Intelligent buoy dynamic early warning method and platform for environment monitoring

Abstract

The invention provides an intelligent buoy dynamic early warning method and platform for environmental monitoring, and relates to the technical field of environmental monitoring early warning, and the method comprises the steps: carrying out the interpolation prediction through reading the real-time position coordinates of a buoy and multi-source monitoring data, and generating multi-source prediction data distribution; recording multi-source prediction data distribution under continuous P monitoring points to construct a multi-source prediction data distribution sequence, carrying out water quality pollution analysis on a target water area in a future time zone, and generating a plurality of water quality pollution probabilities; obtaining a maximum pollution probability and a pollution probability mean value according to the plurality of water quality pollution probabilities; and if the maximum pollution probability is greater than a first probability threshold and / or the pollution probability mean value is greater than a second probability threshold, performing water quality pollution early warning on the target water area. According to the method and the device, the technical problem that the early warning accuracy is influenced due to the fact that monitoring data is difficult to accurately obtain because the buoy is easily interfered by environmental factors and has position fluctuation in the prior art is solved, and the environmental monitoring accuracy and the water quality early warning accuracy are improved.

Description

Technical Field

[0001] This application relates to the technical field of environmental monitoring and early warning, and specifically to an intelligent buoy dynamic early warning method and platform for environmental monitoring. Background Art

[0002] As an important tool for environmental monitoring, buoys are widely used in fields such as water quality monitoring, meteorological data collection, and ecological environment assessment. The traditional buoy environmental monitoring and early warning method mainly involves equipping the buoy with sensors and placing it at a fixed position to collect water quality parameters in real time and issue early warnings based on preset thresholds.

[0003] However, due to the complex water environment, buoys are easily interfered by environmental factors such as water flow and wind movement during actual operation, and their positions may fluctuate, making it difficult to maintain a set predetermined position for a long time. This position deviation will lead to a decrease in the spatial consistency of the collected data, thereby affecting the accuracy and integrity of environmental monitoring data. In addition, most methods rely strongly on single monitoring data and fail to fully integrate multi-source data for comprehensive analysis. This limitation results in insufficient spatial distribution accuracy of the monitoring data, and in a complex water environment, the early warning results may be distorted or delayed, reducing the reliability of water pollution monitoring and early warning. Summary of the Invention

[0004] This application provides an intelligent buoy dynamic early warning method and platform for environmental monitoring, which solves the technical problem in the prior art that due to the buoy being easily interfered by environmental factors and having position fluctuations, it is difficult to accurately obtain monitoring data at the predetermined position, thereby affecting the accuracy of early warning, and achieves the technical effect of improving the accuracy of environmental monitoring and the accuracy of water quality early warning.

[0005] In view of the above problems, on the one hand, this application provides an intelligent buoy dynamic early warning method for environmental monitoring. The method includes: reading the real-time position coordinates and multi-source monitoring data of a number of buoys in the target water area at fixed points, performing interpolation prediction based on the real-time position coordinates and multi-source monitoring data to obtain multi-source prediction data of the calibrated position coordinates of a number of buoys, and generating a multi-source prediction data distribution; recording the multi-source prediction data distributions under consecutive P monitoring points to construct a multi-source prediction data distribution sequence, performing water pollution analysis on the target water area in the future time zone based on the multi-source prediction data distribution sequence to generate multiple water pollution probabilities; respectively performing maximum probability extraction and probability mean calculation based on the multiple water pollution probabilities to obtain the maximum pollution probability and the pollution probability mean; if the maximum pollution probability is greater than the first probability threshold and / or the pollution probability mean is greater than the second probability threshold, then issue a water pollution early warning for the target water area.

[0006] On the other hand, the present application also provides an intelligent buoy dynamic warning platform for environmental monitoring. The platform includes: an interpolation prediction module, configured to read the real-time position coordinates and multi-source monitoring data of a plurality of buoys in a target water area at fixed points, perform interpolation prediction based on the real-time position coordinates and multi-source monitoring data, obtain multi-source prediction data of the calibrated position coordinates of the plurality of buoys, and generate a multi-source prediction data distribution; a water quality pollution analysis module, configured to record the multi-source prediction data distributions at consecutive P monitoring points to construct a multi-source prediction data distribution sequence, perform water quality pollution analysis of the target water area in a future time zone based on the multi-source prediction data distribution sequence, and generate a plurality of water quality pollution probabilities; a probability extraction and calculation module, configured to perform maximum probability extraction and probability mean calculation respectively based on the plurality of water quality pollution probabilities, and obtain a maximum pollution probability and a pollution probability mean; a pollution warning module, configured to issue a water quality pollution warning for the target water area if the maximum pollution probability is greater than a first probability threshold and / or the pollution probability mean is greater than a second probability threshold.

[0007] One or more technical solutions provided in the present application have at least the following technical effects or advantages:

[0008] By reading the real-time position coordinates and multi-source monitoring data of a plurality of buoys in a target water area at fixed points, the actual positions of the buoys when their positions fluctuate due to factors such as sea waves, wind speed, and water flow velocity can be accurately grasped. The multi-source monitoring data can provide more comprehensive environmental information, and these data provide the original data for subsequent analysis, avoiding data acquisition errors caused by buoy position fluctuations and improving the timeliness and accuracy of the data. By performing interpolation prediction based on the real-time position coordinates and multi-source monitoring data to obtain multi-source prediction data of the calibrated position coordinates of the buoys, the data that has not been collected can be estimated according to the position fluctuation of the buoys, so as to estimate the data that the buoys should have at the calibrated positions and ensure the spatial consistency of the data. By recording the multi-source prediction data distributions at consecutive P monitoring points to construct a multi-source prediction data distribution sequence, the environmental change trend can be better captured, the temporal characteristics of water quality changes can be analyzed, and the accuracy of pollution warning can be improved. By analyzing the multi-source prediction data distribution sequence to predict the future water quality pollution risk and generating a plurality of water quality pollution probabilities, multi-dimensional information can be provided for decision-making, helping to analyze the possibility of water quality changes and making early emergency response preparations. By performing maximum probability extraction and probability mean calculation based on the plurality of water quality pollution probabilities, the water quality pollution is quantified from multiple dimensions and the error caused by single probability judgment is reduced, making the warning result more reliable. When the pollution probability exceeds the set threshold, a water quality pollution warning is issued. By setting two probability thresholds and issuing a water quality pollution warning when the pollution probability exceeds the set threshold, it is ensured that an alarm can be issued in a timely manner when the water quality pollution risk reaches a certain level, providing timely information support for environmental protection and treatment.

