Integrated water quality monitoring treatment method and system

Abstract

The invention discloses an integrated water quality monitoring processing method and system, and the method comprises the following steps: setting a high-resolution camera, collecting water storage pool solution image data with a timestamp in a preset direction, and carrying out the preprocessing of an image; and S2, extracting image features of the data acquired in the step S1 by adopting a pre-trained neural network, and transmitting the image features to a long-short-term memory network to obtain a turbidity prediction value output by the model. The system and the method have remarkable advantages in the aspects of accuracy, intelligent level, flexibility and the like of water quality monitoring treatment.

Description

Technical Field

[0001] The present invention relates to the technical field of water quality monitoring, and more specifically, to a method and system for integrated water quality monitoring and processing. Background Art

[0002] With the continuous global attention to water resources and the increasing demand for the protection of the natural ecological environment, the protection and management of water resources in mountainous areas has become particularly urgent. However, mountain waters face multiple environmental pressures, including complex terrain, soil erosion, human activities and other factors, which may threaten water quality, further emphasizing the urgent need for water quality monitoring and treatment systems in mountainous areas. Among them, turbidity is an important physical property that describes water bodies and an important parameter for water quality testing. The size of turbidity is inversely proportional to the quality of water. Reducing turbidity is conducive to improving water quality and reducing the content of bacteria, viruses and other harmful substances in water. Therefore, water turbidity detection is particularly important and has important significance for natural environmental protection and human health and safety.

[0003] At present, the turbidity detection technology on the market usually adopts two testing technologies: laboratory chemical analysis and traditional online measuring instruments, but there are also certain shortcomings. Among them, laboratory chemical analysis detection requires the use of more reagents and equipment, and requires a high level of professional skills of the operator. In addition, this method requires the water sample to be brought back to the laboratory for analysis, which limits the timeliness of the measurement results and makes it impossible to achieve real-time monitoring; and traditional online instruments are expensive, complex in structure, and have high spare parts and maintenance costs. Therefore, although these two turbidity detection technologies already exist on the market, they still cannot fully meet the needs of users.

[0004] Therefore, improving the accuracy and intelligence of water quality monitoring in mountainous areas has become a problem that still needs to be addressed. This will not only help to better understand and protect the ecosystem of mountainous waters, but also provide technical support for scientific and effective water quality management. Summary of the invention

[0005] The purpose of the present invention is to overcome the above-mentioned deficiencies of the prior art and to provide a method and system for integrated water quality monitoring and processing.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] The present invention discloses a method for integrated water quality monitoring and processing, comprising the following steps:

[0008] S1: Setting a high-resolution camera to collect image data of the water tank solution with a time stamp in a preset direction and preprocessing the image;

[0009] S2: using a pre-trained neural network to extract the image features of the data collected in step S1, passing the image features into the long short-term memory network, and obtaining the turbidity prediction value output by the model;

[0010] S3: Setting an immersion turbidity meter in the water storage tank, selecting a corresponding turbidity detection method according to the turbidity value predicted in step S2, and realizing wide-range and accurate measurement of turbidity;

[0011] S4: using the turbidity prediction value obtained in step S2 to establish a system state equation, combining the measured value of the turbidity meter in step S3 to establish a system observation equation, and realizing accurate prediction of turbidity through Kalman filtering algorithm;

[0012] S5: The current timestamp and turbidity value are stored, and whether to maintain the operation of the water pumping system is selected according to the turbidity value. If the turbidity exceeds the preset threshold, the pool solution is processed; if the turbidity does not exceed the preset threshold, no processing is performed.

[0013] Furthermore, step S1 includes the following steps:

[0014] S11: Setting a high-resolution camera in a preset direction of the water reservoir and initializing it, while collecting images of the water reservoir solution;

[0015] S12: For each channel of the image, noise is removed by performing a blur operation using a [5×5] Gaussian kernel given by a standard normal distribution;

[0016] S13: Standardize the data by the mean and standard deviation of the standard normal distribution;

[0017] S14: Normalize the values.

[0018] Further, step S2 includes the following steps:

[0019] S21: pass the preprocessed image sequence to the convolutional layer of the neural network, where the convolution kernel will slide on the image and perform a convolution operation on each position to extract local features;

[0020] S22: Activate the feature map after the convolution operation through an activation function to introduce nonlinearity;

[0021] S23: The feature map after the activation function will be passed to the pooling layer, where the pooling operation will downsample each feature map to reduce the dimension of the feature map and retain the most significant features;

[0022] S24: Flatten the feature map after the pooling operation into a one-dimensional vector, and pass it as input to the fully connected layer of the pre-trained neural network, so as to obtain the features of the image sequence;

[0023] S25: The image feature sequence is passed to the LSTM network. The hidden layer of the LSTM network receives and processes the input sequence, and the hidden state and memory unit are updated at each time step.

