A method and system for early warning of reverse osmosis membrane fouling

Abstract

The present invention relates to a reverse osmosis membrane fouling warning method and system, which is characterized by comprising: adopting a machine learning method to construct a reverse osmosis membrane fouling warning model based on the historical measured data of the reverse osmosis device to be measured; obtaining the measured data of the reverse osmosis device to be measured, and selecting characteristic variables for denoising and preprocessing; inputting the denoised and preprocessed characteristic variables into the constructed reverse osmosis membrane fouling warning model to obtain the warning result of the reverse osmosis device to be measured. The present invention can monitor the fouling condition of the reverse osmosis membrane and can be widely applied to the information field of water treatment systems.

Description

Technical Field

[0001] The present invention relates to the field of informatization of water treatment systems, and particularly to a method and system for early warning of reverse osmosis membrane fouling. Background Art

[0002] Currently, in the production processes of high-standard industrial pure water (such as boiler chemical water) and wastewater treatment processes, reverse osmosis (RO) devices are widely used as desalination treatment processes. Although the pretreatment process of reverse osmosis feed water has been very mature, the occurrence of reverse osmosis membrane fouling is inevitable. As the fouling of the reverse osmosis membrane gradually intensifies, the inter-stage pressure difference of the reverse osmosis device also increases accordingly. When the inter-stage pressure difference accumulates to a set threshold, it is necessary to stop the chemical cleaning of the membrane components of the reverse osmosis device, which directly affects the operating efficiency of the reverse osmosis device. Therefore, early warning of reverse osmosis membrane fouling is particularly important.

[0003] Existing reverse osmosis devices are all equipped with several sensors to collect a large amount of operating data, but they do not have the function of early warning of reverse osmosis membrane fouling. Field operators often stop the chemical cleaning of the membrane components of the reverse osmosis device based on experience and when the inter-stage pressure difference reaches the threshold. How to use the massive historical operating data and combine artificial intelligence methods to warn of the fouling of the reverse osmosis membrane, timely push warning information to the operators, guide the operators to adjust the operating conditions of the reverse osmosis device, delay the trend of reverse osmosis membrane fouling, and increase the operating cycle duration of the reverse osmosis device is of great significance for controlling reverse osmosis membrane fouling and improving the operating efficiency of the reverse osmosis device. Summary of the Invention

[0004] In view of the above problems, the purpose of the present invention is to provide a method and system for early warning of reverse osmosis membrane fouling that can improve the operating efficiency of reverse osmosis devices.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions: In the first aspect, a method for early warning of reverse osmosis membrane fouling is provided, including:

[0006] Using a machine learning method, based on the historical measured data of the reverse osmosis device to be measured, construct a reverse osmosis membrane fouling early warning model;

[0007] Obtain the measured data of the reverse osmosis device to be measured, and select characteristic variables for denoising and preprocessing;

[0008] Input the denoised and preprocessed characteristic variables into the constructed reverse osmosis membrane fouling early warning model to obtain the early warning result of the reverse osmosis device to be measured.

[0009] Further, the step of using a machine learning method to construct a reverse osmosis membrane fouling early warning model based on the historical measured data of the reverse osmosis device includes:

[0010] Obtain the historical measured data of the reverse osmosis device to be tested for feature engineering and feature selection, determine the correlation coefficients between the operating parameters of the reverse osmosis device in the historical measured data, and select one or more operating parameters that have the greatest impact on the reverse osmosis membrane fouling problem as feature variables according to the determined correlation coefficients;

[0011] Denoise and preprocess the selected feature variables;

[0012] Adopt a machine learning method to construct a reverse osmosis membrane fouling warning model based on the preprocessed feature variables, and determine the evaluation indicators of the reverse osmosis membrane fouling warning model, including the accuracy rate indicator and the AUC indicator;

[0013] Train, validate and test the constructed reverse osmosis membrane fouling warning model according to the preprocessed feature variables and the determined evaluation indicators to obtain the constructed reverse osmosis membrane fouling warning model.

[0014] Further, the correlation coefficient ρ is:

[0015]

[0016] where x i and y i represent the values of two variables; and represent the means of the two variable populations.

[0017] Further, the operating parameters of the reverse osmosis device include one or more combinations of the conductivity, pH value, ORP value, temperature, flow rate and pressure of the concentrated water and the produced water, the operating frequency feedback and current of the high-pressure pump, and the differential pressure at the inlet and outlet of the reverse osmosis device.

