Water Quality Prediction System Based on Machine Learning Algorithm

Abstract

The present invention relates to the technical field of water quality prediction, and specifically, to a water quality prediction system based on a machine learning algorithm. It includes a detection and judgment unit, a comparison and prediction unit, and a predicted water quality unit. The predicted water quality unit of the present invention is used to receive the cyanobacteria concentration trend data, physical indicators, and chemical indicator data in the data judgment module, receive the predicted pH value trend data in the trend prediction module and the water source data obtained in the detection data module, and input the cyanobacteria concentration trend data, physical indicator data, chemical indicator data, and predicted pH value trend data into a multi-source regression model for learning, and output the intercept term and the deviation term. At the same time, the multi-source regression model predicts the water quality at a certain time point according to the learning results. By predicting future water quality changes, potential pollution problems can be identified in advance, so as to take preventive measures, reduce pollution sources or increase water quality treatment, thereby improving the efficiency of water quality management.

Description

Technical Field

[0001] The present invention relates to the technical field of water quality prediction, and more specifically, to a water quality prediction system based on machine learning algorithms. Background Art

[0002] Water quality prediction is a data-driven method that analyzes and processes historical water quality data to predict future water quality changes, which is used to monitor the health of water bodies, warn of water quality deterioration events, and support water resource management decisions. Since there are cyanobacteria in water sources, cyanobacteria can produce toxins that are harmful to humans, fish, and other aquatic organisms, resulting in a decrease in the oxygen content of the water body. When cyanobacteria reproduce rapidly, it can cause algal blooms in the water body, causing odors and color changes in the water source. Since the changes in water quality indicators cannot be accurately predicted based on the concentration trend of cyanobacteria, the water quality deteriorates. Therefore, we provide a water quality prediction system based on machine learning algorithms. Summary of the Invention

[0003] The purpose of the present invention is to provide a water quality prediction system based on machine learning algorithms to solve the problems raised in the above background art.

[0004] To achieve the above purpose, the present invention provides a water quality prediction system based on machine learning algorithms, including a detection and judgment unit, a comparison and prediction unit, and a predicted water quality unit;

[0005] The detection and judgment unit obtains the water source data and historical data of the water quality to be predicted, uses a fluorescence sensor to detect the concentration data of cyanobacteria in the water source, and then predicts the predicted value of the cyanobacteria concentration trend based on the cyanobacteria concentration data, and uses the predicted value of the cyanobacteria concentration trend to judge the growth trend of cyanobacteria with a set standard threshold;

[0006] The comparison and prediction unit is used to receive the data in the detection and judgment unit and detect the concentrations of bicarbonate ions and carbonate ions, then calculate the PH value based on the detected concentrations of bicarbonate ions and carbonate ions, judge whether the growth of cyanobacteria has caused an increase in the PH value based on the calculated PH value, and then predict the trend of the PH value based on historical data using a linear regression formula;

[0007] The predicted water quality unit is used to receive the data in the detection and judgment unit and the comparison and prediction unit, and predict the water quality at a certain time point based on the cyanobacteria concentration trend data in the detection and judgment unit and the predicted PH value trend data in the comparison and prediction unit.

[0008] As a further improvement of this technical solution, the detection and judgment unit includes a detection data module and a data judgment module;

[0009] The detection data module obtains the water source data and historical data of the water quality to be predicted, and obtains the environmental factor data around the water source through sensors. The environmental factor data includes water temperature and light. A fluorescence sensor is used to detect the presence of cyanobacteria in the water source and the concentration data of cyanobacteria;

[0010] The historical data includes historical pH values;

[0011] The data judgment module is used to receive the environmental factor data and cyanobacteria concentration data in the detection data module, detect the physical and chemical indicators in the water quality through various instruments, and use a decision tree to predict the predicted value of the cyanobacteria concentration trend based on the environmental factor data, cyanobacteria concentration data, physical indicators and chemical indicator data. Then, the predicted value of the cyanobacteria concentration trend is used to judge the cyanobacteria growth trend with the set standard threshold.