[0024] S26: After being processed by the hidden layer, the LSTM network outputs a hidden state sequence, which contains information for modeling the input sequence;

[0025] S27: The hidden state sequence is used as the input of the output layer of the LSTM network, and the output layer generates a predicted sequence of turbidity values ​​based on the hidden state sequence.

[0026] Further, step S3 includes the following steps:

[0027] S31: Setting an immersion turbidity meter in the water storage tank and initializing it;

[0028] S32: If the turbidity value is within the range of 0 to 400 NTU, it is determined to be low turbidity; if the turbidity value is within the range of 400 to 1000 NTU, it is determined to be high turbidity;

[0029] S33: According to the result determined in step S32, a corresponding measurement method is selected for the turbidity meter. If the turbidity is low, a scattering method is used to further accurately measure the turbidity; if the turbidity is high, a transmission method is used to further accurately measure the turbidity.

[0030] Further, step S4 includes the following steps:

[0031] S41: Taking the turbidity value predicted in step S2 as the estimated value x of the system state k , establish the system state equation:

[0032] x k =f(x k-1 )

[0033] Where f(x) represents the LSTM network fitting function;

[0034] S42: The turbidity value measured by the turbidity meter in step S3 is used as the observed value z k Establish the observation equation:

[0035] z k =Hx k +v k

[0036] Among them, H represents the observation matrix, v k represents the observation noise;

[0037] S43: Initialize the estimated value of the system state at the initial moment and the state covariance matrix P0;

[0038] S44: At each time step k, the estimated value of the system state at the next moment is predicted based on the system state equation and the state covariance matrix

[0039] S45: At each time step k, the measurement residual at the current moment is calculated according to the observation equation:

[0040]

[0041] S46: At each time step k, the Kalman gain is calculated based on the system prediction residual and the system prediction covariance matrix:

[0042]

[0043] Where R is the covariance matrix of observation noise;

[0044] S47: At each time step k, update the estimated value of the system state and the state covariance matrix to obtain the optimal estimated value at the current moment and the state covariance matrix

[0045]

[0046]

[0047] S48: Return to step S44 and repeat the prediction and update until all observation data have been processed.

[0048] Further, step S5 includes the following steps:

[0049] S51: The system records the current timestamp and the corresponding turbidity value, and records this information into the system database or log for subsequent analysis and monitoring;

[0050] S52: Decision rules are formulated based on the preset water quality turbidity standard. If the turbidity value exceeds a certain threshold, the water quality is considered unqualified, the water pumping system is shut down, and the water storage tank is treated. If the turbidity value is lower than the threshold, the water pumping system is maintained in operation, so that clean water that meets the standard is lifted to the required height for water supply.

[0051] S53: Provides remote access and management interface, system administrators can view and manage recorded data anytime and anywhere.

[0052] A system integrating water quality monitoring and processing, comprising:

[0053] An image acquisition module is used to set a high-resolution camera to acquire image data of the water reservoir solution with a time stamp in a preset direction and pre-process the image;

[0054] The turbidity prediction module uses a pre-trained neural network to extract the image features of the collected data, and then passes the image features to the long short-term memory network to obtain the turbidity prediction value output by the model;

[0055] The turbidity detection module is used to set up an immersion turbidity meter in the water storage tank, select the corresponding turbidity detection method according to the turbidity value predicted by the turbidity prediction module, and further realize the wide range and accurate measurement of turbidity;

[0056] Data fusion module, establishes the system state equation, combines the measured values ​​of the turbidity meter to establish the system observation equation, and realizes the turbidity prediction through the Kalman filter algorithm;

[0057] The water supply module is used to store the current timestamp and turbidity value, and choose whether to maintain the operation of the water extraction system according to the turbidity value. If the turbidity exceeds the threshold, the pool solution is processed.