[0009] In summary, the present application effectively solves the problem of insufficient data accuracy caused by the position fluctuation of the buoy in buoy monitoring, improves the environmental monitoring accuracy and the accuracy of water pollution early warning, and provides more accurate decision-making support for water pollution prevention and emergency response.

[0010] The above description is only an overview of the technical solution of the present application. In order to be able to more clearly understand the technical means of the present application, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present application more obvious and understandable, the specific embodiments of the present application are specifically exemplified below. Brief Description of the Drawings

[0011] Figure 1 It is a schematic flow chart of an intelligent buoy dynamic early warning method for environmental monitoring provided by an embodiment of the present application.

[0012] Figure 2 It is a schematic flow chart of interpolating and predicting multiple unmarked calibration position coordinates according to real-time position coordinates and multi-source monitoring data in the intelligent buoy dynamic early warning method for environmental monitoring provided by an embodiment of the present application.

[0013] Figure 3 It is a schematic structural diagram of an intelligent buoy dynamic early warning platform for environmental monitoring provided by an embodiment of the present application.

[0014] Description of the reference numerals: interpolation prediction module 10, water pollution analysis module 20, probability extraction and calculation module 30, pollution early warning module 40. Detailed Embodiments

[0015] The embodiment of the present application provides an intelligent buoy dynamic early warning method and platform for environmental monitoring, solves the technical problem in the prior art that it is difficult to accurately obtain the monitoring data at the predetermined position due to the position fluctuation of the buoy caused by the easy interference of the buoy by environmental factors, and further affects the accuracy of early warning, and achieves the technical effect of improving the environmental monitoring accuracy and the water quality early warning accuracy.

[0016] Embodiment 1, as Figure 1 shown, the embodiment of the present application provides an intelligent buoy dynamic early warning method for environmental monitoring, and the method includes:

[0017] Step S1: Read the real-time position coordinates and multi-source monitoring data of several buoys in the target water area at fixed points, perform interpolation prediction according to the real-time position coordinates and multi-source monitoring data, obtain multi-source prediction data of the calibration position coordinates of several buoys, and generate a multi-source prediction data distribution.

[0018] Specifically, the buoy is equipped with a variety of sensors, such as water temperature sensors, pH sensors, dissolved oxygen sensors, salinity sensors, turbidity sensors, etc., which are used to monitor water environment parameters in real time. The real-time position coordinates of several buoys in the target waters are read at a fixed point through a positioning system (such as GPS), and multi-source monitoring data are obtained using various sensors on the buoys. These multi-source monitoring data are obtained by multiple different types of sensors on the buoy. Using interpolation prediction methods (such as linear interpolation or spline interpolation), data prediction is performed for unobserved positions based on real-time position coordinates and multi-source monitoring data. Through interpolation prediction, multi-source prediction data of multiple buoys can be obtained, and these multi-source prediction data are integrated according to spatial distribution to generate multi-source prediction data distribution. Multi-source prediction data distribution includes multi-source prediction data at different locations at the same time, reflecting the spatial distribution of water environment data.

[0019] By dynamically reading the buoy position and monitoring data and using interpolation prediction methods to correct data errors caused by buoy position fluctuations, the accuracy of the data can be improved, ensuring that the generated environmental data distribution is more reliable, and providing basic data for subsequent pollution analysis.

[0020] Step S2: Record the distribution of multi-source prediction data at P consecutive monitoring points to construct a multi-source prediction data distribution sequence, perform water quality pollution analysis on the target water area in the future time zone according to the multi-source prediction data distribution sequence, and generate multiple water quality pollution probabilities.

[0021] Specifically, record the multi-source prediction data distribution of P monitoring points, which are specific time points for data collection in the target waters, and the time intervals between two adjacent monitoring points are the same, such as setting a monitoring point every two hours from the time point of starting monitoring, and setting a total of P monitoring points, wherein P is a positive integer representing the number of monitoring points, which can be adjusted according to specific monitoring needs. Arrange the multi-source prediction data distribution of each monitoring point according to the time sequence of the P monitoring points to generate a multi-source prediction data distribution sequence. Then, use a model (such as a regression analysis model based on statistical principles or a model based on machine learning) to perform water pollution analysis of the target waters in the future time zone according to the multi-source prediction data distribution sequence, and generate a water pollution probability for each monitoring indicator to obtain multiple water pollution probabilities. These water pollution probabilities are the probability values ​​of the water pollution risk of each location in the target waters at a certain moment in the future.

[0022] By constructing a multi-source prediction data distribution sequence, the water quality of the target waters can be grasped from a dynamic and overall perspective, overcoming the limitations of single time point and single location data. Water pollution analysis based on this sequence can more accurately predict water pollution in future time zones, providing a forward-looking data basis for water quality management and pollution warning.

[0023] Step S3: Perform maximum probability extraction and probability mean calculation respectively according to the multiple water quality pollution probabilities, and obtain the maximum pollution probability and the pollution probability mean.

