Big data fusion-based offshore water quality problem mining, early warning and forecasting method

Through big data fusion technology, automatic audit and prediction models are built, and the problem of insufficient audit and prediction of water quality monitoring data in the existing technology is solved, the excavation and early warning of water quality problems is realized, and the quality and prediction accuracy of water quality monitoring data are improved. Sea area management.

CN119005397AInactive Publication Date: 2024-11-22CHINA SHIPPING ENVIRONMENT SCI & TECH (SHANGHAI) CO LTD +1

Patent Information

Application Number
CN202411010173.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-26
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing technology lacks effective water quality monitoring data review methods, cannot fully utilize the high temporal and spatial resolution of the marine automatic monitoring station, cannot effectively excavate water quality problems in nearshore waters, and lacks water quality prediction and forecasting mechanisms, making it difficult to support water quality protection in the sea area. and management.

Method used

Using nearshore water quality problems excavation and early warning and forecasting methods based on big data fusion, we can build automatic audit models and prediction models by obtaining and initially processing online monitoring data to realize automatic audit of water quality parameters and hourly and daily resolution of future water quality parameters. Forecast and warning based on the prediction results.

Benefits of technology

It improves the quality of monitoring data, explores potential water quality problems, improves the accuracy of water quality prediction, meets the time accuracy requirements of different application scenarios, and supports nearshore water quality protection and management.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119005397A_ABST
    Figure CN119005397A_ABST
Patent Text Reader

Abstract

The invention discloses an offshore water quality problem mining, early warning and forecasting method and system based on big data fusion, electronic equipment and a computer readable storage medium, and the method comprises the steps: further upgrading and optimizing an existing water quality monitoring data automatic auditing method, and mining a water quality problem which cannot be recognized by conventional auditing; seawater quality automatic monitoring data, tide data and meteorological data are fused, a machine learning model is utilized to predict water quality data after specific time, and early warning is carried out when a threshold condition is reached. According to the method, multi-source data are fused, offshore water quality potential problem excavation, water quality parameter prediction and water quality early warning and forecasting are carried out, and a basis is provided for supporting target area water quality control and pollution prevention.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of automatic review of water quality monitoring data and water quality prediction, and in particular to a method, system, electronic device and computer-readable storage medium for mining and early warning of water quality problems in nearshore waters based on big data fusion. Background Art

[0002] The ocean is an important source of material for human development and an important support for economic development. Fully understanding the quality and changing trends of the marine environment in nearshore waters is crucial to serving the development of the regional marine economy.

[0003] The routine monitoring frequency of "manual sampling-laboratory analysis" is low and cannot meet the needs of coastal waters environmental management. For this reason, the online water quality monitoring system in key sea areas has been gradually established and improved, and its long-term operation has accumulated a large amount of monitoring data. However, there are few studies on the audit methods of marine monitoring data, and there is no complete set of guidance documents for the audit rules of marine water quality monitoring data. The "Technical Specifications for Automatic Monitoring of Water Quality in Coastal Seas" (HJ731-2014) only puts forward simple regulations on data validity. If we can give full play to the advantages of high temporal resolution and spatial resolution of automatic monitoring stations at sea, optimize the audit methods to improve the quality of monitoring data, explore potential water quality problems, and carry out water quality prediction and forecasting, it will be of great significance to support the protection and management of water quality in coastal waters. Summary of the invention

[0004] The present invention proposes a method, system, electronic device and computer-readable storage medium for mining and early warning of water quality problems in coastal waters based on big data fusion. The purpose of the present invention is to make full use of online monitoring data of coastal waters, upgrade the automatic review method, improve the quality of monitoring data, and at the same time mine potential water quality problems, carry out water quality prediction and forecasting, and provide a basis for supporting water quality protection and management in coastal waters of target areas.