[0058] The beneficial effects of the present invention are: according to the preliminary turbidity, the corresponding turbidity detection method is adaptively selected to have a more accurate turbidity measurement value, and on this basis, a system observation equation is established, and based on the Kalman filtering technology, the fusion of state and observation is realized to obtain a more accurate turbidity measurement result, thereby realizing the intelligent management of water quality monitoring and treatment and the continuous improvement of water supply quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a schematic diagram of the overall process of the method for water quality monitoring and treatment in this embodiment;

[0060] Figure 2 This is a schematic diagram of the process of achieving accurate turbidity measurement in this embodiment;

[0061] Figure 3 Schematic diagram of the structure of an integrated water quality monitoring and processing system in this embodiment. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0063] As attached Figure 1 -Attached Figure 3 As shown, a method for integrated water quality monitoring and processing comprises the following steps:

[0064] S1: Set up a high-resolution camera to collect image data of the water tank solution with a time stamp in a preset direction and pre-process the image. The specific process is as follows:

[0065] S11: Setting a high-resolution camera in a preset direction of the water reservoir and initializing it, while collecting images of the water reservoir solution;

[0066] S12: For each channel of the image, a [5×5] Gaussian kernel (i.e., μ = 0 and σ 2 =1) Perform blur operation to remove noise;

[0067] S13: Standardize the data by the mean and standard deviation of the standard normal distribution;

[0068] S14: Normalize the values ​​and center them at zero to the interval [-1, 1].

[0069] S2: Use a pre-trained neural network to extract the image features of the data collected in step S1, and then pass the image features to a long short-term memory network (LSTM) to obtain the turbidity prediction value output by the model. The specific process is as follows:

[0070] S21: pass the preprocessed image sequence to the convolutional layer of the neural network, where the convolution kernel will slide on the image and perform a convolution operation on each position to extract local features;

[0071] S22: The feature map after the convolution operation is activated by an activation function to introduce nonlinearity;

[0072] S23: The feature map after the activation function will be passed to the pooling layer, where the pooling operation will downsample each feature map to reduce the dimension of the feature map and retain the most significant features;

[0073] S24: Flatten the feature map after the pooling operation into a one-dimensional vector, and pass it as input to the fully connected layer of the pre-trained neural network, so as to obtain the features of the image sequence;

[0074] S25: The image feature sequence is passed to the LSTM network. The hidden layer of the LSTM network receives and processes the input sequence, and the hidden state and memory unit are updated at each time step.

[0075] S26: The hidden layer of the LSTM network receives and processes the input sequence, updating the hidden state and memory cells at each time step;

[0076] S27: After being processed by the hidden layer, the LSTM network outputs a hidden state sequence, which contains information for modeling the input sequence;

[0077] S28: The hidden state sequence is used as the input of the output layer of the LSTM network, and the output layer generates a predicted sequence of turbidity values ​​based on the hidden state sequence.

[0078] S3: Install an immersion turbidity meter in the water storage tank, select a corresponding turbidity detection method according to the turbidity value predicted in step S2, and further achieve wide-range and accurate measurement of turbidity. The specific process is as follows:

[0079] S31: Setting an immersion turbidity meter in the water storage tank and initializing it;

[0080] S32: The turbidity value predicted in step S2 is judged according to the following formula. If the turbidity value is within the range of 0 to 400 NTU, it is judged as low turbidity; if the turbidity value is within the range of 400 to 1000 NTU, it is judged as high turbidity:

[0081]

[0082] Wherein, the unit NTU is nephelometric turbidity;

[0083] S33: According to the turbidity determined in step S32, the turbidity meter selects a corresponding measurement method. If the turbidity is low, the scattering method is used to further accurately measure the turbidity; if the turbidity is high, the transmission method is used to further accurately measure the turbidity.

[0084] Scattering method, the main steps are as follows:

[0085] (1) Set the Digital-to-Analog Converter (DAC) and turn on the light source;

[0086] (2) The analog-to-digital converter (ADC) samples 90 times and performs median average filtering on the scattered voltage to obtain V s1 ;

[0087] (3) DAC output is set to 0 and the light source is turned off;

[0088] (4) ADC samples 90 times and averages and filters the scattered voltage to obtain V s2 ;

[0089] (5) Substitute into the fitting model and calculate the liquid turbidity T measured by the scattering method s :

[0090]

[0091] Among them, k s is the scattering coefficient, R s is the signal amplification factor of the scattered light detection circuit.

[0092] Transmission method, the main steps are as follows:

[0093] (1) Set DAC and turn on the light source;

[0094] (2) ADC samples 90 times and filters the transmission voltage median average to obtain V t1 ;

[0095] (3) DAC output is set to 0 and the light source is turned off;

[0096] (4) ADC samples 90 times and averages and filters the transmission voltage to obtain V t2 ;

[0097] (5) Substitute into the fitting model and calculate the liquid turbidity T measured by the transmission method t :

[0098]

[0099] Among them, k t is the transmission coefficient, R t is the signal amplification factor of the transmitted light detection circuit, and b is a constant.

