A method and system for auxiliary prediction of chlorophyll content in small sample water bodies

By assisted prediction method of chlorophyll content in small sample water bodies, water body data is collected and processed and trained using weak regression models, the problems of high modeling cost and insufficient data in the existing technology are solved, and low cost is achieved for different waters Chlorophyll prediction of high generalization performance.

CN117763508BActive Publication Date: 2024-11-01JIANGNAN UNIV
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202311780040.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-12-22
Publication Date
2024-11-01
Estimated Expiration
2043-12-22

AI Technical Summary

Technical Problem

In the prior art, the monitoring or prediction methods of chlorophyll content have problems such as high modeling costs, neglecting the correlation of environmental image factors, and insufficient data.

Method used

A chlorophyll content auxiliary prediction method for small sample water bodies is adopted. By collecting water body information in different environments, the data set is constructed, data preprocessing and dimensionality reduction is performed, a mixed training set and test set are set, and a weak regression model is used for training. And constantly update the sample weights, and finally select the regression model with the best prediction effect for chlorophyll prediction.

Benefits of technology

It reduces the requirements for data quantity and quality in the modeling process, reduces the costs of data acquisition and model training, and establishes a low-cost and high generalization performance chlorophyll prediction model, which is suitable for environmental monitoring in different waters.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117763508B_ABST
    Figure CN117763508B_ABST
Patent Text Reader

Abstract

The present invention provides a method and system for auxiliary prediction of chlorophyll content in a small sample water body, which relates to the field of biological environmental science and technology. The method includes obtaining water body related data; preprocessing the data to obtain an ideal data set; calculating the average impact value, and performing dimensionality reduction operation processing on the ideal data set; setting a mixed training set and a test set, and setting a weight and a sample update coefficient for each sample; combining the data set and the weight vector, training a basic weak regression model, and continuously updating the sample weight; repeatedly establishing the model, saving and recording each weak regression model, and finally selecting the regression model with the best effect as a strong regression model; inputting the test sample into the strong regression model to obtain the chlorophyll prediction value of the test sample. The present invention utilizes the correlation between water environment influencing factors and chlorophyll, and between different water environments, to establish a chlorophyll prediction model with low cost and high generalization performance for the chlorophyll prediction process.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of biological environmental science and technology, and in particular to a method and system for auxiliary prediction of chlorophyll content in a small sample water body. Background Art

[0002] When measuring chlorophyll content, traditional fluorescence detection technology is usually used to measure the required parameters, such as fluorescence, to calculate chlorophyll content, or hyperspectral imaging methods are used to measure chlorophyll content. These methods usually require specific physical / chemical parameters and have high costs and complex installation requirements. Therefore, it is of great significance to study the impact of influencing factors in water bodies on chlorophyll content and related prediction models, and then promote their application in carbon emission control.

[0003] The influencing factors in water bodies include water temperature, pH, dissolved oxygen, conductivity, turbidity, permanganate index, ammonia nitrogen, total phosphorus, total nitrogen, etc. The existing technology establishes a data-driven chlorophyll prediction model for each unit, which has high requirements for data quantity and quality. When the collected data is small or the data is seriously polluted, it is difficult to establish an accurate model. In addition, the environmental influencing factors of each water area are different, and it is difficult to use a universal model to meet the needs of chlorophyll prediction for each water area unit. Therefore, modeling each water area unit separately requires repeated data collection and processing, model training, etc., and the modeling cost is high.

[0004] In summary, the existing chlorophyll content monitoring or prediction methods have the problems of high modeling cost, neglect of the correlation of environmental imaging factors, and difficulty in modeling when there is insufficient data. Summary of the invention

[0005] To this end, an embodiment of the present invention provides an auxiliary prediction method and system for chlorophyll content in small sample water bodies, which is used to solve the problems in the prior art of chlorophyll content monitoring or prediction methods, such as high modeling cost, neglect of the correlation of environmental image factors, and difficulty in modeling when there is insufficient data.