[0012] As a further improvement of this technical solution, the specific judgment situations in the data judgment module include:

[0013] Situation ①: When the predicted value of the cyanobacteria concentration trend is greater than the set standard threshold, it indicates that the environmental factors, physical indicators and chemical indicator data promote the growth of cyanobacteria, causing the cyanobacteria concentration trend to rise. The command of the rising cyanobacteria concentration trend is transmitted to the data comparison module;

[0014] Situation ②: When the predicted value of the cyanobacteria concentration trend is less than the set standard threshold, it indicates that the cyanobacteria concentration trend is decreasing.

[0015] As a further improvement of this technical solution, the comparison and prediction unit includes a data comparison module and a trend prediction module;

[0016] The data comparison module is used to receive the command data of the rising cyanobacteria concentration trend in the data judgment module. The data comparison module obtains the water source data from the detection data module, detects the concentrations of bicarbonate ions and carbonate ions in the obtained water source data, calculates the pH value based on the detected concentrations of bicarbonate ions and carbonate ions, and compares the calculated pH value with the set standard pH value;

[0017] The trend prediction module is used to receive the command data of the rising pH value and the calculated pH value in the data comparison module. When the trend prediction module receives the command data of the rising pH value, the trend prediction module obtains the historical pH value in the historical data from the detection data module, and updates the calculated pH value and the historical pH value into complete historical pH value data. The complete historical pH value data is input into a linear regression model. The linear regression model outputs the corresponding linear regression coefficients and the error terms generated by the linear regression model according to the different pH values in the complete historical pH value data, and predicts the trend of the pH value using the linear regression formula based on the complete historical pH value data, the corresponding regression coefficients and the error terms.

[0018] As a further improvement of this technical solution, the specific comparison situations in the data comparison module include:

[0019] Situation 1: When the calculated pH value is greater than the set standard pH value, it indicates that the growth of cyanobacteria has caused the pH value to increase. The command data of the increased pH value is transmitted to the trend prediction module;

[0020] Situation 2: When the calculated pH value is less than the set standard pH value, it indicates that the growth of cyanobacteria has not caused the pH value to increase.

[0021] As a further improvement of this technical solution, the implementation principle of using the linear regression formula to predict the pH value trend in the trend prediction module:

[0022] Collect the complete historical pH value data WPH, the corresponding regression coefficient β, and the error term ε to predict the pH value trend, and obtain the predicted pH value trend data QS PH , and the specific algorithm formula:

[0023] QS PH = PH0+(PH1×β1)+(PH2×β2)+...+(PH n ×β n )+ε;

[0024] Among them, PH0 is the constant term, PH1 refers to the first pH value data in the complete historical pH value data, PH n refers to the nth pH value data in the complete historical pH value data, β1 refers to the linear regression coefficient corresponding to the first pH value data, and β n refers to the linear regression coefficient corresponding to the nth pH value data.

[0025] As a further improvement of this technical solution, the predicted water quality unit is used to receive the cyanobacteria concentration trend data, physical indicators, and chemical indicator data in the data judgment module, receive the predicted pH value trend data in the trend prediction module and the water source data obtained in the detection data module, and input the cyanobacteria concentration trend data, physical indicator data, chemical indicator data, and predicted pH value trend data into the multi-source regression model for learning, and output the intercept term and deviation term. At the same time, the multi-source regression model predicts the water quality at a certain time point according to the learning results.

[0026] As a further improvement of this technical solution, the implementation steps of using the multi-source regression model to predict the water quality at a certain time point in the predicted water quality unit:

[0027] Step 1: Collect the cyanobacteria concentration trend data Nd qs, physical index data Wz, chemical index data Hz, obtained water source data, and predicted pH value trend data QS PH Input into the multi-source regression model. The multi-source regression model learns the input data and extracts key feature data. Using the multi-source regression model, based on the extracted key feature data Gj tz Obtain the impact of the extracted key features on the water source data

[0028] Step 2: The multi-source regression model calculates the lag value of the extracted key features and the average value δ of the extracted key features within a time window i , and obtain the impact γ of the lag value of the extracted key features on the water source data i ;

[0029] Step 3: Use the impact of the extracted key features on the water source data The impact γ of the lag value of the extracted key features on the water source data i , the average value δ of the extracted key features within a time window i , intercept term α0, and bias term θ to predict the water quality at a certain time point, and obtain the predicted water quality W at a certain time point t , the specific algorithm formula:

[0030]

[0031] Among them, t refers to the time point, and m refers to the number of extracted key features

[0032] Compared with the prior art, the beneficial effects of the present invention are:

[0033] In the water quality prediction system based on the machine learning algorithm, the predicted water quality unit is used to receive the cyanobacteria concentration trend data, physical indicators, and chemical indicator data in the data judgment module, receive the predicted pH value trend data in the trend prediction module, and the obtained water source data in the detection data module, and input the cyanobacteria concentration trend data, physical index data, chemical index data, and predicted pH value trend data into the multi-source regression model for learning, and output the intercept term and the bias term. At the same time, the multi-source regression model predicts the water quality at a certain time point according to the learning results. By predicting the future water quality changes, possible pollution problems can be identified in advance, so as to take preventive measures, reduce pollution sources or increase water quality treatment, and improve the water quality management efficiency Description of the Drawings

[0034] Figure 1 is the overall block diagram of the present invention;

[0035] Figure 2 is the block diagram of the detection and judgment unit of the present invention;

[0036] Figure 3 This is the block diagram of the comparison and prediction unit of the present invention.

[0037] The meanings of the various labels in the figure are as follows:

[0038] 1. Detection and judgment unit; 11. Detection data module; 12. Data judgment module;

[0039] 2. Comparison and prediction unit; 21. Data comparison module; 22. Trend prediction module; 3. Predicted water quality unit. Specific implementation mode

[0040] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of the present invention.

[0041] Embodiment 1

[0042] The present invention provides a water quality prediction system based on a machine learning algorithm. Please refer to Figures 1 - 3 , including a detection and judgment unit 1, a comparison and prediction unit 2, and a predicted water quality unit 3;

[0043] The detection and judgment unit 1 obtains the water source data and historical data of the water quality to be predicted, uses a fluorescence sensor to detect the concentration data of cyanobacteria in the water source, and then predicts the predicted value of the cyanobacteria concentration trend based on the cyanobacteria concentration data, and uses the predicted value of the cyanobacteria concentration trend to judge the cyanobacteria growth trend with a set standard threshold; the comparison and prediction unit 2 is used to receive the data in the detection and judgment unit 1 and detect the concentrations of bicarbonate ions and carbonate ions, then calculate the PH value according to the detected concentrations of bicarbonate ions and carbonate ions, judge that the growth of cyanobacteria has caused the PH value to increase according to the calculated PH value, and then predict the trend of the PH value according to the historical data using a linear regression formula; the predicted water quality unit 3 is used to receive the data in the detection and judgment unit 1 and the comparison and prediction unit 2, and predict the water quality at a certain time point based on the cyanobacteria concentration trend data in the detection and judgment unit 1 and the predicted PH value trend data in the comparison and prediction unit 2.

[0044] The following is a refinement of the above units. Please refer to Figures 1 - 3 ;

[0045] The detection and judgment unit 1 includes a detection data module 11 and a data judgment module 12;

[0046] The detection data module 11 obtains the water source data and historical data of the water quality to be predicted, and obtains the environmental factor data around the water source through sensors. The environmental factor data includes water temperature and light. A fluorescence sensor is used to detect the presence of cyanobacteria in the water source and the concentration data of cyanobacteria. The fluorescence sensor uses specific fluorescent substances (such as chlorophyll a) contained in cyanobacterial cells to detect cyanobacteria. Chlorophyll a is the main photosynthetic pigment of cyanobacteria and many other algae, and it can emit fluorescence under the irradiation of light with a specific wavelength. The fluorescence sensor judges the presence and corresponding concentration of cyanobacteria in the water sample by detecting these fluorescence signals;

[0047] The historical data includes historical pH values;

[0048] The data judgment module 12 is used to receive the environmental factor data and cyanobacteria concentration data in the detection data module 11, and detect the physical indexes and chemical indexes in the water quality through various instruments. The physical indexes include chromaticity and turbidity data, and the chemical indexes include nitrogen, phosphorus concentration and dissolved oxygen data. Then, a decision tree is used to predict the predicted value of the cyanobacteria concentration trend based on the environmental factor data, cyanobacteria concentration data, physical indexes and chemical index data. Next, the predicted value of the cyanobacteria concentration trend is used to judge the growth trend of cyanobacteria with a set standard threshold. The outbreak of cyanobacteria may lead to eutrophication of water bodies and the release of harmful substances. Predicting the trend and combining the threshold judgment can help take preventive measures before the cyanobacteria concentration is too high, such as controlling pollutant emissions or increasing water body replacement;