[0100] S4: Using the turbidity prediction value obtained in step S2 to establish the system state equation, combining the measured value of the turbidity meter in step 3 to establish the system observation equation, and using the Kalman filter algorithm to achieve further accurate prediction of turbidity. The specific process is as follows:

[0101] S41: Taking the turbidity value predicted in step S2 as the estimated value x of the system state k , establish the system state equation:

[0102] x k =f(x k-1 )

[0103] Where f(x) represents the LSTM network fitting function;

[0104] S42: The turbidity value measured by the turbidity meter in step S3 is used as the observed value z k Establish the observation equation:

[0105] z k =Hx k +v k

[0106] Among them, H represents the observation matrix, v k represents the observation noise;

[0107] S43: Initialize the estimated value of the system state at the initial moment and the state covariance matrix P0;

[0108] S44: At each time step k, the estimated value of the system state at the next moment is predicted based on the system state equation and the state covariance matrix

[0109] S45: At each time step k, the measurement residual at the current moment is calculated according to the observation equation:

[0110]

[0111] S46: At each time step k, the Kalman gain is calculated based on the system prediction residual and the system prediction covariance matrix:

[0112]

[0113] Where R is the covariance matrix of observation noise;

[0114] S47: At each time step k, update the estimated value of the system state and the state covariance matrix to obtain the optimal estimated value at the current moment and the state covariance matrix

[0115]

[0116]

[0117] S48: Return to step S44 and repeat the prediction and update until all observation data have been processed.

[0118] S5: Store the current timestamp and turbidity value, and choose whether to maintain the operation of the water extraction system according to the turbidity value. If the turbidity exceeds the threshold, process the pool solution. The specific process is as follows:

[0119] S51: The system records the current timestamp and the corresponding turbidity value, and records this information into the system database or log for subsequent analysis and monitoring;

[0120] S52: Decision rules are formulated based on the preset water quality turbidity standard. If the turbidity value exceeds a certain threshold, the water quality is considered unqualified, the water pumping system is shut down, and the water storage tank is treated. If the turbidity value is lower than the threshold, the water pumping system is maintained in operation, so that clean water that meets the standard is lifted to the required height for water supply.

[0121] S53: Provides remote access and management interface, system administrators can view and manage recorded data anytime and anywhere.

[0122] A system integrating water quality monitoring and processing, comprising:

[0123] An image acquisition module is used to set a high-resolution camera to acquire image data of the water reservoir solution with a time stamp in a preset direction and pre-process the image;

[0124] The turbidity prediction module uses a pre-trained neural network to extract the image features of the collected data, and then passes the image features to the long short-term memory network (LSTM) to obtain the turbidity prediction value output by the model;

[0125] The turbidity detection module is used to set up an immersion turbidity meter in the water storage tank, select the corresponding turbidity detection method according to the turbidity value predicted by the turbidity prediction module, and further realize the wide range and accurate measurement of turbidity;

[0126] The data fusion module uses the turbidity prediction value obtained in step S2 to establish the system state equation, combines the measured value of the turbidity meter in step 3 to establish the system observation equation, and realizes further accurate prediction of turbidity through the Kalman filter algorithm;

[0127] The water supply module is used to store the current timestamp and turbidity value, and choose whether to maintain the operation of the water extraction system according to the turbidity value. If the turbidity exceeds the threshold, the pool solution is processed.

[0128] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method for integrated water quality monitoring and processing, characterized in that: The following steps are involved: S1: Setting a high-resolution camera to collect image data of the water tank solution with a time stamp in a preset direction and preprocessing the image; S2: using a pre-trained neural network to extract the image features of the data collected in step S1, passing the image features into the long short-term memory network, and obtaining the turbidity prediction value output by the model; S3: Setting an immersion turbidity meter in the water storage tank, selecting a corresponding turbidity detection method according to the turbidity value predicted in step S2, and realizing wide-range and accurate measurement of turbidity; S4: using the turbidity prediction value obtained in step S2 to establish a system state equation, combining the measured value of the turbidity meter in step S3 to establish a system observation equation, and realizing accurate prediction of turbidity through Kalman filtering algorithm; S5: The current timestamp and turbidity value are stored, and whether to maintain the operation of the water pumping system is selected according to the turbidity value. If the turbidity exceeds the preset threshold, the pool solution is processed; if the turbidity does not exceed the preset threshold, no processing is performed.

2. The method of integrated water quality monitoring and processing according to claim 1, characterized in that: Step S1 includes the following steps: S11: Setting a high-resolution camera in a preset direction of the water reservoir and initializing it, while collecting images of the water reservoir solution; S12: For each channel of the image, noise is removed by performing a blur operation using a [5×5] Gaussian kernel given by a standard normal distribution; S13: Standardize the data by the mean and standard deviation of the standard normal distribution; S14: Normalize the values.