[0018] Further, the training, validating and testing the constructed reverse osmosis membrane fouling warning model according to the preprocessed feature variables and the determined evaluation indicators to obtain the constructed reverse osmosis membrane fouling warning model includes:

[0019] Divide the preprocessed feature variables into a training set, a validation set and a test set, and set the hyperparameters of the constructed reverse osmosis membrane fouling warning model;

[0020] Based on the set hyperparameters, train, cross-validate and test the constructed reverse osmosis membrane fouling warning model through the training set, the validation set and the test set, calculate the accuracy rates of the constructed reverse osmosis membrane fouling warning model on the training set, the validation set and the test set, draw the ROC curve of the reverse osmosis membrane fouling warning model on the test set, calculate the AUC of the test set, and output the warning result;

[0021] If both the accuracy rate and the AUC are higher than the pre-set thresholds, the constructed reverse osmosis membrane fouling warning model is obtained; otherwise, the reverse osmosis membrane fouling warning model is improved until a reverse osmosis membrane fouling warning model that meets the on-site deployment requirements is obtained.

[0022] Further, the improvement of the reverse osmosis membrane fouling warning model includes:

[0023] Reset the hyperparameters;

[0024] If the accuracy rate and the AUC of the reverse osmosis membrane fouling warning model after resetting the hyperparameters are still not higher than the pre-set thresholds, other machine learning methods are used to construct the reverse osmosis membrane fouling warning model until a reverse osmosis membrane fouling warning model that meets the on-site deployment requirements is obtained.

[0025] Further, if the reverse osmosis device to be tested starts to foul, the warning result is to issue a warning signal, and the operating conditions of the reverse osmosis device are checked and adjusted in a timely manner according to the warning result; after adjusting the operating conditions, it is judged whether the reverse osmosis device to be tested still has fouling through the constructed reverse osmosis membrane fouling warning model; if it still exists, the warning result continues to issue a warning signal; if not, the warning result issues a normal signal.

[0026] In a second aspect, a reverse osmosis membrane fouling warning system is provided, including:

[0027] A model construction module, configured to use a machine learning method to construct a reverse osmosis membrane fouling warning model based on the historical measured data of the reverse osmosis device to be tested;

[0028] A data acquisition module, configured to acquire the measured data of the reverse osmosis device to be tested, and select characteristic variables for denoising and preprocessing;

[0029] A warning module, configured to input the denoised and preprocessed characteristic variables into the constructed reverse osmosis membrane fouling warning model to obtain the warning result of the reverse osmosis device to be tested.

[0030] In a third aspect, a processing device is provided, including computer program instructions, wherein when the computer program instructions are executed by the processing device, they are used to implement the steps corresponding to the above reverse osmosis membrane fouling warning method.

[0031] In a fourth aspect, a computer-readable storage medium is provided, on which computer program instructions are stored, wherein when the computer program instructions are executed by a processor, they are used to implement the steps corresponding to the above reverse osmosis membrane fouling warning method.

[0032] Due to the above technical solutions adopted by the present invention, it has the following advantages:

[0033] 1. When the present invention is in specific operation, according to the measured data of the reverse osmosis device collected by the sensor, if the reverse osmosis membrane starts to be fouled, the on-site deployed reverse osmosis membrane fouling warning model is used to judge in real time whether fouling occurs. When fouling occurs, the model will timely push a warning signal to the operator, so as to guide the operator to adjust the operating conditions of the reverse osmosis device, delay the trend of reverse osmosis membrane fouling, and increase the operating cycle duration of the reverse osmosis device.

[0034] 2. The reverse osmosis membrane fouling warning model constructed in the present invention can also judge whether the membrane fouling has improved after the operator adjusts the operating conditions of the reverse osmosis device, so as to judge whether the operation adjustment of the operator is effective. The present invention can monitor the fouling condition of the reverse osmosis membrane, timely adjust the operating conditions, delay the chemical cleaning time of the reverse osmosis membrane, and improve the operating efficiency.