[0024] Specifically, obtain the multiple water quality pollution probabilities generated in Step S2, find the maximum probability among them through a comparison algorithm, denoted as the maximum pollution probability, and at the same time calculate the mean of all probabilities, that is, add up all the water quality pollution probabilities and then divide by the number of water quality pollution probabilities to obtain the pollution probability mean. Through maximum probability extraction and mean calculation, two different dimensions of information are provided for pollution warning: the maximum pollution risk in the worst case and the overall pollution trend, so as to more comprehensively evaluate the potential risk of water quality pollution.

[0025] Step S4: If the maximum pollution probability is greater than the first probability threshold and / or the pollution probability mean is greater than the second probability threshold, issue a water quality pollution warning for the target water area.

[0026] Specifically, the first probability threshold and the second probability threshold are preset pollution warning thresholds for judging whether a warning needs to be issued. Compare the maximum pollution probability with the first probability threshold and compare the pollution probability mean with the second probability threshold. If the maximum pollution probability is greater than the first probability threshold or the pollution probability mean is greater than the second probability threshold, trigger a water quality pollution warning for the target water area. By setting two probability thresholds for double judgment, the accuracy and reliability of water quality pollution warning are improved. It takes into account both the possibility of the most serious pollution situation (maximum pollution probability) and the overall pollution situation (pollution probability mean), and can more reasonably judge whether a warning needs to be issued for the target water area, avoiding misjudgment or missed judgment that may be caused by a single threshold judgment, and improving the accuracy and rationality of water quality pollution warning.

[0027] Further, Step S1 includes:

[0028] Step S11: Obtain the coordinate positions of several calibration points of several buoys in the target water area, calculate the relative distances between the coordinate positions of the several calibration points and the nearest real-time position coordinates respectively. If the relative distance is less than the predetermined distance threshold, set the multi-source monitoring data corresponding to the real-time position coordinates as the multi-source prediction data corresponding to the corresponding calibration position coordinates.

[0029] Step S12: Screen and obtain multiple unmarked calibration position coordinates, and perform interpolation prediction on the multiple unmarked calibration position coordinates respectively according to the real-time position coordinates and the multi-source monitoring data to obtain multiple multi-source prediction data of the multiple unmarked calibration position coordinates.

[0030] Step S13: Arrange the multiple multi-source prediction data of the several calibration position coordinates according to the positions to generate the multi-source prediction data distribution.

[0031] Specifically, obtain the calibration position coordinates of multiple buoys in the target water area, which are the positions where the buoys should be under no interference. Calculate the relative distance between the actual position (real-time position coordinates) of each buoy and the calibration position coordinates, which can be calculated by the Euclidean distance formula. Compare the relative distance with a predetermined distance threshold. If the calculated relative distance is less than the predetermined distance threshold, it indicates that the actual position of the buoy is relatively close to the calibration position, and the multi-source monitoring data corresponding to the real-time position coordinates can be directly regarded as the multi-source prediction data of the buoy at the calibration position. Among them, the predetermined distance threshold is a preset value used to determine whether the calibration position coordinates and the real-time position coordinates are close enough.

[0032] Mark the positions of all buoys whose relative distances are less than the predetermined distance threshold, and identify all unmarked buoy coordinates in the target water area, whose positions have changed significantly. For these buoy coordinates without marked calibration positions, interpolation prediction is performed based on the real-time position coordinates of nearby buoys and the existing multi-source monitoring data. Through the interpolation method, the water quality data at these unmarked positions can be predicted. The interpolation prediction method fills in the missing data at the unmarked positions, thereby ensuring that the water quality monitoring data of all buoys in the target water area can be effectively supplemented, providing complete monitoring data for subsequent pollution analysis.

[0033] Arrange the calibration position coordinates of multiple buoys and their corresponding multi-source prediction data, that is, perform spatial sorting according to the calibration position coordinates (such as longitude and latitude coordinates) of each buoy in the water area to generate the distribution of multi-source prediction data at the current moment in the target water area. For example, the position of each buoy in the water area and its monitoring data can form a data set of coordinate points, forming a complete spatial data map to display the spatial distribution of water quality changes. Through spatial arrangement, a complete distribution of multi-source prediction data is formed, which can accurately reflect the environmental monitoring situation of each point in the target water area, ensuring the spatial accuracy and timeliness of the data, and providing accurate spatial data support for subsequent water quality pollution analysis and early warning.

[0034] Furthermore, as Figure 2 shown, step S12 includes:

[0035] Step S121: Randomly select the first unmarked calibration position coordinates. Based on the first unmarked calibration position coordinates, search within a predetermined distance range to obtain multiple first adjacent real-time position coordinates, and analyze to obtain multiple position correlation information, where the position correlation information includes correlation distance and correlation azimuth.

[0036] Step S122: Obtain multi-source monitoring indicators, conduct distance and azimuth correlation analysis on the multi-source monitoring indicators, and establish a multi-source position mapping correlation between distance, azimuth, and monitoring indicators, where the multi-source monitoring indicators at least include dissolved oxygen, pH value, turbidity, conductivity, temperature, heavy metal concentration, chemical oxygen demand, and ammonia nitrogen.

[0037] Step S123: Based on the multi-source position mapping correlation, match and obtain multiple sets of correlation coefficients according to the multiple position correlation information, and analyze and determine multiple sets of correlation weights.

[0038] Step S124: Obtain multiple first multi-source monitoring data of multiple first adjacent real-time position coordinates, perform weighted interpolation prediction on the multiple first multi-source monitoring data according to the multiple sets of correlation weights, and output first multi-source prediction data.