[0006] In order to solve the above problems, an embodiment of the present invention provides a method for auxiliary prediction of chlorophyll content in a small sample water body, the method comprising:

[0007] S1: Collect water body information in different environments, obtain relevant data of water bodies, and construct a data set, wherein the data set includes relevant data of small sample water bodies and relevant data of auxiliary sample water bodies, and the small sample water bodies and the auxiliary sample water bodies come from different environments;

[0008] S2: preprocessing the relevant data to obtain an ideal data set;

[0009] S3: Based on the ideal data set, calculate the average influence value, and perform dimensionality reduction processing on the ideal data set;

[0010] S4: According to the dataset after dimension reduction, set the mixed training set and test set, and set the weight and sample update coefficient for each sample;

[0011] S5: Combine the reduced-dimensional dataset and the weight vector to train the basic weak regression model and continuously update the sample weights;

[0012] S6: Repeat S4-S5 continuously, save and record each weak regression model, and finally select the regression model with the best prediction effect as the strong regression model;

[0013] S7: Input the test sample into the strong regression model to obtain the chlorophyll prediction value of the test sample.

[0014] Preferably, the method for collecting water body information under different environments and obtaining water body related data is:

[0015] Sensors are used to collect water information in different environments to obtain water-related data, including water temperature, pH value, dissolved oxygen content, potassium permanganate content, ammonia nitrogen content, total phosphorus content, total nitrogen content, conductivity, turbidity and chlorophyll A content at the previous moment.

[0016] Preferably, the preprocessing of the relevant data to obtain an ideal data set specifically includes:

[0017] S21: Remove non-numerical samples from the data set;

[0018] S22: The chlorophyll A content in each sample in the data set is evaluated using the 3σ standard to remove outliers in the sample;

[0019] S23: The Z-score method is used to standardize the data to eliminate the dimensional effects between different variables. The standardization formula is:

[0020]

[0021] In the formula, x i is the i-th variable in the original sample x, is x i The average value, σ i is x i The standard deviation of is x i The standardized value of .

[0022] Preferably, the step of calculating the average influence value based on the ideal data set and performing dimensionality reduction processing on the ideal data set specifically includes:

[0023] S31: Use the original data in the ideal data set to train a BP neural network g D (x), the ideal data set is in the form of T(T={(x i ,y i )}, where x i is the training sample, y i For test samples;

[0024] S32: Increase the value of one of the independent variables by 10% in turn, and keep other variables unchanged. After obtaining a new training sample, input it into the BP neural network for simulation to obtain the result

[0025]

[0026] In the formula, is the value of the jth variable in the ith sample, t represents the number of samples in the ideal training set, and k represents the number of variables in each sample;

[0027] S33: Reduce the value of one of the independent variables by 10% in turn, and keep other variables unchanged. After obtaining a new training sample, input it into the BP neural network for simulation to obtain the result

[0028]

[0029] S34: Calculate the difference of the results obtained for each independent variable, average them according to the number of training samples, and calculate the average influence value of the j-th independent variable on the network:

[0030]

[0031]

[0032] In the formula, Get the difference of the results for each independent variable, MIV j is the average influence of the j-th independent variable on the network;

[0033] S35: Determine the absolute value of the average impact value according to the calculation result of S34, screen out the independent variables with cumulative contribution value>90%, and complete the selection of input variables.

[0034] Preferably, the setting of the mixed training set and the test set, and setting a weight and a sample update coefficient for each sample, specifically includes:

[0035] S41: Set the initial weight of each sample in the mixed training set T

[0036]

[0037] Where n is the number of samples from the auxiliary sample set in the mixed training set, and m is the number of samples from the small sample water body in the mixed training set;

[0038] S42: Define the sample weight update coefficient β for the auxiliary sample set:

[0039]

[0040] Preferably, the step of combining the reduced-dimensional data set with the weight vector, training the basic weak regression model, and continuously updating the sample weights specifically includes:

[0041] S51: Set sample weight distribution

[0042]

[0043] In the formula, is the sample weight of the tth iteration, n is the number of samples from the auxiliary sample set in the mixed training set, and m is the number of samples from the small sample water body in the mixed training set;

[0044] S52: Multiply the sample weight distribution by the sample, and train a weak regressor f based on the multiplied data set t (x);

[0045] S53: Calculate function error

[0046]

[0047] in

[0048]

[0049] In the formula, x i is the training sample, y i For test samples;

[0050] S54: Calculate the average error e t :

[0051]

[0052] S55: Define the update coefficient β of the target sample set t :

[0053] β t =e t / (1-e t );

[0054] S56: Update sample weights:

[0055]

[0056] S57: Based on the new sample weights, a new BP neural network is obtained.