[0035] In summary, the present invention can be widely applied to the information field of water treatment systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] By reading the detailed description of the preferred embodiments below, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0037] Figure 1 is a schematic flow chart of the method provided by an embodiment of the present invention;

[0038] Figure 2 is a schematic diagram of the curve comparison of a section of differential pressure data before and after denoising by using the exponential moving average method provided by an embodiment of the present invention;

[0039] Figure 3 is a schematic diagram of the curve comparison of a section of differential pressure data before and after denoising by using the wavelet transform method provided by an embodiment of the present invention;

[0040] Figure 4 is a schematic diagram of the curve comparison of a section of differential pressure data before and after denoising by using the method of the operating frequency of the high-pressure pump provided by an embodiment of the present invention;

[0041] Figure 5 is a schematic diagram of the form of the image data after denoising and preprocessing provided by an embodiment of the present invention;

[0042] Figure 6 is a schematic diagram of the ROC curve and AUC of the model constructed by using the method of the present invention on the test set provided by an embodiment of the present invention;

[0043] Figure 7It is a schematic diagram of the prediction results of the model constructed by the method of the present invention in an embodiment of the present invention on the test set. Detailed implementation manners

[0044] Hereinafter, the exemplary embodiments of the present invention will be described in more detail with reference to the accompanying drawings. Although the exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present invention can be more thoroughly understood and the scope of the present invention can be fully conveyed to those skilled in the art.

[0045] It should be understood that the terms used herein are for the purpose of describing specific exemplary embodiments only and are not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms "a", "an", and "the" as used herein may also include the plural forms. The terms "comprising", "including", "containing", and "having" are inclusive and thus specify the presence of the stated features, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not to be construed as necessarily requiring them to be performed in the particular order described or illustrated, unless the order of performance is explicitly stated. It should also be understood that additional or alternative steps may be used.

[0046] The reverse osmosis membrane fouling warning method and system provided by the embodiments of the present invention are based on the historical measured data related to reverse osmosis membrane fouling, establish a reverse osmosis membrane fouling warning model based on machine learning, perform denoising processing and feature engineering on the data, and perform training, verification, testing, and on-site deployment of the model. This model can be used for the warning of reverse osmosis membrane fouling, reminding the operator to handle it, guiding the operator to adjust the operating conditions of the reverse osmosis device, delaying the trend of reverse osmosis membrane fouling, thereby delaying the reverse osmosis membrane fouling, and further using this model to judge whether the membrane fouling has been improved.

[0047] Embodiment 1

[0048] As Figure 1 shown, this embodiment provides a reverse osmosis membrane fouling warning method, including the following steps:

[0049] 1) Using a machine learning method, based on the historical measured data of the reverse osmosis device to be tested, construct a reverse osmosis membrane fouling warning model, specifically:

[0050] 1.1) Extract the historical measured data of the reverse osmosis device to be tested stored in the data extraction platform for feature engineering and feature selection, determine the correlation coefficients between the operating parameters of the reverse osmosis device in the historical measured data, and select one or more operating parameters that have the greatest impact on the reverse osmosis membrane fouling problem as feature variables according to the determined correlation coefficients.

[0051] Specifically, the operating parameters of the reverse osmosis device include indicators such as the conductivity, pH value (hydrogen ion concentration index), ORP value (oxidation-reduction potential), temperature, flow rate, and pressure of the concentrated water and the produced water, and one or more combinations of the operating frequency feedback, current of the high-pressure pump, and the differential pressure at the inlet and outlet of the reverse osmosis device.

[0052] Specifically, the purpose of the present invention is the early warning judgment of reverse osmosis membrane fouling, which is a discrete variable with labels of early warning / normal. Therefore, the Spearman correlation coefficient is used to calculate the correlation between variables. The Spearman correlation coefficient ρ is:

[0053]

[0054] where x i and y i represent the values of two variables; and represent the overall means of the two variables.

[0055] More specifically, if y tends to increase when x increases, the Spearman correlation coefficient is positive; if y tends to decrease when x increases, the Spearman correlation coefficient is negative. A Spearman correlation coefficient of 0 indicates that there is no tendency for y when x increases; when x and y are closer and closer to a completely monotonic correlation, the Spearman correlation coefficient will increase in absolute value.

[0056] 1.2) Denoise and preprocess the selected feature variables.

[0057] 1.2.1) Denoise the selected feature variables.

[0058] Specifically, due to the start-stop of the reverse osmosis device and the errors of the data instruments, there are often obvious noises in the measured data collected by the sensors. To improve the signal-to-noise ratio of the data and prevent the noises from interfering with the model effect, methods are needed to reduce the noises in the measured data, such as mean filtering, exponential moving average, or denoising using parameters associated with the noises.

[0059] 1.2.2) Preprocess the denoised feature variables.