[0039] Specifically, the specific process of interpolating and predicting multiple unmarked calibration position coordinates based on real-time position coordinates and multi-source monitoring data includes: First, randomly select one from all unmarked calibration position coordinates as the first unmarked calibration position coordinate. Then, based on this coordinate, search for buoys within a predetermined distance range through coordinate calculation and search algorithms to obtain the real-time position coordinates of these buoys, that is, multiple first adjacent real-time position coordinates. The predetermined distance range is a distance range of a circular or square area centered on a certain point set in advance. For example, it can be set as a circular area with a radius of 100 meters centered on a certain coordinate. After determining multiple first adjacent real-time position coordinates, analyze the position correlation information between multiple first adjacent real-time position coordinates and the first unmarked calibration position coordinate respectively. The position correlation information describes the relationship between two positions, including the correlation distance and correlation azimuth. The correlation distance can calculate the straight-line distance between two positions using a coordinate calculation tool, and the correlation azimuth can be determined through azimuth angle calculation. By determining the adjacent real-time position coordinates of the first unmarked calibration position coordinate and their position correlation information, the position relationship of relevant monitoring points around the target position is clarified, which helps to more accurately conduct location-based monitoring indicator analysis and interpolation prediction.

[0040] Obtain the multi-source monitoring indicators of the buoys. These multi-source monitoring indicators are different indicators collected by various sensors on the buoys to reflect the environmental conditions of the target water area, and at least include dissolved oxygen, pH value, turbidity, conductivity, temperature, heavy metal concentration, chemical oxygen demand, and ammonia nitrogen. For each monitoring indicator, conduct correlation analysis according to the previously obtained position correlation information (distance and azimuth). For example, analyze the change law of dissolved oxygen content at different distances and azimuths, establish the relationship between distance, azimuth, and dissolved oxygen content, and so on to establish the multi-source position mapping correlation between distance, azimuth, and all monitoring indicators.

[0041] Based on the established multi-source location mapping associations, match according to multiple location association information. Substitute the location association information corresponding to each first adjacent real-time location coordinate into the corresponding multi-source location mapping association to obtain the correlation coefficients corresponding to multiple monitoring indicators of each first adjacent real-time location coordinate. Integrate the correlation coefficients of multiple monitoring indicators corresponding to the same first adjacent real-time location coordinate to obtain multiple association weight sets.

[0042] Obtain the multi-source monitoring data corresponding to multiple first adjacent real-time location coordinates, that is, multiple first multi-source monitoring data. The first multi-source monitoring data is the actual monitoring values of each monitoring indicator monitored at each first adjacent real-time location coordinate. Then, according to the determined multiple association weight sets, perform weighted interpolation prediction on the first multi-source monitoring data and output the first multi-source prediction data. The formaldehyde interpolation can be calculated by the following formula: where D is the first multi-source prediction data, ω i is the association weight corresponding to the i-th monitoring indicator of location P i , D ( P i) is the actual monitoring value of the i-th monitoring indicator at location P i .

[0043] Randomly select another unmarked calibrated location coordinate from all unmarked calibrated location coordinates, and repeat the above interpolation prediction steps to obtain the multi-source prediction data corresponding to this unmarked calibrated location coordinate. Continuously repeat the steps of coordinate extraction and interpolation prediction to obtain multiple multi-source prediction data of multiple unmarked calibrated location coordinates. Through weighted interpolation prediction, the target water area can be predicted more accurately according to the data of adjacent buoys, ensuring the accuracy and reliability of data prediction, effectively improving the accuracy and reliability of the buoy environmental monitoring system, and providing accurate and reliable data for subsequent pollution early warning.

[0044] Further, step S122 includes:

[0045] Step S122-1: Randomly select a first monitoring indicator from the multi-source monitoring indicators, and retrieve the water quality pollution monitoring log of the target water area according to the first monitoring indicator to obtain a sample initial monitoring data set, a sample relative distance set, a sample relative azimuth set, and a sample associated monitoring data set.

[0046] Step S122-2: According to the sample initial monitoring data set and the sample associated monitoring data set, perform correlation analysis on the sample initial monitoring data and the sample associated monitoring data respectively to obtain a sample correlation coefficient set.

[0047] Step S122-3: Construct a first location mapping association according to the sample relative distance set, the sample relative azimuth set, and the sample correlation coefficient set, and add it to the multi-source location mapping association.

[0048] Specifically, randomly select one from the multi-source monitoring indicators as the first monitoring indicator. Then, according to the first monitoring indicator, use a database query tool or a data retrieval algorithm to retrieve the water quality pollution monitoring logs of the target water area, and query all the monitoring data related to this indicator as sample data, including the sample initial monitoring data set, the sample relative distance set, the sample relative azimuth set, and the sample associated monitoring data set. Among them, the sample initial monitoring data set refers to the monitoring values of the first monitoring indicator directly collected by the sensors on the buoy during the historical monitoring process; the sample relative distance set is the distances between multiple adjacent real-time position coordinates and the initial measurement point searched within a predetermined distance range; the sample relative azimuth set is the azimuths between multiple adjacent real-time position coordinates and the initial measurement point searched within a predetermined distance range, and there is a one-to-one correspondence between the sample relative distance set and the sample relative azimuth set; the sample associated monitoring data set is the monitoring values of the first monitoring indicator corresponding to multiple adjacent real-time position coordinates.

[0049] Use the correlation analysis method to analyze the sample initial monitoring data set and the sample associated monitoring data set to obtain the correlation coefficients between each sample associated monitoring data and the sample initial monitoring data, thus forming a sample correlation coefficient set. Correlation analysis can use the correlation coefficient calculation method in statistics, such as the Pearson correlation coefficient method. The sample correlation coefficient set quantifies the correlation degree between the sample initial monitoring data and the sample associated monitoring data, helps to more accurately understand the relationship between different monitoring data, provides an important data basis for constructing the position mapping association subsequently, and can integrate the association relationship between the position information and the monitoring data more scientifically.