[0057] Preferably, the root mean square error is used to evaluate the performance of the model, where the root mean square error formula is:

[0058]

[0059] Where RMSE is the root mean square error, yi is the true value of chlorophyll A in the test set, is the predicted value of Chlorophyll A in the test set.

[0060] The embodiment of the present invention further provides a chlorophyll content auxiliary prediction system for a small sample water body, which is used to implement the above-mentioned chlorophyll content auxiliary prediction method for a small sample water body, specifically including:

[0061] A data collection module collects water body information under different environments, obtains relevant data of the water body, and constructs a data set, wherein the data set includes relevant data of a small sample water body and relevant data of an auxiliary sample water body, and the small sample water body and the auxiliary sample water body come from different environments;

[0062] A data preprocessing module, used for preprocessing the relevant data to obtain an ideal data set;

[0063] An average influence value calculation module, used to calculate the average influence value based on the ideal data set, and perform dimensionality reduction processing on the ideal data set;

[0064] The sample weight setting and sample coefficient updating module is used to set the mixed training set and test set according to the reduced-dimensional data set, and set the weight and sample update coefficient for each sample;

[0065] The sample weight update module is used to combine the reduced-dimensional data set with the weight vector, train the basic weak regression model, and continuously update the sample weights;

[0066] A prediction model selection module is used to continuously repeat the sample weight setting and sample coefficient updating module and the sample weight updating module, save and record each weak regression model, and finally select the regression model with the best prediction effect as the strong regression model;

[0067] The prediction module is used to input the test sample into the strong regression model to obtain the chlorophyll prediction value of the test sample.

[0068] An embodiment of the present invention also provides an electronic device, which includes a processor, a memory and a bus system, wherein the processor and the memory are connected through the bus system, the memory is used to store instructions, and the processor is used to execute the instructions stored in the memory to implement the above-mentioned auxiliary prediction method for chlorophyll content in small sample water bodies.

[0069] An embodiment of the present invention also provides a computer storage medium, which stores a computer software product. The computer software product includes several instructions for enabling a computer device to execute the above-mentioned auxiliary prediction method for chlorophyll content in a small sample water body.

[0070] It can be seen from the above technical solutions that the present invention has the following advantages:

[0071] The embodiment of the present invention provides a method and system for auxiliary prediction of chlorophyll content in small sample water bodies. The present invention makes full use of the knowledge information contained in the data of adjacent purification units, reduces the requirements for data quantity and quality in the modeling process, reduces a large amount of data collection, processing and model training work in the modeling process, and establishes a low-cost and high generalization performance chlorophyll prediction model for the chlorophyll prediction process. According to the specific conditions in the small sample water body, the present invention establishes a better prediction model with the assistance of a large sample water body. Different from the commonly used measurement methods such as photobiological reaction devices, the present invention is mainly data-driven, collects various variable information in the water body, and applies machine learning and transfer learning methods to predict chlorophyll content, which provides assistance for water quality monitoring and environmental governance. The method is suitable for biological monitoring processes with known process requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the implementation cases of the present invention or the technical solutions in the prior art, the following is a brief description of the drawings required for use in the embodiments. By referring to the drawings, the features and advantages of the present invention will be more clearly understood. The drawings are schematic and should not be understood as limiting the present invention in any way. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. Among them:

[0073] Figure 1 A flowchart of a method for auxiliary prediction of chlorophyll content in a small sample of water provided in an embodiment;

[0074] Figure 2 A brief flow chart of the method of the present invention in the embodiment;

[0075] Figure 3 A schematic diagram of the average impact value of each influencing variable in the embodiment;

[0076] Figure 4This is a schematic diagram showing the comparison of RMSE values ​​between the method of the present invention and other algorithms in the embodiments;

[0077] Figure 5 It is a block diagram of a chlorophyll content auxiliary prediction system for a small sample water body provided in an embodiment. DETAILED DESCRIPTION

[0078] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0079] Embodiment 1

[0080] like Figure 1 , Figure 2 As shown, the embodiment of the present invention proposes a method for auxiliary prediction of chlorophyll content in a small sample water body, the method comprising:

[0081] S1: Collect water body information in different environments, obtain relevant data of water bodies, and construct a data set, wherein the data set includes relevant data of small sample water bodies and relevant data of auxiliary sample water bodies, and the small sample water bodies and the auxiliary sample water bodies come from different environments;