[0005] According to the present invention, a method for mining and early warning of water quality problems in coastal waters based on big data fusion is proposed, and the method comprises the following steps:

[0006] Step S1: obtaining online monitoring data of water quality in coastal waters, and performing preliminary processing on the online monitoring data to form a standard data table, thereby obtaining a standard data table that meets the requirements of automatic audit;

[0007] Step S2: construct an automatic audit model for online water quality monitoring data, which is trained and verified and meets the prediction accuracy requirements. The standard data table of step S1 is input into the automatic audit model, and each water quality parameter is automatically audited to discover water quality problems;

[0008] Step S3: Output of water quality automatic audit results, including water quality problem type, problem site, problem water quality parameters, and problem time period;

[0009] Step S4: construct hourly resolution and daily resolution water quality prediction models. Based on the standard data table of step S1, the tidal data and weather data of the seawater quality automatic monitoring station are integrated to predict the future water quality parameters after a certain period of time. For different water quality parameters at different sites, prediction models with different input contents are designed, and the mean absolute percentage error (MAPE) is used as the prediction accuracy of the model for each water quality parameter at each site to determine the optimal prediction model;

[0010] Step S5: Based on the optimal prediction result of the water quality parameters in step S4, an early warning is issued when the preset threshold conditions are reached.

[0011] Preferably, the preliminary processing of the automatic water quality monitoring data in step S1 specifically includes the following steps:

[0012] Step S2.1: Convert the site name, time format, and water quality parameter name of the data table into a specific format to meet the automatic audit recognition requirements;

[0013] Step S2.2: Clear the marker characters after the data values ​​in the data table and convert all the data in a column into numbers;

[0014] Step S2.3: Identify the abnormal values ​​in the data table that are beyond the range and replace them with null values;

[0015] Step S2.4: Obtain the time resolution of the data table.

[0016] Preferably, the automatic audit model for online water quality monitoring data in step S2 includes but is not limited to the following categories:

[0017] (1) Water quality parameters remain constant: a certain water quality parameter remains constant for N1 consecutive data values;

[0018] (2) Algal proliferation: pH greater than or equal to C1, dissolved oxygen greater than or equal to C2, and chlorophyll greater than or equal to C3;

[0019] (3) Abnormal water quality: pH is less than C4 or greater than C5, or dissolved oxygen is less than C6 or greater than C7, or active phosphate is greater than C8, or total nitrogen is greater than C9;

[0020] (4) Water quality is questionable: nitrite is less than C at two consecutive time points 10 .

[0021] Preferably, in step S4, hourly resolution and daily resolution water quality prediction models are constructed, and the machine learning models used include:

[0022] (1) Linear model: The linear model predicts a continuous target based on the linear relationship between the target water quality parameter and one or more other water quality parameters;

[0023] (2) Random Forest Model: The random forest model is a model that statistically calculates the relationship between each vector (x) and its category (y). Its advantage is that it can accurately give the relationship between the vectors covered in the training set and their categories.

[0024] According to the present invention, a nearshore water quality problem mining and early warning forecasting system based on big data fusion is also proposed, and the system includes the following modules:

[0025] Module M1: Water quality online monitoring data acquisition and standardization module, used to obtain the online monitoring data of the water quality of the nearshore waters, perform preliminary processing on the online monitoring data to form a standard data table, and obtain a standard data table that meets the requirements of automatic audit;

[0026] Module M2: Water quality data automatic review and problem mining module, which is used to automatically review each water quality parameter in the standard data table of module M1 based on the constructed water quality online monitoring data automatic review model to mine water quality problems;

[0027] Module M3: Water quality problem output module, used to output water quality automatic audit results, including water quality problem type, problem site, problem water quality parameters, and problem time period;

[0028] Module M4: Water quality parameter prediction module, used to use the hourly resolution and daily resolution water quality prediction models constructed, the automatic audit model has been trained and verified and meets the prediction accuracy requirements, based on the standard data table of module M1, integrating the tidal data and weather data of the seawater quality automatic monitoring station to predict future water quality parameters after a certain period of time;

[0029] Module M5: Water quality early warning and forecasting module, used to issue early warning and forecast after the water quality parameter prediction results of module M4 reach a certain threshold condition.