[0050] Obtain the sample relative distance set, the sample relative azimuth set, and the sample correlation coefficient set. Then, construct the first position mapping association based on these data. The relative distance, relative azimuth, and correlation coefficient can be related by establishing a functional relationship or in the form of a data table. For example, taking the relative distance and relative azimuth as independent variables and the correlation coefficient as the dependent variable, construct a functional relationship, and this functional relationship is the first position mapping association, and then add it to the multi-source position mapping association.

[0051] For other indicators in the multi-source monitoring indicators, use the same method to construct the corresponding position mapping associations, so as to obtain a complete multi-source position mapping association. By constructing the multi-source position mapping association, it can more comprehensively reflect the association situation of different monitoring indicators under different position relationships, provide a more detailed and accurate basis for subsequent data matching and interpolation prediction according to the position association information, and help improve the accuracy of predicting the environmental conditions of the target water area.

[0052] Furthermore, step S2 includes:

[0053] Step S21: Randomly select the first prediction data distribution sequence of the first monitoring index from the multi-source prediction data distribution sequence, and match the first analysis plugin in the water quality pollution analysis channel according to the first monitoring index.

[0054] Step S22: Perform water quality pollution analysis on the first prediction data distribution sequence for the future time zone through the first analysis plugin, output the first water quality pollution probability, and add it to the multiple water quality pollution probabilities.

[0055] Specifically, randomly select a monitoring index from the multi-source prediction data distribution sequence as the first monitoring index, and then extract the data related to the first monitoring index from the multi-source prediction data distribution sequence to form the first prediction data distribution sequence. Then, according to the first monitoring index, perform a matching operation in the water quality pollution analysis channel to find the first analysis plugin suitable for analyzing the first monitoring index. The water quality pollution analysis channel is a machine learning model for water quality pollution analysis, which contains multiple analysis plugins for pollution analysis of different monitoring indexes. The first analysis plugin is a module in the water quality pollution analysis channel for analyzing the data of the first monitoring index.

[0056] Use the algorithm and parameter settings in the first analysis plugin to perform water quality pollution analysis on the first prediction data distribution sequence for the future time zone. For example, if the first analysis plugin is based on a machine learning algorithm, it will predict the water quality pollution situation in the future time zone according to the historical data (such as turbidity data at different past times) in the first prediction data distribution sequence and the pre-trained model, and output a value representing the possibility of water quality pollution, that is, the first water quality pollution probability. Finally, add this first water quality pollution probability to the multiple water quality pollution probabilities for subsequent comprehensive processing.

[0057] Randomly select another monitoring index from the multi-source prediction data distribution sequence, extract the data related to this monitoring index to form a prediction data distribution sequence, and select the analysis plugin corresponding to the new monitoring index from the water quality pollution analysis channel to predict the water quality pollution probability. Repeat this step to predict the water quality pollution probability according to each monitoring index in the multi-source prediction data distribution sequence respectively, and obtain the final multiple water quality pollution probabilities.

[0058] By analyzing the prediction data of each monitoring index and performing targeted calculations of pollution probabilities through the analysis plugin, it provides a scientific basis for subsequent pollution warnings, ensures that different monitoring indexes are reasonably analyzed and processed, finally outputs accurate water quality pollution probabilities, and enhances the accuracy and real-time performance of water quality pollution prediction.

[0059] Furthermore, the method described in the embodiment of the present application further includes constructing a water quality pollution analysis channel:

[0060] Step S2-1: According to the water quality pollution monitoring log of the target water area, with the first monitoring index and P monitoring points as constraints, collect the sample monitoring data distribution sequence set.

[0061] Step S2-2: Based on a predetermined similarity tolerance interval, count the water quality pollution frequencies of different sample monitoring data distribution sequences in the historical time zone, set as the sample water quality pollution probability, and obtain the sample water quality pollution probability set, where the time interval between the historical time zone and the future time zone is the same.

[0062] Step S2-3: Use the sample monitoring data distribution sequence set and the sample water quality pollution probability set as supervised data to train the BP neural network until convergence, obtain the first analysis plug-in, and sequentially construct multiple analysis plug-ins for multiple monitoring indicators to build the water quality pollution analysis channel.

[0063] Specifically, the process of constructing the water quality pollution analysis channel includes: determining the water quality pollution monitoring log of the target water area. The water quality pollution monitoring log records information such as various water quality index data of the target water area at different times and different monitoring points. According to the first monitoring index (the same as in step S21) and P monitoring points as constraints, from the water quality pollution monitoring data of the first monitoring index at the same time in the monitoring log, arrange the retrieved data in the same monitoring in the time order of the monitoring points to form a sample monitoring data distribution sequence. Summarize all the sample monitoring data distribution sequences to obtain the sample monitoring data distribution sequence set.

[0064] Obtain a preset similarity tolerance interval, which is the interval range for judging data similarity. Then, for each sample monitoring data distribution sequence in the sample monitoring data distribution sequence set, count the frequency of water quality pollution in the historical time zone. The time interval of this historical time zone is the same as that of the aforementioned future time zone. If the future time zone is from 5 days to 10 days in the future, the historical time zone is from 5 days to 10 days after the monitoring time period corresponding to the sample monitoring data. During the counting process, if the data in different sample monitoring data distribution sequences is within the preset similarity tolerance interval, it is considered a similar water quality pollution situation, and the frequency of its occurrence is counted. For example, for a sample monitoring data distribution sequence with dissolved oxygen as the first monitoring index, the number of times the dissolved oxygen value is within a certain range (determined by the preset similarity tolerance interval) is the frequency of water quality pollution. Dividing this frequency by the total number of statistical times gives the sample water quality pollution probability. Conduct water quality pollution frequency statistics for all sample monitoring data distribution sequences to obtain the sample water quality pollution probability set. By setting the similarity tolerance interval to count the sample water quality pollution probability set, the similarity of the data is considered, which can more reasonably evaluate the water quality pollution possibility of different sample monitoring data distribution sequences in the historical time zone, provide meaningful supervised data for subsequent training of the neural network, and help improve the accuracy of the constructed analysis plug-in for water quality pollution analysis.