[0082] S2: preprocessing the relevant data to obtain an ideal data set;

[0083] S3: Based on the ideal data set, calculate the average influence value, and perform dimensionality reduction processing on the ideal data set;

[0084] S4: According to the dataset after dimension reduction, set the mixed training set and test set, and set the weight and sample update coefficient for each sample;

[0085] S5: Combine the reduced-dimensional dataset and the weight vector to train the basic weak regression model and continuously update the sample weights;

[0086] S6: Repeat S4-S5 continuously, save and record each weak regression model, and finally select the regression model with the best prediction effect as the strong regression model;

[0087] S7: Input the test sample into the strong regression model to obtain the chlorophyll prediction value of the test sample.

[0088] From the above technical scheme, it can be seen that the present invention provides an auxiliary prediction method for the chlorophyll content of a small sample water body, by obtaining water body related data; preprocessing the data to obtain the ideal data set required for modeling; calculating the average impact value, and performing dimensionality reduction operation on the ideal data set; according to the reduced dimensionality data set, setting a mixed training set and a test set to prepare for training a neural network, and setting weights and sample update coefficients for each sample; combining the data set and the weight vector, training the basic weak regression model, and continuously updating the sample weights; repeatedly building the model, saving and recording each weak regression model, and finally selecting the regression model with the best effect as the strong regression model. The present invention utilizes the correlation between water environment influencing factors and chlorophyll, and between different water environments, makes full use of the knowledge information contained in the adjacent purified unit data, reduces the requirements for data quantity and quality in the modeling process, reduces a large amount of data collection, processing and model training work in the modeling process, and establishes a low-cost and high generalization performance chlorophyll prediction model for the chlorophyll prediction process.

[0089] In this embodiment, in step S1, sensors (sensor 1 and sensor 2) are used to collect water body information in different environments, obtain relevant data of the water body, and construct a data set. The data set includes relevant data (data 1) of a small sample water body (water body A) and relevant data (data 2) of an auxiliary sample water body (water body B), and the small sample water body and the auxiliary sample water body come from different environments. It should be noted that the auxiliary sample water body in this embodiment is a large sample water body, which is used to assist the establishment of a small water body sample to establish a prediction model.

[0090] Furthermore, the relevant data of the water body include the water body's temperature, pH value, dissolved oxygen content, potassium permanganate content, ammonia nitrogen content, total phosphorus content, total nitrogen content, conductivity, turbidity and chlorophyll A content at the last moment.

[0091] In this embodiment, in step S2, the relevant data is preprocessed to obtain an ideal data set, which specifically includes:

[0092] S21: Remove non-numerical samples from the data set;

[0093] S22: The chlorophyll A content in each sample in the data set is evaluated using the 3σ standard to remove outliers in the sample;

[0094] S23: The Z-score method is used to standardize the data to eliminate the dimensional effects between different variables. The standardization formula is:

[0095]

[0096] In the formula, x i is the i-th variable in the original sample x, is xi The average value, σ i is x i The standard deviation of is x i The standardized value of .

[0097] In this embodiment, in step S3, based on the ideal data set, the average influence value is calculated, and the dimensionality reduction process is performed on the ideal data set, which specifically includes:

[0098] S31: Use the original data in the ideal data set to train a BP neural network g D (x), the ideal data set is in the form of T(T={(x i ,y i )}, where x i is the training sample, y i For test samples;

[0099] S32: Increase the value of one of the independent variables by 10% in turn, and keep other variables unchanged. After obtaining a new training sample, input it into the BP neural network for simulation to obtain the result P i j :

[0100]

[0101] In the formula, is the value of the jth variable in the ith sample, t represents the number of samples in the ideal training set, and k represents the number of variables in each sample;

[0102] S33: Reduce the value of one of the independent variables by 10% in turn, and keep other variables unchanged. After obtaining a new training sample, input it into the BP neural network for simulation to obtain the result

[0103]

[0104] S34: Calculate the difference of the results obtained for each independent variable, average them according to the number of training samples, and calculate the average influence value of the j-th independent variable on the network:

[0105]

[0106]

[0107] In the formula, Get the difference of the results for each independent variable, MIV j is the average influence of the j-th independent variable on the network;

[0108] S35: Determine the absolute value of the average impact value according to the calculation result of S34, screen out the independent variables with cumulative contribution value>90%, and complete the selection of input variables.