[0030] Preferably, the module M2 includes the following modules:

[0031] Module M2.1: Convert the site name, time format, and water quality parameter name of the data table into a specific format to meet the requirements of automatic audit identification;

[0032] Module M2.2: Clear the marker characters after the data values ​​in the data table and convert all the data in a column into numbers;

[0033] Module M2.3: Identify abnormal values ​​in the data table that are out of range and replace them with null values;

[0034] Module M2.4: Get the time resolution of the data table.

[0035] Preferably, the water quality online monitoring data automatic audit model constructed in the module M2 includes but is not limited to the following categories:

[0036] (1) Water quality parameters remain constant: a certain water quality parameter remains constant for N1 consecutive data values;

[0037] (2) Algal proliferation: pH greater than or equal to C1, dissolved oxygen greater than or equal to C2, and chlorophyll greater than or equal to C3;

[0038] (3) Abnormal water quality: pH is less than C4 or greater than C5, or dissolved oxygen is less than C6 or greater than C7, or active phosphate is greater than C8, or total nitrogen is greater than C9;

[0039] (4) Water quality is questionable: nitrite is less than C at two consecutive time points 10 .

[0040] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for mining and early warning of water quality problems in nearshore waters based on big data fusion as described above is implemented.

[0041] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for mining and early warning of water quality problems in nearshore waters based on big data fusion as described above is implemented.

[0042] Compared with the prior art, the present invention has the following beneficial technical effects:

[0043] 1. The present invention upgrades and optimizes the existing automatic audit method for online water quality monitoring data, supplements the construction of an automatic audit model for water quality data, improves the quality of monitoring data, and simultaneously mines potential water quality problems;

[0044] 2. The present invention combines tidal data and weather data to predict water quality parameters, which improves the accuracy of prediction results compared with traditional prediction methods based only on historical monitoring data;

[0045] 3. The present invention can make predictions with time accuracy of hours and days, and carry out water quality early warning and forecasting to meet the needs of different application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0047] Figure 1 It is a flow chart of the method for mining and early warning of water quality problems in coastal waters based on big data fusion in the present invention;

[0048] Figure 2 It is a structural schematic diagram of the nearshore waters water quality problem mining and early warning forecasting system based on big data fusion in the present invention. DETAILED DESCRIPTION

[0049] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0050] Embodiment 1

[0051] like Figure 1 As shown, this embodiment provides a method for mining and early warning of water quality problems in coastal waters based on big data fusion, including:

[0052] Step S1: obtaining online monitoring data of water quality in coastal waters, performing preliminary processing on the online monitoring data to form a standard data table, and obtaining a standard data table that meets the requirements of automatic audit. In this embodiment, the water quality parameters include water temperature, conductivity, salinity, dissolved oxygen, pH, turbidity, chlorophyll, cyanophyll, nitrite, nitrate, ammonia nitrogen, and active phosphate;

[0053] Step S2: construct an automatic audit model for online water quality monitoring data, which is trained and verified and meets the prediction accuracy requirements. The standard data table of step S1 is input into the automatic audit model, and each water quality parameter is automatically audited to discover water quality problems;

[0054] Step S3: Output of water quality automatic audit results, including water quality problem type, problem site, problem water quality parameters, and problem time period;

[0055] Step S4: construct hourly resolution and daily resolution water quality prediction models, based on the standard data table of step S1, integrating the tide data and weather data of the seawater quality automatic monitoring station to predict the future water quality parameters after a certain period of time;

[0056] Step S5: Based on the water quality parameter prediction result of step S4, an early warning is issued when the preset threshold condition is reached.

[0057] The preliminary processing of the water quality automatic monitoring data in step S1 specifically includes the following steps:

[0058] Step S2.1: Convert the site name, time format, and water quality parameter name of the data table into a specific format to meet the requirements of automatic audit recognition. In this embodiment, the time format is "year-month-day hour: minute: second", such as "2023-9-2616:00:00", and the water quality parameter name format is "indicator unit", such as "water temperature ℃", "conductivity mS / cm", "salinity ‰", "dissolved oxygen mg / L", "pH dimensionless", "turbidity NTU", "chlorophyll μg / L", "blue-green algae ce lls / μL", "nitrite mg / L", "nitrate mg / L", "ammonia nitrogen mg / L", "active phosphate mg / L";