[0060] Specifically, the denoised feature variables are organized into the format required by the following reverse osmosis membrane fouling warning model, such as one-dimensional data, two-dimensional graphics, or multi-dimensional data, etc.

[0061] 1.3) Using machine learning methods, construct a reverse osmosis membrane fouling warning model based on the preprocessed feature variables, and determine the evaluation indicators of the reverse osmosis membrane fouling warning model, specifically:

[0062] 1.3.1) Using machine learning methods, construct a reverse osmosis membrane fouling warning model based on the preprocessed feature variables. Among them, the input of the reverse osmosis membrane fouling warning model is the preprocessed feature variables, and the output of the reverse osmosis membrane fouling warning model is the warning result, including an alarm signal and a normal signal. When the reverse osmosis membrane of the reverse osmosis device is fouled, the warning result is an alarm signal; otherwise, it is a normal signal.

[0063] Specifically, the reverse osmosis membrane fouling warning model can be a linear regression model, a support vector machine model, a random forest model, a general deep neural network model, a convolutional neural network model, or a recurrent neural network model, etc.

[0064] 1.3.2) Determine the evaluation indicators of the reverse osmosis membrane fouling warning model, including the accuracy rate indicator and the AUC indicator.

[0065] Specifically, the reverse osmosis membrane fouling warning problem involved in the present invention is a binary classification problem, that is, to judge whether the reverse osmosis membrane unit is fouled according to the formatted data. Considering the timeliness of the prediction results, the network parameters in the reverse osmosis membrane fouling warning model constructed in the present invention should not be too many to avoid affecting the judgment speed after the model is deployed. The most commonly used evaluation indicator for determining the effect of the binary classification model is accuracy, which is defined as the proportion of correctly classified samples in all samples to the total number of samples, and can be used as an evaluation indicator for the training process and the final effect:

[0066]

[0067] Among them, T (True) and P (Positive) respectively represent true and positive classes; F (False) and N (Negative) respectively represent false and negative classes; TP is true positive, indicating the number of correctly predicted positive samples; FP is false positive, indicating the number of incorrectly predicted positive samples; TN is true negative, indicating the number of correctly predicted negative samples; FN is false negative, indicating the number of incorrectly predicted negative samples.

[0068] Specifically, the accuracy metric is convenient and intuitive. However, for samples with imbalanced distributions, the accuracy metric cannot reflect the true performance of the model, while the ROC curve (Receiver Operating Characteristic Curve) remains unchanged when the distribution of positive and negative samples in the sample changes. In machine learning tasks, the results output by the classifier are generally probability values, and setting different thresholds may lead to different prediction results. The curve obtained by plotting the true positive rate predicted by the model under different thresholds on the horizontal axis and the false positive rate on the vertical axis is the ROC curve. The area under the ROC curve is the AUC (Area Under Curve), and its value range is [0, 1]. Ideally, the true positive rate is 1 and the false positive rate is 0, and at this time the AUC is 1. Therefore, the closer the AUC is to 1, the better the performance of the model. If the AUC is less than 0.5, it means the model is ineffective. If the AUC reaches above 0.9, it means the model has a relatively good effect.

[0069] 1.4) According to the preprocessed feature variables and the determined evaluation metrics, train, validate, test, and evaluate the constructed reverse osmosis membrane fouling warning model to obtain the constructed reverse osmosis membrane fouling warning model. Specifically:

[0070] 1.4.1) Divide the preprocessed feature variables into a training set, a validation set, and a test set, and set the hyperparameters of the constructed reverse osmosis membrane fouling warning model.

[0071] 1.4.2) Based on the set hyperparameters, train, cross-validate, and test the constructed reverse osmosis membrane fouling warning model using the training set, validation set, and test set, calculate the accuracy of the constructed reverse osmosis membrane fouling warning model on the training set, validation set, and test set, plot the ROC curve of the reverse osmosis membrane fouling warning model on the test set, calculate the AUC of the test set, and output the warning result.

[0072] 1.4.3) If both the accuracy and the AUC are higher than a pre-set threshold, such as 90%, it indicates that the constructed reverse osmosis membrane fouling warning model can meet the requirements for on-site deployment, and proceed to step 1.4.4); otherwise, proceed to step 1.4.1) to reset the hyperparameters. If both the accuracy and the AUC are still not higher than the pre-set threshold, proceed to step 1.3.1) to construct a reverse osmosis membrane fouling warning model using other machine learning methods until a reverse osmosis membrane fouling warning model that meets the requirements for on-site deployment is obtained.