[0109] In this embodiment, in step S4, a mixed training set and a test set are set according to the data set after dimensionality reduction, and a weight and a sample update coefficient are set for each sample.

[0110] The weight and sample update coefficient are set for each sample, including:

[0111] S41: Set the initial weight of each sample in the mixed training set T

[0112]

[0113] Where n is the number of samples from the auxiliary sample set in the mixed training set, and m is the number of samples from the small sample water body in the mixed training set;

[0114] S42: Define the sample weight update coefficient β for the auxiliary sample set:

[0115]

[0116] In this embodiment, in step S5, the reduced-dimensional data set and the weight vector are combined to train the basic weak regression model, and the sample weights are continuously updated, specifically including:

[0117] S51: Set sample weight distribution

[0118]

[0119] In the formula, is the sample weight of the tth iteration, n is the number of samples from the auxiliary sample set in the mixed training set, and m is the number of samples from the small sample water body in the mixed training set;

[0120] S52: Multiply the sample weight distribution by the sample, and train a weak regressor f based on the multiplied data set t (x);

[0121] S53: Calculate function error

[0122]

[0123] in

[0124]

[0125] In the formula, x i is the training sample, yi For test samples;

[0126] S54: Calculate the average error e t :

[0127]

[0128] S55: Define the update coefficient β of the target sample set t :

[0129] β t =e t / (1-e t );

[0130] S56: Update sample weights:

[0131]

[0132] S57: Based on the new sample weights, a new BP neural network is obtained.

[0133] In this embodiment, in step S6, S4-S5 are continuously repeated, the weak regression model of each time is saved and recorded, and finally the regression model with the best prediction effect is selected as the strong regression model.

[0134] It should be noted that, in this embodiment, both the weak regression model and the strong regression model are BP neural network models, and the only difference is the prediction accuracy.

[0135] In this embodiment, in step S7, the test sample is input into the strong regression model to obtain the chlorophyll prediction value of the test sample.

[0136] Furthermore, the present invention uses the root mean square error to evaluate the performance of the model, wherein the root mean square error formula is:

[0137]

[0138] Where RMSE is the root mean square error, yi is the true value of chlorophyll A in the test set, is the predicted value of Chlorophyll A in the test set.

[0139] In order to further illustrate the advantages of the method of the present invention, it is described below in conjunction with specific experiments.

[0140] This experiment analyzed the original samples from the National Water Quality Automatic Comprehensive Supervision Platform and obtained 851 samples from the Huaihe River Basin. Among them, the first 172 samples came from Jiangba Town and the last 679 samples came from Gaoliangjian Town. We calculated the average impact value of the relevant variables for these data to obtain the input of the established neural network. The absolute value of the average impact value of each variable is as follows Figure 3As shown. Later, in the process of setting up the mixed data set, the number of 679 samples added from the auxiliary sample water body area B in the mixed data set remained unchanged. Then we extracted 10%-80% of the samples from the 172 samples in the small sample water body area A and added them to the mixed data set, and the remaining samples not extracted from area A were used as the test set. We used the traditional BP network and the BP network without the average influence value algorithm for comparison. In order to clearly see the regression effect of transfer learning. We compared the RMSE of the three models, and the comparison effect is shown in Figure 4 shown.

[0141] Embodiment 2

[0142] like Figure 5 As shown, the present invention provides a chlorophyll content auxiliary prediction system for a small sample water body, which is used to implement the chlorophyll content auxiliary prediction method for a small sample water body in the above embodiment 1, and specifically includes:

[0143] The data collection module 10 collects water body information in different environments, obtains relevant data of the water body, and constructs a data set, wherein the data set includes relevant data of a small sample water body and relevant data of an auxiliary sample water body, and the small sample water body and the auxiliary sample water body come from different environments;

[0144] A data preprocessing module 20, used for preprocessing the relevant data to obtain an ideal data set;

[0145] An average influence value calculation module 30, used to calculate the average influence value based on the ideal data set, and perform dimensionality reduction processing on the ideal data set;

[0146] The sample weight setting and sample coefficient updating module 40 is used to set a mixed training set and a test set according to the data set after dimension reduction, and set a weight and a sample update coefficient for each sample;

[0147] A sample weight updating module 50 is used to combine the reduced-dimensional data set with the weight vector, train the basic weak regression model, and continuously update the sample weight;

[0148] The prediction model selection module 60 is used to continuously repeat the sample weight setting and sample coefficient updating module 40 and the sample weight updating module 50, save and record each weak regression model, and finally select the regression model with the best prediction effect as the strong regression model;

[0149] The prediction module 70 is used to input the test sample into the strong regression model to obtain the chlorophyll prediction value of the test sample.