[0049] Steps for realizing the predicted value of the cyanobacteria concentration trend using a decision tree:

[0050] Step ①: First, obtain the environmental factor data Hj, cyanobacteria concentration data Ln, physical index data Wz and chemical index data Hz, and perform missing value and outlier processing. Then, extract the features of the processed data;

[0051] Step ②: Integrate the extracted feature data into a data set through a data fusion tool, and divide the data set into a training set and a test set (the common ratio is 80% training set and 20% test set) for model training and evaluation;

[0052] Step ③: Use the training set data to train the decision tree regression model. The decision tree regression model will learn based on the extracted feature data and output the predicted value of the cyanobacteria concentration trend and the cyanobacteria concentration trend data;

[0053] Specific judgment situations include:

[0054] Situation ①: When the predicted value of the cyanobacteria concentration trend is greater than the set standard threshold, it indicates that the environmental factors, physical indexes and chemical index data promote the growth of cyanobacteria, causing the cyanobacteria concentration trend to rise. The command of the rising cyanobacteria concentration trend is transmitted to the data comparison module 21;

[0055] Case ②: When the predicted value of the cyanobacteria concentration trend is less than the set standard threshold, it indicates that the cyanobacteria concentration trend is decreasing;

[0056] The comparison and prediction unit 2 includes a data comparison module 21 and a trend prediction module 22;

[0057] The data comparison module 21 is used to receive the command data indicating an increase in the cyanobacteria concentration trend in the data judgment module 12. The data comparison module 21 obtains water source data from the detection data module 11, detects the concentrations of bicarbonate ions and carbonate ions in the obtained water source data, calculates the pH value based on the detected concentrations of bicarbonate ions and carbonate ions, and compares the calculated pH value with the set standard pH value;

[0058] The implementation principle of calculating the pH value:

[0059] First, collect the detected concentration of bicarbonate ions TSq and the detected concentration of carbonate ions Ts to calculate the pH value, and obtain the calculated pH value PHz. The specific algorithm formula:

[0060]

[0061] where pK a is the acid dissociation constant of bicarbonate ions;

[0062] The specific comparison situations include:

[0063] Situation 1: When the calculated pH value is greater than the set standard pH value, it indicates that the growth of cyanobacteria has promoted an increase in the pH value, and the command data indicating an increase in the pH value is transmitted to the trend prediction module 22;

[0064] Situation 2: When the calculated pH value is less than the set standard pH value, it indicates that the growth of cyanobacteria has not promoted an increase in the pH value;

[0065] The trend prediction module 22 is used to receive the command data of the rising pH value and the calculated pH value in the data comparison module 21. When the trend prediction module 22 receives the command data of the rising pH value, the trend prediction module 22 obtains the historical pH value in the historical data from the detection data module 11, and updates the calculated pH value and the historical pH value into complete historical pH value data. The complete historical pH value data is input into the linear regression model. The linear regression model outputs the corresponding linear regression coefficients and the error terms generated by the linear regression model according to different pH values in the complete historical pH value data, and uses the linear regression formula to predict the trend of the pH value based on the complete historical pH value data, the corresponding regression coefficients and the error terms. By predicting the trend of the pH value, the long-term trend of water quality change can be identified. For example, finding a long-term downward or upward trend in the pH value may indicate a systematic change in the acidity or alkalinity of the water body, which may be related to factors such as pollution sources and industrial activities;

[0066] The implementation principle of predicting the trend of the pH value using the linear regression formula:

[0067] Collect the complete historical pH value data WPH, the corresponding regression coefficients β and the error terms ε to predict the trend of the pH value, and obtain the predicted pH value trend data QS PH , and the specific algorithm formula:

[0068] QS PH = PH0+(PH1×β1)+(PH2×β2)+...+(PH n ×β n );

[0069] Among them, PH0 is the constant term, PH1 refers to the first pH value data in the complete historical pH value data, PH n refers to the nth pH value data in the complete historical pH value data, β1 refers to the linear regression coefficient corresponding to the first pH value data, β n refers to the linear regression coefficient corresponding to the nth pH value data;