[0065] Use the sample monitoring data distribution sequence set and the sample water quality pollution probability set as supervised data to train the BP neural network. First, use a deep learning framework, such as TensorFlow or PyTorch, etc., to construct an initialized BP neural network structure. The number of input layer nodes is the same as the dimension of the sample monitoring data distribution sequence; the number of hidden layer nodes is determined according to experience or the trial-and-error method; and the number of output layer nodes is 1, representing the water quality pollution probability. Then, use the supervised data to train the BP neural network. By continuously adjusting the weights and biases of the network, the error between the output of the network and the sample water quality pollution probability set gradually decreases until it converges (i.e., the error reaches an acceptable minimum value). Output the trained BP neural network as the first analysis plug-in. In the same way, construct the corresponding analysis plug-ins for other monitoring indicators in turn, and combine these analysis plug-ins together to form a water quality pollution analysis channel.

[0066] The above steps use the powerful non-linear fitting ability of the BP neural network to train the analysis plug-in and build the water quality pollution analysis channel, which can accurately construct the analysis plug-ins applicable to different monitoring indicators based on historical data. The water quality pollution analysis channel composed of these plug-ins can comprehensively and accurately analyze the water quality pollution of the target water area, improving the scientificity and reliability of water quality pollution analysis.

[0067] Further, the method according to the embodiment of the present application further includes: configuring a first probability threshold and a second probability threshold, wherein the second probability threshold is less than the first probability threshold.

[0068] Specifically, when warning of water quality pollution in a target water area, two different probability thresholds are configured according to the actual situation of the target water area (such as the function of the water area, the surrounding environment, the past water quality conditions, etc.) to verify the water quality pollution probability, namely the first probability threshold and the second probability threshold. The first probability threshold is used to judge serious pollution events, while the second probability threshold is used to remind of the risk of possible pollution. The second probability threshold should be less than the first probability threshold. When the water quality pollution probability reaches the second threshold, a warning or prompt is issued, but a full-scale emergency response is not triggered; while when the pollution probability exceeds the first threshold, stronger alarms and emergency measures will be triggered. For example, if the target water area is a drinking water source, the first probability threshold can be set relatively low, such as 0.3, indicating that when the water quality pollution probability reaches 0.3, high attention needs to be paid and strict protection measures should be taken; while the second probability threshold may be set to 0.1. When the water quality pollution probability is between 0.1 and 0.3, relatively mild measures such as increasing the monitoring frequency can be taken. The configuration of the dual thresholds helps to grade and judge the degree of water quality pollution, so as to formulate response strategies more pertinently and protect the water quality safety of the target water area.

[0069] In summary, the intelligent buoy dynamic warning method for environmental monitoring provided by the embodiment of the present application has the following technical effects:

[0070] In the embodiment of the present application, by dynamically reading the real-time position and multi-source monitoring data of the buoy and combining the interpolation prediction technology to correct the fluctuations in the buoy position, the spatial consistency and accuracy of the data are ensured; by recording the multi-source data distribution of multiple consecutive monitoring points and analyzing the data sequence, the water quality change trend is effectively captured, improving the timeliness and accuracy of pollution warning; further, through the comprehensive calculation of the maximum pollution probability and the average pollution probability, a reliable judgment basis is provided for water quality pollution warning, and the warning mechanism of the dual probability thresholds further improves the scientificity and accuracy of the warning, reducing false alarms and missed alarms. Generally speaking, the embodiment of the present application realizes the dynamic monitoring and early warning of water quality pollution, significantly improves the accuracy of environmental monitoring and the accuracy and reliability of water quality pollution warning, and provides strong technical support for the protection and management of the water area environment.

[0071] Embodiment 2, as Figure 3 shown, based on the same inventive concept as the foregoing Embodiment 1, the embodiment of the present application provides an intelligent buoy dynamic warning platform for environmental monitoring, and the platform includes:

[0072] The interpolation prediction module 10 is configured to fixedly read the real-time position coordinates and multi-source monitoring data of a plurality of buoys in the target water area, perform interpolation prediction based on the real-time position coordinates and multi-source monitoring data, obtain the multi-source prediction data of the calibrated position coordinates of the plurality of buoys, and generate a multi-source prediction data distribution.

[0073] The water quality pollution analysis module 20 is configured to record the multi-source prediction data distribution under continuous P monitoring points to construct a multi-source prediction data distribution sequence, perform water quality pollution analysis on the target water area in the future time zone according to the multi-source prediction data distribution sequence, and generate a plurality of water quality pollution probabilities.

[0074] The probability extraction and calculation module 30 is configured to respectively perform maximum probability extraction and probability mean calculation according to the plurality of water quality pollution probabilities, and obtain the maximum pollution probability and the pollution probability mean.

[0075] The pollution warning module 40 is configured to perform a water quality pollution warning on the target water area if the maximum pollution probability is greater than the first probability threshold and / or the pollution probability mean is greater than the second probability threshold.