[0059] Step S2.2: Clear the marker characters after the data values ​​in the data table and convert all the data in a column into numbers;

[0060] Step S2.3: Identify the abnormal values ​​in the data table that are beyond the range and replace them with null values. In this embodiment, the abnormal values ​​less than 0 or greater than 7999, the data with an order of magnitude ten times greater than the previous data, the pH greater than 15, the dissolved oxygen concentration less than 2 mg / L, the water temperature not less than 40°C, and the conductivity not less than 55 mS / cm are replaced with null values;

[0061] Step S2.4: Obtain the time resolution of the data table. Some judgments in the automatic audit in this embodiment require that the data comes from continuous monitoring. Therefore, the time resolution of the data series is obtained by subtracting the minimum value obtained by staggered time series. If the time difference between the data of two time points is greater than 1.5 times the resolution, the time of the two points is considered to be discontinuous, and some judgments on the speed of change of water quality parameters cannot be made.

[0062] Among them, the automatic audit model of water quality online monitoring data in step S2 includes but is not limited to the following categories:

[0063] (1) Water quality parameters remain constant: In this embodiment, the judgment standard is that for pH, 12 consecutive data values, or for other parameters, 6 consecutive parameter values, if the water quality parameters do not change, the problem of constant water quality parameters occurs;

[0064] (2) Algal proliferation: In this embodiment, the judgment criteria are pH greater than or equal to 8.5, dissolved oxygen greater than or equal to 9.0 mg / L, and chlorophyll greater than or equal to 18 μg / L;

[0065] (3) Abnormal water quality: In this embodiment, the judgment criteria are pH less than 7.5 or greater than 8.5, or dissolved oxygen less than 5.5 mg / L or greater than 9.5 mg / L, or active phosphate greater than C8, or total nitrogen greater than C9;

[0066] (4) Suspicious water quality: In this embodiment, the judgment standard for a certain site is that the nitrite is less than 0.001 mg / L for 12 consecutive time points.

[0067] Among them, in step S4, hourly resolution and daily resolution water quality prediction models are constructed, and the machine learning models used include:

[0068] (1) Linear model: The linear model predicts a continuous target based on the linear relationship between the target water quality parameter and one or more other water quality parameters;

[0069] (2) Random Forest Model: The random forest model is a model that statistically calculates the relationship between each vector (x) and its category (y). Its advantage is that it can accurately give the relationship between the vectors covered in the training set and their categories.

[0070] Combining the machine learning linear model Sc i kit-l earn and the random forest model, an artificial intelligence water quality prediction model based on the hourly and daily resolution water quality, tide, climate and meteorological big data of each station was constructed. The historical water quality, tide, climate and meteorological data of each station were used to train and learn the model, and the prediction accuracy of the model of each station was evaluated by the mean absolute error and mean absolute percentage error. Finally, the optimal prediction model suitable for different water quality factors of each station was screened out.

[0071] The general linear model or multiple regression model is a statistical linear model. The formula is:

[0072] Y=XB+U

[0073] Where Y is a matrix with a set of multivariate measurements (each column is a set of measurements of a dependent variable), X is the observation matrix of the independent variables, which can be a design matrix (each column is about one independent variable), B is a matrix containing the parameters that are usually to be estimated, and U is a matrix containing the errors (noise). The errors are usually assumed to be uncorrelated measurements and follow a multivariate normal distribution. If the errors do not follow a multivariate normal distribution, generalized linear models can be used to relax the assumptions about Y and U.

[0074] Random forest is an ensemble learning model based on the Bagging (Bootstrap Aggregation) strategy. It can effectively handle nonlinear problems and is good at handling a large number of samples and features. In addition, random forest also focuses on reducing variance. During the training process of the Bagging method, there is no dependency between the base learners, and parallel training can be achieved. By integrating multiple models, it can effectively deal with overfitting problems and improve the prediction accuracy and generalization ability of the model.