[0070] The water quality prediction unit 3 is used to receive the cyanobacteria concentration trend data, physical index and chemical index data in the data judgment module 12, receive the predicted pH value trend data in the trend prediction module 22 and the water source data obtained from the detection data module 11, and input the cyanobacteria concentration trend data, physical index data, chemical index data, and predicted pH value trend data into the multi-source regression model for learning, and output the intercept term and the deviation term. At the same time, the multi-source regression model predicts the water quality at a certain time point according to the learning results. By predicting future water quality changes, possible pollution problems can be identified in advance, so as to take preventive measures. For example, if it is predicted that the water quality index may deteriorate at a certain moment, measures can be taken in advance to reduce pollution sources or increase water quality treatment;

[0071] Implementation steps for predicting water quality at a certain time point using a multi-source regression model:

[0072] Step 1: Collect the cyanobacteria concentration trend data Nd qs , physical index data Wz, chemical index data Hz, obtained water source data, and predicted pH value trend data QS PH Input them into the multi-source regression model. The multi-source regression model learns the input data and extracts key feature data. Based on the extracted key feature data Gj tz obtain the influence of the extracted key features on the water source data

[0073] Step 2: The multi-source regression model calculates the lag value of the extracted key features and the average value δ of the extracted key features within a time window based on the extracted key features i , and obtain the influence γ of the lag value of the extracted key features on the water source data i ;

[0074] Step 3: Use the influence of the extracted key features on the water source data the influence γ of the lag value of the extracted key features on the water source data i , the average value δ of the extracted key features within a time window i , intercept term α0, and bias term θ to predict the water quality at a certain time point, and obtain the predicted water quality W at a certain time point t , the specific algorithm formula:

[0075]

[0076] where t refers to the time point and m refers to the number of extracted key features. This formula is used to predict the water quality at a certain time point. By predicting water quality changes, the treatment process can be adjusted according to the expected water quality load, thereby improving treatment efficiency and reducing costs.

[0077] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art of this industry should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions in the specification are only preferred examples of the present invention and are not used to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A water quality prediction system based on machine learning algorithms, characterized in that: It includes a detection and judgment unit (1), a comparison and prediction unit (2), and a predicted water quality unit (3); The detection and judgment unit (1) obtains the water source data and historical data of the water quality to be predicted, uses a fluorescence sensor to detect the concentration data of cyanobacteria in the water source, and then predicts the predicted value of the cyanobacteria concentration trend based on the cyanobacteria concentration data, and uses the predicted value of the cyanobacteria concentration trend and the set standard threshold to judge the growth trend of cyanobacteria; The comparison and prediction unit (2) is used to receive the data in the detection and judgment unit (1) and detect the concentrations of bicarbonate ions and carbonate ions, then calculate the pH value according to the detected concentrations of bicarbonate ions and carbonate ions, judge that the growth of cyanobacteria has promoted the increase of the pH value according to the calculated pH value, and then use the linear regression formula to predict the trend of the pH value based on historical data; The predicted water quality unit (3) is used to receive the data in the detection and judgment unit (1) and the comparison and prediction unit (2), and predict the water quality at a certain time point according to the cyanobacteria concentration trend data in the detection and judgment unit (1) and the predicted pH value trend data in the comparison and prediction unit (2); The steps for the predicted water quality unit (3) to predict the water quality at a certain time point using a multi-source regression model are as follows: Step 1: Collect the cyanobacteria concentration trend data Nd qs , physical index data Wz, chemical index data Hz, obtained water source data, and predicted pH value trend data QS PH Input them into the multi-source regression model. The multi-source regression model learns the input data and extracts key feature data. According to the extracted key feature data Gj tz Obtain the impact of the extracted key features on the water source data Step 2: The multi-source regression model calculates the lag value of the extracted key features and the average value δ of the extracted key features within a time window, and obtains the influence γ of the lag value of the extracted key features on the water source data. i i ; Step 3. Influence of the extracted key features on water source data Influence γ of the lag value of the extracted key features on water source data i , average value δ of the extracted key features within a time window i , intercept term α0 and deviation term θ are used to predict the water quality at a certain time point, and the predicted water quality W at a certain time point is obtained t , specific algorithm formula: Among them, t refers to the time point, and m refers to the number of key features extracted.