[0076] Furthermore, the interpolation prediction module 10 in the embodiment of the present application is further configured to perform the following steps:

[0077] Obtain the calibrated position coordinates of a plurality of buoys in the target water area, respectively calculate the relative distances between the calibrated position coordinates and the nearest real-time position coordinates. If the relative distance is less than a predetermined distance threshold, set the multi-source monitoring data corresponding to the real-time position coordinates as the multi-source prediction data corresponding to the calibrated position coordinates; screen and obtain a plurality of unmarked calibrated position coordinates, and perform interpolation prediction on the plurality of unmarked calibrated position coordinates respectively according to the real-time position coordinates and multi-source monitoring data to obtain a plurality of multi-source prediction data of the plurality of unmarked calibrated position coordinates; arrange the plurality of multi-source prediction data of the plurality of calibrated position coordinates according to the positions to generate the multi-source prediction data distribution.

[0078] Furthermore, the interpolation prediction module 10 in the embodiment of the present application is further configured to perform the following steps:

[0079] Randomly select the coordinates of the first unmarked calibration position. Based on the coordinates of the first unmarked calibration position, search within a predetermined distance range to obtain multiple sets of coordinates of the first adjacent real-time positions, and analyze to obtain multiple position correlation information, where the position correlation information includes the correlation distance and the correlation azimuth; obtain multi-source monitoring indicators, and perform distance and azimuth correlation analysis on the multi-source monitoring indicators to establish a multi-source position mapping correlation between distance, azimuth and the monitoring indicators, where the multi-source monitoring indicators at least include dissolved oxygen, pH value, turbidity, conductivity, temperature, heavy metal concentration, chemical oxygen demand and ammonia nitrogen; based on the multi-source position mapping correlation, match and obtain multiple sets of correlation coefficients according to the multiple position correlation information, and analyze and determine multiple sets of correlation weights; obtain multiple sets of first multi-source monitoring data of the multiple sets of coordinates of the first adjacent real-time positions, and perform weighted interpolation prediction on the multiple sets of first multi-source monitoring data according to the multiple sets of correlation weights, and output the first multi-source prediction data.

[0080] Further, the interpolation prediction module 10 of the embodiment of the present application is further configured to perform the following steps:

[0081] Randomly select a first monitoring indicator from the multi-source monitoring indicators, retrieve the water quality pollution monitoring log of the target water area according to the first monitoring indicator, and obtain a sample initial monitoring data set, a sample relative distance set, a sample relative azimuth set and a sample associated monitoring data set; according to the sample initial monitoring data set and the sample associated monitoring data set, perform correlation analysis on the sample initial monitoring data and the sample associated monitoring data respectively to obtain a sample correlation coefficient set; construct a first position mapping correlation according to the sample relative distance set, the sample relative azimuth set and the sample correlation coefficient set, and add it to the multi-source position mapping correlation.

[0082] Further, the water quality pollution analysis module 20 of the embodiment of the present application is further configured to perform the following steps:

[0083] Randomly select a first prediction data distribution sequence of a first monitoring indicator from the multi-source prediction data distribution sequence, and match a first analysis plug-in in the water quality pollution analysis channel according to the first monitoring indicator; perform water quality pollution analysis on the first prediction data distribution sequence in the future time zone through the first analysis plug-in, output a first water quality pollution probability, and add it to the multiple water quality pollution probabilities.

[0084] Further, the platform of the embodiment of the present application further includes an analysis channel construction module, and the analysis channel construction module is configured to perform the following steps:

[0085] According to the water quality pollution monitoring log of the target water area, a sample monitoring data distribution sequence set is collected with the first monitoring index and P monitoring points as constraints; based on a predetermined similarity tolerance interval, the water quality pollution frequencies of different sample monitoring data distribution sequences in the historical time zone are statistically calculated, set as the sample water quality pollution probability, and a sample water quality pollution probability set is obtained, where the time interval between the historical time zone and the future time zone is the same; using the sample monitoring data distribution sequence set and the sample water quality pollution probability set as supervised data, a BP neural network is trained until convergence to obtain the first analysis plug-in, and multiple analysis plug-ins for multiple monitoring indicators are sequentially constructed to build the water quality pollution analysis channel.

[0086] Further, the platform in the embodiment of the present application further includes a threshold configuration module, and the threshold configuration module is used to perform the following steps:

[0087] Configure a first probability threshold and a second probability threshold, where the second probability threshold is less than the first probability threshold.

[0088] Through the foregoing detailed description of the intelligent buoy dynamic warning method for environmental monitoring in this specification, those skilled in the art can clearly know the intelligent buoy dynamic warning platform for environmental monitoring in this embodiment. For the platform disclosed in Embodiment 2, since it corresponds to the method disclosed in Embodiment 1, it has corresponding functional modules and beneficial effects. For the relevant parts, refer to the description in the method part.

[0089] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A smart buoy dynamic early warning method for environmental monitoring, characterized in that: Methods include: Fixed-point reading of real-time position coordinates and multi-source monitoring data of several buoys in the target waters, interpolation prediction based on the real-time position coordinates and multi-source monitoring data, obtaining multi-source prediction data of the calibrated position coordinates of several buoys, and generating a multi-source prediction data distribution; Record the distribution of multi-source prediction data at P consecutive monitoring points to construct a multi-source prediction data distribution sequence, perform water quality pollution analysis on the target water area in the future time zone according to the multi-source prediction data distribution sequence, and generate multiple water quality pollution probabilities; According to the multiple water quality pollution probabilities, maximum probability extraction and probability mean calculation are performed respectively to obtain the maximum pollution probability and the pollution probability mean; If the maximum pollution probability is greater than the first probability threshold and / or the mean pollution probability is greater than the second probability threshold, a water pollution warning is issued for the target water area.