3. The method of integrated water quality monitoring and processing according to claim 1, characterized in that: Step S2 includes the following steps: S21: pass the preprocessed image sequence to the convolutional layer of the neural network, where the convolution kernel will slide on the image and perform a convolution operation on each position to extract local features; S22: Activate the feature map after the convolution operation through an activation function to introduce nonlinearity; S23: The feature map after the activation function will be passed to the pooling layer, where the pooling operation will downsample each feature map to reduce the dimension of the feature map and retain the most significant features; S24: Flatten the feature map after the pooling operation into a one-dimensional vector, and pass it as input to the fully connected layer of the pre-trained neural network, so as to obtain the features of the image sequence; S25: The image feature sequence is passed to the LSTM network. The hidden layer of the LSTM network receives and processes the input sequence, and the hidden state and memory unit are updated at each time step. S26: After being processed by the hidden layer, the LSTM network outputs a hidden state sequence, which contains information for modeling the input sequence; S27: The hidden state sequence is used as the input of the output layer of the LSTM network, and the output layer generates a predicted sequence of turbidity values ​​based on the hidden state sequence.

4. The method of integrated water quality monitoring and processing according to claim 1, characterized in that: Step S3 includes the following steps: S31: Setting an immersion turbidity meter in the water storage tank and initializing it; S32: If the turbidity value is within the range of 0 to 400 NTU, it is determined to be low turbidity; if the turbidity value is within the range of 400 to 1000 NTU, it is determined to be high turbidity; S33: According to the result determined in step S32, a corresponding measurement method is selected for the turbidity meter. If the turbidity is low, a scattering method is used to further accurately measure the turbidity; if the turbidity is high, a transmission method is used to further accurately measure the turbidity.

5. The method of integrated water quality monitoring and processing according to claim 1, characterized in that: Step S4 includes the following steps: S41: Taking the turbidity value predicted in step S2 as the estimated value x of the system state k , establish the system state equation: x k =f(x k-1 ) Where f(x) represents the LSTM network fitting function; S42: The turbidity value measured by the turbidity meter in step S3 is used as the observed value z k Establish the observation equation: z k =Hx k +v k Among them, H represents the observation matrix, v k represents the observation noise; S43: Initialize the estimated value of the system state at the initial moment and the state covariance matrix P0; S44: At each time step k, the estimated value of the system state at the next moment is predicted based on the system state equation and the state covariance matrix S45: At each time step k, the measurement residual at the current moment is calculated according to the observation equation: S46: At each time step k, the Kalman gain is calculated based on the system prediction residual and the system prediction covariance matrix: Where R is the covariance matrix of observation noise; S47: At each time step k, update the estimated value of the system state and the state covariance matrix to obtain the optimal estimated value at the current moment and the state covariance matrix S48: Return to step S44 and repeat the prediction and update until all observation data have been processed.

6. The method of integrated water quality monitoring and processing according to claim 1, characterized in that: Step S5 includes the following steps: S51: The system records the current timestamp and the corresponding turbidity value, and records this information into the system database or log for subsequent analysis and monitoring; S52: According to the preset water quality turbidity standard, a decision rule is formulated. If the turbidity value exceeds the threshold, the water quality is considered unqualified, the water pumping system is shut down, and the water storage tank is treated; if the turbidity value is lower than the threshold, the water pumping system is maintained in operation; S53: Provides remote access and management interface, system administrators can view and manage recorded data anytime and anywhere.

7. A system integrating water quality monitoring and processing, characterized in that: include: An image acquisition module is used to set a high-resolution camera to acquire image data of the water reservoir solution with a time stamp in a preset direction and pre-process the image; The turbidity prediction module uses a pre-trained neural network to extract the image features of the collected data, and then passes the image features to the long short-term memory network to obtain the turbidity prediction value output by the model; The turbidity detection module is used to set up an immersion turbidity meter in the water storage tank, select the corresponding turbidity detection method according to the turbidity value predicted by the turbidity prediction module, and further realize the wide range and accurate measurement of turbidity; Data fusion module, establishes the system state equation, combines the measured values ​​of the turbidity meter to establish the system observation equation, and realizes the turbidity prediction through the Kalman filter algorithm; The water supply module is used to store the current timestamp and turbidity value, and choose whether to maintain the operation of the water extraction system according to the turbidity value. If the turbidity exceeds the threshold, the pool solution is processed.

Images