[0073] 1.4.4) Conduct on-site testing on the constructed reverse osmosis membrane fouling warning model:

[0074] Build a reverse osmosis membrane fouling warning model deployed in the on-site distributed control system (DCS) or data platform, and use the historical measured data of the on-site distributed control system to select, denoise, and preprocess the characteristic variables, and then test the test effect of the built reverse osmosis membrane fouling warning model. If the test results do not meet the expectations, modify the interface program of the on-site distributed control system or data platform.

[0075] 2) Obtain the measured data of the reverse osmosis device to be tested through sensors, and upload the obtained measured data to the distributed control system and store it in the data platform.

[0076] 3) Select the characteristic variables of the obtained measured data, and perform denoising and preprocessing.

[0077] 4) Input the denoised and preprocessed characteristic variables into the built reverse osmosis membrane fouling warning model to obtain the warning result of the reverse osmosis device to be tested.

[0078] Specifically, if the reverse osmosis device to be tested starts to be fouled, the warning result is to issue a warning signal, and the operator checks and adjusts the operating conditions of the reverse osmosis device in a timely manner according to the warning result.

[0079] Specifically, after adjusting the operating conditions, the built reverse osmosis membrane fouling warning model can also be called to determine whether the reverse osmosis device to be tested is still fouled; if it is still fouled, the warning result continues to issue a warning signal; if not, the warning result issues a normal signal.

[0080] The following takes the reverse osmosis device for chemical water treatment in a certain thermal power plant as a specific embodiment to detail the reverse osmosis membrane fouling warning method of the present invention:

[0081] 1) Based on the historical measured data of the reverse osmosis device for chemical water treatment, build a reverse osmosis membrane fouling warning model:

[0082] ① Extract the measured historical data of the reverse osmosis device for chemical water treatment stored in the data platform, including the data of the heating seasons from 2019 to 2020 and from 2020 to 2021, and a total of four reverse osmosis devices, namely A, B, C, and E. Analyze the data, perform feature engineering and feature selection, and study the correlation coefficients between variables. According to the calculation results of the correlation coefficients, select one or more variables that have a greater impact on the reverse osmosis membrane fouling problem as characteristic variables.

[0083] From the calculation results of the correlation coefficient, it can be seen that the characteristics that have the greatest impact on the fouling of the reverse osmosis membrane are the differential pressure between the inlet and outlet of the first stage and the current signal of the chemical dosing metering pump. In addition, the conductivity of the produced water, the inlet / outlet / produced water flow rate, and the frequency feedback of the high-pressure pump all have a certain impact on the fouling of the reverse osmosis membrane. However, the correlation between other variables such as the pH, ORP, temperature, etc. of the inlet main pipe water quality and membrane fouling is close to 0, with almost no correlation and no value to be utilized in the machine learning model. Therefore, these indicators can be discarded during the training of the subsequent model.

[0084] ② Denoise and preprocess the selected characteristic variables.

[0085] Denoising: Taking the exponential moving average method for denoising as an example, set the smoothing coefficient to 0.3. The exponential moving average denoising result is as Figure 2 shown (where raw data is the original data and denoised data is the data after denoising). It can be seen from the figure that the noise of the data is improved to a certain extent, but there are still obvious "spike" outliers. The denoising effect cannot meet the requirements of the model, and a better algorithm still needs to be found to denoise the data. The denoising effect of wavelet transform is as Figure 3 shown. Compared with the exponential moving average denoising method, the denoising effect of wavelet transform is improved, the amplitude of the "spike" appears significantly reduced, and the trend of mutation decreases. However, wavelet transform denoising cannot reduce the number of "spikes", so the effect is still not satisfactory.

[0086] Combined with the operating conditions of the reverse osmosis device, analyzing the measured data found that the data noise in the measured data collected by the sensor mainly comes from the start-stop and frequency adjustment of the reverse osmosis high-pressure pump. When the high-pressure pump stops, the differential pressure of the first stage will drop to zero; during normal operation, the inter-stage differential pressure of the reverse osmosis will increase with the increase of the high-pressure pump frequency and decrease with the decrease. According to the above characteristics, the differential pressure data of the first stage can be processed according to the operating frequency of the high-pressure pump to remove noise and outliers. The specific process is as follows:

[0087] A) After obtaining the differential pressure data of the first stage, obtain the high-pressure pump frequency data corresponding to the differential pressure data at the same point and in the same time period.