[0075] In this embodiment, the determination of relevant judgment parameters involved in the automatic review model of online water quality monitoring data is based on the long-term actual water quality conditions, management needs and existing water quality data analysis of the target nearshore waters; by changing different input contents and the selection of linear models and random forest models, multiple prediction models are trained for each site, and the model with the lowest mean absolute percentage error (MAPE) is selected as the optimal prediction model.

[0076] The optimal hourly resolution prediction model determined in this embodiment is to input 8 conventional water quality parameters and 4 nutrient parameter values ​​at a certain time point, as well as the weather and tidal data at the input time point and 24 and 48 hours thereafter, a total of 3 time points. The model outputs 8 conventional water quality parameters and 4 nutrient parameter values ​​24 and 48 hours after the input time point; the optimal daily resolution water quality prediction model is to input 8 conventional water quality parameters and 4 nutrient parameter values ​​for 7 consecutive days, as well as the weather and tidal data for the above 7 days and the next two days, and outputs 8 conventional water quality parameters and 4 nutrient parameter values ​​for two days after the above 7 days.

[0077] Embodiment 2

[0078] The same principle as the above-mentioned embodiment of a method for mining and early warning of water quality problems in coastal waters based on big data fusion, such as Figure 2 As shown, this embodiment provides a nearshore water quality problem mining and early warning forecasting system based on big data fusion, including:

[0079] Module M1: Water quality online monitoring data acquisition and standardization module, used to obtain online monitoring data of water quality in coastal waters, preliminarily process the online monitoring data to form a standard data table, and obtain a standard data table that meets the requirements of automatic audit. In this embodiment, the water quality parameters include water temperature, conductivity, salinity, dissolved oxygen, pH, turbidity, chlorophyll, cyanophyll, nitrite, nitrate, ammonia nitrogen, and active phosphate;

[0080] Module M2: Water quality data automatic review and problem mining module, which is used to automatically review each water quality parameter in the standard data table of module M1 based on the constructed water quality online monitoring data automatic review model to mine water quality problems;

[0081] Module M3: Water quality problem output module, used to output water quality automatic audit results, including water quality problem type, problem site, problem water quality parameters, and problem time period;

[0082] Module M4: Water quality parameter prediction module, used to use the hourly resolution and daily resolution water quality prediction models constructed, the automatic audit model has been trained and verified and meets the prediction accuracy requirements, based on the standard data table of module M1, integrating the tidal data and weather data of the seawater quality automatic monitoring station to predict future water quality parameters after a certain period of time;

[0083] Module M5: Water quality early warning and forecasting module, used to issue early warning and forecast after the water quality parameter prediction results of module M4 reach a certain threshold condition.

[0084] Wherein, the module M2 includes the following modules:

[0085] Module M2.1: Convert the site name, time format, and water quality parameter name of the data table into a specific format to meet the requirements of automatic audit and identification. In this embodiment, the time format is "year-month-day hour: minute: second", such as "2023-9-2616:00:00", and the water quality parameter name format is "indicator unit", such as "water temperature ℃". Any field can be automatically converted into a standard water quality parameter name, such as "temperature", "Temp", and "℃" can be converted into "water temperature ℃";

[0086] Module M2.2: Clear the marker characters after the data values ​​in the data table and convert all the data in a column into numbers;

[0087] Module M2.3: Identify abnormal values ​​in the data table that are beyond the range and replace them with null values. In this embodiment, abnormal values ​​less than 0 or greater than 7999, data with an order of magnitude ten times greater than the previous data, pH greater than 15, dissolved oxygen concentration less than 3 mg / L, water temperature not less than 40°C, and conductivity not less than 55 mS / cm are replaced with null values;

[0088] Module M2.4: Obtain the time resolution of the data table. Some judgments in the automatic audit in this embodiment require that the data comes from continuous monitoring. Therefore, the time resolution of the data series is obtained by subtracting the minimum value obtained by staggered time series. If the time difference between the data at two time points is greater than 1.5 times the resolution, the time of the two points is considered to be discontinuous, and some judgments on the speed of change of water quality parameters cannot be made.