[0150] The present embodiment is an auxiliary prediction system for chlorophyll content of small sample water bodies, which is used to implement the aforementioned auxiliary prediction method for chlorophyll content of small sample water bodies. Therefore, the specific implementation method of the auxiliary prediction system for chlorophyll content of small sample water bodies can be seen in the embodiment part of the auxiliary prediction method for chlorophyll content of small sample water bodies in the previous text, for example, the data acquisition module 10, the data preprocessing module 20, the average influence value calculation module 30, the sample weight setting and sample coefficient update module 40, the sample weight update module 50, the prediction model selection module 60, and the prediction module 70 are respectively used to implement the above-mentioned steps S1, S2, S3, S4, S5, S6, and S7 in the auxiliary prediction method for chlorophyll content of small sample water bodies. Therefore, its specific implementation method can refer to the description of the corresponding embodiments of each part. In order to avoid redundancy, it will not be repeated here.

[0151] Embodiment 3

[0152] An embodiment of the present invention also provides an electronic device, which includes a processor, a memory and a bus system, wherein the processor and the memory are connected through the bus system, the memory is used to store instructions, and the processor is used to execute the instructions stored in the memory to implement the above-mentioned auxiliary prediction method for chlorophyll content in small sample water bodies.

[0153] Embodiment 4

[0154] An embodiment of the present invention also provides a computer storage medium, which stores a computer software product. The computer software product includes several instructions for enabling a computer device to execute the above-mentioned auxiliary prediction method for chlorophyll content in a small sample water body.

[0155] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0156] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0157] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0158] Obviously, the above embodiments are merely examples for clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from these are still within the protection scope of the invention.

Claims

1. A method for auxiliary prediction of chlorophyll content in small sample water bodies, characterized in that: include: S1: Collect water body information in different environments, obtain relevant data of water bodies, and construct a data set, wherein the data set includes relevant data of small sample water bodies and relevant data of auxiliary sample water bodies, and the small sample water bodies and the auxiliary sample water bodies come from different environments; S2: preprocessing the relevant data to obtain an ideal data set; S3: Based on the ideal data set, calculate the average influence value, and perform dimensionality reduction processing on the ideal data set; S4: According to the reduced-dimensional data set, set the mixed training set and test set, and set the weight and sample update coefficient for each sample, including: S41: Set the initial weight of each sample in the mixed training set T Where n is the number of samples from the auxiliary sample set in the mixed training set, and m is the number of samples from the small sample water body in the mixed training set; S42: Define the sample weight update coefficient β for the auxiliary sample set: S5: Combine the reduced-dimensional dataset and the weight vector to train the basic weak regression model and continuously update the sample weights, including: S51: Set sample weight distribution In the formula, is the sample weight of the tth iteration, n is the number of samples from the auxiliary sample set in the mixed training set, and m is the number of samples from the small sample water body in the mixed training set; S52: Multiply the sample weight distribution by the sample, and train a weak regressor f based on the multiplied data set t (x); S53: Calculate function error in In the formula, x i is the training sample, y i For test samples; S54: Calculate the average error e t : S55: Define the update coefficient β of the target sample set t : β t =and t / (1-e t ); S56: Update sample weights: S57: Based on the new sample weights, a new BP neural network is obtained; S6: Repeat S4-S5 continuously, save and record each weak regression model, and finally select the regression model with the best prediction effect as the strong regression model; S7: Input the test sample into the strong regression model to obtain the chlorophyll prediction value of the test sample.

2. The chlorophyll content auxiliary prediction method for a small sample water body according to claim 1 is characterized in that: The method for collecting water body information under different environments and obtaining water body related data is as follows: Sensors are used to collect water information in different environments to obtain water-related data, including water temperature, pH value, dissolved oxygen content, potassium permanganate content, ammonia nitrogen content, total phosphorus content, total nitrogen content, conductivity, turbidity and chlorophyll A content at the previous moment.