[0088] B) Delete the differential pressure data of the first stage with the high-pressure pump frequency less than the normal operating frequency (such as 20 Hz) to remove the influence of the change in the inter-stage differential pressure caused by the start-stop of the high-pressure pump.

[0089] C) If the high-pressure pump frequency changes, determine whether the duration of the changed frequency is less than a pre-set duration, such as 0.5 - 1 h. If so, delete the corresponding differential pressure data of the first stage to remove the influence of the change in the inter-stage differential pressure caused by the short-term frequency change of the high-pressure pump.

[0090] D) Interpolate and fill in the deleted data to ensure that the total amount of data does not decrease, and obtain the denoised data.

[0091] Select a section of differential pressure data for a period of time after December 14, 2020 from the original measured data. Figure 4 It can be seen that through the above method, the data with abnormal fluctuations in the differential pressure of the first stage caused by the change of the operating conditions of the reverse osmosis high-pressure pump can be removed, and the noise in the measured data can be reduced. The denoising effect of the denoising method combined with the operating frequency of the high-pressure pump is as Figure 4 shown. Compared with wavelet transform denoising, this method can further reduce the amplitude of outliers. Figure 4 It can be seen that after denoising, the curve has basically become smooth, and the outliers in the original measured data have basically been eliminated, and the signal-to-noise ratio of the data has been significantly improved.

[0092] Pretreatment: For example, for a certain time point, select the measured data for a period of time before, plot with time as the horizontal axis and the differential pressure of the first stage as the vertical axis, and perform denoising in combination with the denoising method of the high-pressure pump frequency. Finally, save the picture as a two-dimensional matrix, as Figure 5 shown.

[0093] ③ Build a reverse osmosis membrane fouling warning model and determine the evaluation index of the reverse osmosis membrane fouling warning model.

[0094] ④ According to the preprocessed characteristic variables and the determined evaluation index, train, verify, test and evaluate the constructed reverse osmosis membrane fouling warning model.

[0095] The original historical measured data includes the data of the heating seasons from 2019 to 2020 and from 2020 to 2021. The data of the heating season from 2020 to 2021 includes a total of four sets of devices A, B, C and E. The historical measured data of the first three sets of devices are used as the training set, and the historical measured data of device E are used as the verification set. All the historical measured data from 2019 to 2020 are used as the test set. After training, verification and testing, the accuracy rates of the reverse osmosis membrane fouling warning model on the training set, verification set and test set are shown in Table 1 below. The ROC curve and prediction results of the reverse osmosis membrane fouling warning model on the test set are as Figure 6 and 7 shown.

[0096] It can be seen from Table 1 below that the accuracy rates of the reverse osmosis membrane fouling warning model on the training set, verification set and test set are 100%, 95% and 96% respectively. The AUC on the test set can reach 0.98, and both the accuracy rate and AUC on the test set can meet the requirements of industrial deployment. Figure 7The prediction results of the reverse osmosis membrane fouling warning model during the heating seasons from 2019 to 2020 are shown. Among them, the diamond points represent the time points when the reverse osmosis membrane fouling warning model issues warnings, and the round points represent the time points under normal operating conditions. It can be seen that the warning time is very consistent with the time point when fouling actually occurs:

[0097] Table 1: The accuracy rates of the reverse osmosis membrane fouling warning model on the training set, validation set, and test set respectively

[0098] Accuracy Number of samples Training set 1.00 761 Validation set 0.95 159 Test set 0.96 1110

[0099] The constructed reverse osmosis membrane fouling warning model is tested on-site, and the test results are the same as the above test results.

[0100] 2) Obtain the measured data of the reverse osmosis device through sensors, and upload the obtained measured data to the discrete control system and store it in the data platform.

[0101] 3) Select the characteristic variables of the obtained measured data, and perform denoising and preprocessing.

[0102] 4) Input the denoised and preprocessed characteristic variables into the constructed reverse osmosis membrane fouling warning model to obtain the warning result of the reverse osmosis device to be tested.

[0103] Example 2

[0104] This example provides a reverse osmosis membrane fouling warning system, including:

[0105] A model construction module, which is used to construct a reverse osmosis membrane fouling warning model based on the historical measured data of the reverse osmosis device to be tested by using machine learning methods.

[0106] A data acquisition module, which is used to acquire the measured data of the reverse osmosis device to be tested, and select characteristic variables for denoising and preprocessing.