[0089] Among them, the automatic audit model for online water quality monitoring data constructed in the module M2 includes but is not limited to the following categories:

[0090] (1) Water quality parameters remain constant: In this embodiment, the judgment standard is that for pH, 12 consecutive data values, or for other parameters, 6 consecutive parameter values, if the water quality parameters do not change, the problem of constant water quality parameters occurs;

[0091] The automated judgment method is to screen out the time points where the water quality parameters have not changed compared with the next time period or the previous time period, that is, the time points with a difference of 0, as a new time series. In this time series, the time point with a time difference of no more than 1.5 times the time resolution with the previous time data is the beginning of each continuous time period; similarly, the time point with a time difference of no more than 1.5 times the resolution with the next time data is the end of the continuous time period; the series of beginnings and ends obtained by this method are one-to-one corresponding, and each group of beginnings and ends belongs to the same continuous time period; and the difference between this group of beginnings and ends is the length of this continuous time period; for pH, if this continuous time period is greater than 12.5 times the time resolution, or greater than 6.5 times the time resolution for other parameters, then this parameter has a problem of constant value in this time period, and the start and end time points of the event are obtained;

[0092] (2) Algal proliferation: In this embodiment, the judgment criteria are pH greater than or equal to 8.5, dissolved oxygen greater than or equal to 9.0 mg / L, and chlorophyll greater than or equal to 16 μg / L;

[0093] (3) Abnormal water quality: In this embodiment, the judgment criteria are pH less than 7.5 or greater than 8.5, or dissolved oxygen less than 5.5 mg / L or greater than 9.5 mg / L, or active phosphate greater than C8, or total nitrogen greater than C9;

[0094] (4) Suspicious water quality: In this embodiment, the judgment standard for a certain site is that the nitrite is less than 0.001 mg / L for 12 consecutive time points.

[0095] The protection content of the present invention is not limited to the above embodiments. Without departing from the spirit and scope of the present invention, changes and advantages that can be thought of by those skilled in the art are included in the present invention and are protected by the attached claims.

Claims

1. A method for mining and early warning of water quality problems in coastal waters based on big data fusion, characterized in that: The steps include: Step S1: obtaining online monitoring data of water quality in coastal waters, and performing preliminary processing on the online monitoring data to form a standard data table, thereby obtaining a standard data table that meets the requirements of automatic audit; Step S2: construct an automatic audit model for online water quality monitoring data, which is trained and verified and meets the prediction accuracy requirements. The standard data table of step S1 is input into the automatic audit model, and each water quality parameter is automatically audited to discover water quality problems; Step S3: outputting the water quality automatic audit result, wherein the water quality automatic audit result includes the water quality problem type, problem site, problem water quality parameter, and problem time period; Step S4: Combine the machine learning linear model Scikit-learn and the random forest model to build an artificial intelligence water quality prediction model based on the big data of water quality, tides, climate and meteorology at each station with hourly resolution and daily resolution. Use the historical water quality, tides, climate and meteorology data of each station to train and learn the model, and evaluate the prediction accuracy of the model of each station with the mean absolute error and mean absolute percentage error, and finally screen out the optimal prediction model suitable for different water quality factors of each station; The formulas for mean absolute error and mean absolute percentage error are as follows: Step S5: Based on the optimal prediction result of the water quality parameters in step S4, an early warning is issued when the preset threshold conditions are reached.

2. According to claim 1, a method for mining and early warning of water quality problems in coastal waters based on big data fusion is characterized in that: The preliminary processing of the water quality automatic monitoring data in step S1 specifically includes the following steps: Step S2.1: Convert the site name, time format, and water quality parameter name in the data table of the online monitoring data into a specific format to meet the automatic audit recognition requirements; Step S2.2: Clear the marker characters after the data values ​​in the data table and convert all the data in a column into numbers; Step S2.3: Identify the abnormal values ​​in the data table that are beyond the range and replace them with null values; Step S2.4: Obtain the time resolution of the data table.