2. The water quality prediction system based on a machine learning algorithm according to claim 1, wherein: The detection and judgment unit (1) includes a detection data module (11) and a data judgment module (12); The detection data module (11) obtains the water source data and historical data of the water quality to be predicted, and obtains the environmental factor data around the water source through sensors. The environmental factor data includes water temperature and light, and uses a fluorescence sensor to detect the presence of cyanobacteria in the water source and the concentration data of cyanobacteria; The historical data includes historical pH values; The data judgment module (12) is used to receive the environmental factor data and cyanobacteria concentration data in the detection data module (11), detect the physical and chemical indexes in the water quality through various instruments, and use a decision tree to predict the predicted value of the cyanobacteria concentration trend based on the environmental factor data, cyanobacteria concentration data, physical and chemical index data, and then use the predicted value of the cyanobacteria concentration trend and the set standard threshold to judge the growth trend of cyanobacteria.

3. The water quality prediction system based on machine learning algorithm according to claim 2, characterized in that: The specific judgment situations in the data judgment module (12) are as follows: Situation ①: When the predicted value of the cyanobacteria concentration trend is greater than the set standard threshold, it indicates that the environmental factors, physical and chemical index data promote the growth of cyanobacteria, causing the cyanobacteria concentration trend to rise, and the command of the rising cyanobacteria concentration trend is transmitted to the data comparison module (21); Situation ②: When the predicted value of the cyanobacteria concentration trend is less than the set standard threshold, it indicates that the cyanobacteria concentration trend is decreasing.

4. The water quality prediction system based on a machine learning algorithm according to claim 2, characterized in that: The comparison and prediction unit (2) includes a data comparison module (21) and a trend prediction module (22); The data comparison module (21) is used to receive the command data of the rising trend of cyanobacteria concentration in the data judgment module (12). The data comparison module (21) obtains water source data from the detection data module (11), detects the concentrations of bicarbonate ions and carbonate ions in the obtained water source data, calculates the pH value based on the detected concentrations of bicarbonate ions and carbonate ions, and compares the calculated pH value with the set standard pH value; The trend prediction module (22) is used to receive the command data of the rising pH value and the calculated pH value in the data comparison module (21). When the trend prediction module (22) receives the command data of the rising pH value, the trend prediction module (22) obtains the historical pH value in the historical data from the detection data module (11), updates the calculated pH value and the historical pH value into complete historical pH value data, inputs the complete historical pH value data into a linear regression model. The linear regression model outputs the corresponding linear regression coefficients and the error terms generated by the linear regression model according to different pH values in the complete historical pH value data, and predicts the trend of the pH value using the linear regression formula based on the complete historical pH value data, the corresponding regression coefficients and the error terms.

5. The water quality prediction system based on a machine learning algorithm according to claim 4, wherein: The specific comparison situations in the data comparison module (21) include: Situation 1: When the calculated pH value is greater than the set standard pH value, it indicates that the growth of cyanobacteria has promoted the increase of the pH value, and the command data of the rising pH value is transmitted to the trend prediction module (22); Situation 2: When the calculated pH value is less than the set standard pH value, it indicates that the growth of cyanobacteria has not promoted the increase of the pH value.

6. The water quality prediction system based on a machine learning algorithm according to claim 4, characterized in that: The implementation principle of predicting the trend of the pH value using the linear regression formula in the trend prediction module (22): Collect complete historical pH value data WPH, corresponding regression coefficients β, and error terms ε to predict the trend of the pH value, and obtain the predicted pH value trend data QS PH , and the specific algorithm formula is: QS PH = PH0 + (PH1 × β1) + (PH2 × β2) +... + (PH n × β n ) + ε; Among them, PH0 is the constant term, PH1 refers to the first PH value data in the complete historical PH value data, and PH n refers to the nth PH value data in the complete historical PH value data, β1 refers to the linear regression coefficient corresponding to the first PH value data, and β n refers to the linear regression coefficient corresponding to the nth PH value data.

7. The water quality prediction system based on machine learning algorithm according to claim 4, characterized in that: The predicted water quality unit (3) is used to receive the cyanobacteria concentration trend data, physical index and chemical index data in the data judgment module (12), receive the predicted pH value trend data in the trend prediction module (22) and the water source data obtained from the detection data module (11), and input the cyanobacteria concentration trend data, physical index data, chemical index data, and predicted pH value trend data into a multi-source regression model for learning, and output the intercept term and the deviation term. At the same time, the multi-source regression model predicts the water quality at a certain time point according to the learning results.

Images