the simulated data. The absolute value of the difference between the simulated data and its reference data is obtained as the initial simulated difference. The initial simulated differences of each simulated data in the current monitoring period are obtained, and a difference threshold is preset. If there are simulated data with an initial simulated difference greater than the difference threshold, the simulated data is multiplied by the feedback adjustment coefficient, and the simulated difference is recalculated until there are no simulated data with a difference greater than the difference threshold. The final simulated data is used as the final simulated data of the current evaluation indicator in the current monitoring period of the current water area. Combined with the indicator data, the health data of the current evaluation indicator in the current monitoring period of the current water area is constituted. The health data of all evaluation indicators in the current water area in all monitoring periods constitute the multi-source and multi-dimensional health evaluation data of the current water area.
9. The intelligent dynamic evaluation method for water ecological health based on multi-source data fusion according to claim 1, characterized in that, The specific methods for constructing the indicator-time data matrix for each monitoring location and using a collaborative filtering algorithm to achieve dynamic health assessment of the current water area include: For any monitoring location in the current water area, the health data of each evaluation indicator at each time point is obtained from the multi-source multi-dimensional health evaluation data. A matrix is constructed with the evaluation indicators as columns and time as rows, and the health data is filled into the matrix to obtain the indicator-time data matrix of the monitoring location. Data fusion is performed on the indicator-time data matrix of the monitoring location to obtain the fusion results of each monitoring location in the current water area, and prediction is made based on the indicator-time data matrix; A comprehensive health score is calculated based on the aforementioned indicator-time data matrix; After obtaining the fusion results and health scores, the fusion results and comprehensive health scores of each monitoring location in the current water area are analyzed and recommended based on the collaborative filtering algorithm. The algorithm output results serve as a predictive health assessment of the current water area.
10. A water ecological health intelligent dynamic evaluation system based on multi-source data fusion, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent dynamic evaluation method for water ecological health based on multi-source data fusion as described in any one of claims 1-9.