3. According to claim 1, a method for mining and early warning of water quality problems in coastal waters based on big data fusion is characterized in that: The automatic audit model for online water quality monitoring data in step S2 at least includes: (1) Water quality parameters remain constant: a certain water quality parameter remains constant for N1 consecutive data values; (2) Algal proliferation: pH greater than or equal to C1, dissolved oxygen greater than or equal to C2, and chlorophyll greater than or equal to C3; (3) Abnormal water quality: pH is less than C4 or greater than C5, or dissolved oxygen is less than C6 or greater than C7, or active phosphate is greater than C8, or total nitrogen is greater than C9; (4) Water quality is questionable: nitrite is less than C at two consecutive time points 10 .

4. According to the method of claim 1, the method is characterized by: In step S4, hourly resolution and daily resolution water quality prediction models are constructed, and the machine learning models used include: (1) Linear model: The linear model predicts a continuous target based on the linear relationship between the target water quality parameter and one or more other water quality parameters; (2) Random Forest Model: The random forest model is a model that statistically calculates the relationship between each vector (x) and its category (y). Its advantage is that it can accurately give the relationship between the vectors covered in the training set and their categories.

5. A system for mining and early warning of water quality problems in coastal waters based on big data fusion, characterized in that: Includes the following modules: Module M1: Water quality online monitoring data acquisition and standardization module, used to obtain the online monitoring data of the water quality of the nearshore waters, perform preliminary processing on the online monitoring data to form a standard data table, and obtain a standard data table that meets the requirements of automatic audit; Module M2: Water quality data automatic review and problem mining module, which is used to automatically review each water quality parameter in the standard data table of module M1 based on the constructed water quality online monitoring data automatic review model to mine water quality problems; Module M3: Water quality problem output module, used to output water quality automatic audit results, including water quality problem type, problem site, problem water quality parameters, and problem time period; Module M4: Water quality parameter prediction module, used to use the hourly resolution and daily resolution water quality prediction models constructed, the automatic audit model has been trained and verified and meets the prediction accuracy requirements, based on the standard data table of module M1, integrating the tidal data and weather data of the seawater quality automatic monitoring station to predict future water quality parameters after a certain period of time; Module M5: Water quality early warning and forecasting module, used to issue early warning and forecast after the water quality parameter prediction results of module M4 reach a certain threshold condition.

6. According to claim 5, a system for mining and early warning of water quality problems in coastal waters based on big data fusion is characterized in that: The module M2 includes the following modules: Module M2.1: Convert the site name, time format, and water quality parameter name of the data table into a specific format to meet the requirements of automatic audit identification; Module M2.2: Clear the marker characters after the data values ​​in the data table and convert all the data in a column into numbers; Module M2.3: Identify abnormal values ​​in the data table that are out of range and replace them with null values; Module M2.4: Get the time resolution of the data table.

7. The system for mining and early warning of water quality problems in coastal waters based on big data fusion according to claim 5 is characterized in that: The water quality online monitoring data automatic audit model constructed in the module M2 includes but is not limited to the following categories: (1) Water quality parameters remain constant: a certain water quality parameter remains constant for N1 consecutive data values; (2) Algal proliferation: pH greater than or equal to C1, dissolved oxygen greater than or equal to C2, and chlorophyll greater than or equal to C3; (3) Abnormal water quality: pH is less than C4 or greater than C5, or dissolved oxygen is less than C6 or greater than C7, or active phosphate is greater than C8, or total nitrogen is greater than C9; (4) Water quality is questionable: nitrite is less than C at two consecutive time points 10 .

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements the method for mining and early warning of nearshore waters water quality problems based on big data fusion as described in any one of claims 1 to 4.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the method for mining and early warning of nearshore waters quality problems based on big data fusion as described in any one of claims 1 to 4.

Citation Information

Patent Citations

  • Inshore water quality remote sensing early warning system

    CN106932550A

  • Method for predicting influence of tidal action on river water quality based on river water quality model

    CN110751331A

  • Scientific early warning method based on buoy monitoring in combination with machine learning

    CN111311874A

  • Offshore area ecological environment risk monitoring method, device and equipment and storage medium

    CN114199305A

  • Water environment monitoring method based on multi-source remote sensing and machine learning

    CN114384015A

Cited By

  • Nearshore seawater quality prediction and early warning method based on machine learning

    CN120509522A