3. The chlorophyll content auxiliary prediction method for a small sample water body according to claim 1 is characterized in that: The preprocessing of the relevant data to obtain an ideal data set specifically includes: S21: Remove non-numerical samples from the data set; S22: The chlorophyll A content in each sample in the data set is evaluated using the 3σ standard to remove outliers in the sample; S23: The Z-score method is used to standardize the data to eliminate the dimensional effects between different variables. The standardization formula is: In the formula, x i is the i-th variable in the original sample x, is x i The average value, σ i is x i The standard deviation of is x i The standardized value of .

4. The chlorophyll content auxiliary prediction method for a small sample water body according to claim 1, characterized in that: The step of calculating the average influence value based on the ideal data set and performing dimensionality reduction processing on the ideal data set specifically includes: S31: Use the original data in the ideal data set to train a BP neural network g D (x), the ideal data set is in the form of T, T = {(x i ,y i )}, where x i is the training sample, y i For test samples; S32: Increase the value of one of the independent variables by 10% in turn, and keep other variables unchanged. After obtaining a new training sample, input it into the BP neural network for simulation to obtain the result In the formula, is the value of the jth variable in the ith sample, t represents the number of samples in the ideal training set, and k represents the number of variables in each sample; S33: Reduce the value of one of the independent variables by 10% in turn, and keep other variables unchanged. After obtaining a new training sample, input it into the BP neural network for simulation to obtain the result S34: Calculate the difference of the results obtained for each independent variable, average them according to the number of training samples, and calculate the average influence value of the j-th independent variable on the network: In the formula, Get the difference of the results for each independent variable, MIV j is the average influence of the j-th independent variable on the network; S35: Determine the absolute value of the average impact value based on the calculation result of S34, screen out the independent variables with cumulative contribution value>90%, and complete the selection of input variables.

5. The chlorophyll content auxiliary prediction method for a small sample water body according to claim 1, characterized in that: The root mean square error is used to evaluate the performance of the model, where the root mean square error formula is: In the formula, RMSE is the root mean square error, y i is the true value of chlorophyll A in the test set, is the predicted value of Chlorophyll A in the test set.

6. A chlorophyll content auxiliary prediction system for small sample water bodies, characterized in that: The system is used to implement the auxiliary prediction method for chlorophyll content in a small sample water body according to any one of claims 1 to 5, specifically comprising: A data collection module collects water body information under different environments, obtains relevant data of the water body, and constructs a data set, wherein the data set includes relevant data of a small sample water body and relevant data of an auxiliary sample water body, and the small sample water body and the auxiliary sample water body come from different environments; A data preprocessing module, used for preprocessing the relevant data to obtain an ideal data set; An average influence value calculation module, used to calculate the average influence value based on the ideal data set, and perform dimensionality reduction processing on the ideal data set; The sample weight setting and sample coefficient updating module is used to set the mixed training set and test set according to the reduced-dimensional data set, and set the weight and sample update coefficient for each sample; The sample weight update module is used to combine the reduced-dimensional data set with the weight vector, train the basic weak regression model, and continuously update the sample weights; A prediction model selection module is used to continuously repeat the sample weight setting and sample coefficient updating module and the sample weight updating module, save and record each weak regression model, and finally select the regression model with the best prediction effect as the strong regression model; The prediction module is used to input the test sample into the strong regression model to obtain the chlorophyll prediction value of the test sample.

7. An electronic device, characterized in that: The electronic device includes a processor, a memory and a bus system, the processor and the memory are connected through the bus system, the memory is used to store instructions, and the processor is used to execute the instructions stored in the memory to implement the auxiliary prediction method for chlorophyll content in small sample water bodies as described in any one of claims 1 to 5.

8. A computer storage medium, characterized in that: The computer storage medium stores a computer software product, and the computer software product includes a number of instructions for enabling a computer device to execute the chlorophyll content auxiliary prediction method for a small sample water body as described in any one of claims 1 to 5.

Citation Information

Patent Citations

  • Method for predicting chlorophyll a concentration in water based on BP nerval net

    CN101158674A

  • A perturbation-based chlorophyll a content related factor sensitivity analysis method

    CN109934334A

  • Algae bloom prediction method based on principal component analysis and BP neural network

    CN110046645A

  • Method and device for predicting chlorophyll concentration of water body based on multispectral image

    CN116385820A

  • Reduction of Computation Complexity of Neural Network Sensitivity Analysis

    US20150081606A1