2. The intelligent buoy dynamic early warning method for environmental monitoring according to claim 1 is characterized in that: Performing interpolation prediction based on the real-time position coordinates and multi-source monitoring data, obtaining multi-source prediction data of the calibrated position coordinates of a plurality of buoys, and generating a multi-source prediction data distribution, including: Obtaining a number of calibrated position coordinates of a number of buoys in the target waters, respectively calculating the relative distances between the calibrated position coordinates and the nearest real-time position coordinates, and if the relative distances are less than a predetermined distance threshold, setting the multi-source monitoring data corresponding to the real-time position coordinates as the multi-source prediction data corresponding to the calibrated position coordinates; Filter and obtain multiple unlabeled calibrated position coordinates, and perform interpolation prediction on the multiple unlabeled calibrated position coordinates according to the real-time position coordinates and the multi-source monitoring data to obtain multiple multi-source prediction data of the multiple unlabeled calibrated position coordinates; Arrange a plurality of multi-source prediction data of a plurality of calibrated position coordinates according to positions to generate the multi-source prediction data distribution.

3. The intelligent buoy dynamic early warning method for environmental monitoring according to claim 2 is characterized in that: The method further comprises: performing interpolation prediction on the plurality of unlabeled calibrated position coordinates according to the real-time position coordinates and the multi-source monitoring data, comprising: Randomly select a first unmarked calibrated position coordinate, use the first unmarked calibrated position coordinate as a reference, search within a predetermined distance range, obtain a plurality of first adjacent real-time position coordinates, and analyze to obtain a plurality of position association information, wherein the position association information includes an association distance and an association direction; Acquire multi-source monitoring indicators, and perform distance and azimuth correlation analysis on the multi-source monitoring indicators, and establish multi-source position mapping associations between distance, azimuth and monitoring indicators, wherein the multi-source monitoring indicators include at least dissolved oxygen, pH value, turbidity, conductivity, temperature, heavy metal concentration, chemical oxygen demand and ammonia nitrogen; Based on the multi-source position mapping association, multiple association coefficient sets are obtained according to the matching of the multiple position association information, and multiple association weight sets are determined by analysis; A plurality of first multi-source monitoring data of a plurality of first adjacent real-time position coordinates are obtained, weighted interpolation prediction is performed on the plurality of first multi-source monitoring data according to the plurality of association weight sets, and first multi-source prediction data is output.

4. The intelligent buoy dynamic early warning method for environmental monitoring according to claim 3 is characterized in that: Performing distance and azimuth correlation analysis on the multi-source monitoring indicators, and establishing multi-source position mapping correlation between distance, azimuth and monitoring indicators, including: Randomly selecting a first monitoring indicator from the multi-source monitoring indicators, searching the water quality pollution monitoring log of the target water area according to the first monitoring indicator, and obtaining a sample initial monitoring data set, a sample relative distance set, a sample relative orientation set, and a sample associated monitoring data set; According to the sample initial monitoring data set and the sample associated monitoring data set, respectively, performing correlation analysis on the sample initial monitoring data and the sample associated monitoring data to obtain a sample correlation coefficient set; A first position mapping association is constructed according to the sample relative distance set, the sample relative orientation set and the sample association coefficient set, and is added to the multi-source position mapping association.

5. The intelligent buoy dynamic early warning method for environmental monitoring according to claim 3 is characterized in that: According to the multi-source prediction data distribution sequence, water quality pollution analysis of the target water area in the future time zone is performed to generate multiple water quality pollution probabilities, including: Randomly selecting a first prediction data distribution sequence of a first monitoring indicator from the multi-source prediction data distribution sequence, and matching a first analysis plug-in in a water quality pollution analysis channel according to the first monitoring indicator; The first analysis plug-in is used to perform water pollution analysis on the first predicted data distribution sequence in a future time zone, output a first water pollution probability, and add it to the multiple water pollution probabilities.

6. The intelligent buoy dynamic early warning method for environmental monitoring according to claim 5 is characterized in that: Construct water pollution analysis channel, including: According to the water quality pollution monitoring log of the target water area, with the first monitoring indicator and P monitoring points as constraints, a sample monitoring data distribution sequence set is collected; Based on a predetermined similar tolerance interval, the water pollution frequency of different sample monitoring data distribution sequences in the historical time zone is counted and set as the sample water pollution probability to obtain a sample water pollution probability set, wherein the time interval between the historical time zone and the future time zone is the same; The sample monitoring data distribution sequence set and the sample water quality pollution probability set are used as supervision data to train the BP neural network until convergence to obtain the first analysis plug-in, and then multiple analysis plug-ins for multiple monitoring indicators are constructed in turn to build the water quality pollution analysis channel.

7. The intelligent buoy dynamic early warning method for environmental monitoring according to claim 1 is characterized in that: A first probability threshold and a second probability threshold are configured, wherein the second probability threshold is smaller than the first probability threshold.

8. Intelligent buoy dynamic early warning platform for environmental monitoring, characterized by: The platform is used to execute the intelligent buoy dynamic early warning method for environmental monitoring according to any one of claims 1 to 7, comprising: An interpolation prediction module is used to read the real-time position coordinates and multi-source monitoring data of several buoys in the target waters at a fixed point, perform interpolation prediction based on the real-time position coordinates and multi-source monitoring data, obtain multi-source prediction data of the calibrated position coordinates of several buoys, and generate a multi-source prediction data distribution; A water pollution analysis module is used to record the distribution of multi-source prediction data under P consecutive monitoring points to construct a multi-source prediction data distribution sequence, and to perform water pollution analysis on target waters in future time zones according to the multi-source prediction data distribution sequence to generate multiple water pollution probabilities; A probability extraction and calculation module, used to perform maximum probability extraction and probability mean calculation according to the multiple water pollution probabilities, and obtain the maximum pollution probability and the pollution probability mean; The pollution warning module is used to issue a water pollution warning to the target water area if the maximum pollution probability is greater than a first probability threshold and / or the mean pollution probability is greater than a second probability threshold.

Images