[0107] A warning module, which is used to input the denoised and preprocessed characteristic variables into the constructed reverse osmosis membrane fouling warning model to obtain the warning result of the reverse osmosis device to be tested.

[0108] Example 3

[0109] This example provides a processing device corresponding to the reverse osmosis membrane fouling warning method provided in Example 1. The processing device can be a processing device for a client, such as a mobile phone, laptop computer, tablet computer, desktop computer, etc., to execute the method of Example 1.

[0110] The processing device includes a processor, a memory, a communication interface, and a bus. The processor, the memory, and the communication interface are connected through the bus to complete communication with each other. A computer program that can run on the processing device is stored in the memory. When the processing device runs the computer program, it executes the reverse osmosis membrane fouling warning method provided in Embodiment 1 of the present invention.

[0111] In some implementations, the memory may be a high-speed random access memory (RAM: Random Access Memory), and may also include non-volatile memory, such as at least one disk memory.

[0112] In other implementations, the processor may be various types of general-purpose processors such as a central processing unit (CPU), a digital signal processor (DSP), etc., which are not limited herein.

[0113] Embodiment 4

[0114] This embodiment provides a computer program product corresponding to the reverse osmosis membrane fouling warning method provided in Embodiment 1. The computer program product may include a computer-readable storage medium, on which computer-readable program instructions for executing the reverse osmosis membrane fouling warning method described in Embodiment 1 are uploaded.

[0115] A computer-readable storage medium may be a tangible device that holds and stores instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any combination of the above.

[0116] The above embodiments are only used to illustrate the present invention. The structures, connection methods, manufacturing processes, etc. of each component can all be changed. Any equivalent transformation and improvement made on the basis of the technical solution of the present invention should not be excluded from the protection scope of the present invention.

Claims

1. A method for early warning of reverse osmosis membrane fouling, characterized in that, it includes: Adopting a machine learning method, based on the historical measured data of the reverse osmosis device to be measured, constructing an early warning model for reverse osmosis membrane fouling; Obtaining the measured data of the reverse osmosis device to be measured, and selecting characteristic variables for denoising and preprocessing; Inputting the denoised and preprocessed characteristic variables into the constructed early warning model for reverse osmosis membrane fouling to obtain the early warning result of the reverse osmosis device to be measured; The adopting a machine learning method to construct an early warning model for reverse osmosis membrane fouling based on the historical measured data of the reverse osmosis device includes: Obtaining the historical measured data of the reverse osmosis device to be measured for feature engineering and feature selection, determining the correlation coefficients between the operating parameters of the reverse osmosis device in the historical measured data, and selecting one or more operating parameters that have the greatest impact on the reverse osmosis membrane fouling problem as characteristic variables according to the determined correlation coefficients; Denoising and preprocessing the selected characteristic variables; Adopting a machine learning method to construct an early warning model for reverse osmosis membrane fouling according to the preprocessed characteristic variables, and determining the evaluation indexes of the early warning model for reverse osmosis membrane fouling, including the accuracy index and the AUC index; Training, validating and testing the constructed early warning model for reverse osmosis membrane fouling according to the preprocessed characteristic variables and the determined evaluation indexes to obtain the constructed early warning model for reverse osmosis membrane fouling; The training, validating and testing the constructed early warning model for reverse osmosis membrane fouling according to the preprocessed characteristic variables and the determined evaluation indexes to obtain the constructed early warning model for reverse osmosis membrane fouling includes: Dividing the preprocessed characteristic variables into a training set, a validation set and a test set, and setting the hyperparameters of the constructed early warning model for reverse osmosis membrane fouling; Based on the set hyperparameters, training, cross-validating and testing the constructed early warning model for reverse osmosis membrane fouling through the training set, the validation set and the test set, calculating the accuracy of the constructed early warning model for reverse osmosis membrane fouling on the training set, the validation set and the test set, drawing the ROC curve of the early warning model for reverse osmosis membrane fouling on the test set, calculating the AUC of the test set, and outputting the early warning result; If both the accuracy and the AUC are higher than the pre-set threshold values, the constructed early warning model for reverse osmosis membrane fouling is obtained; otherwise, the early warning model for reverse osmosis membrane fouling is improved until an early warning model for reverse osmosis membrane fouling that meets the on-site deployment requirements is obtained.

2. A method for early warning of reverse osmosis membrane fouling according to claim 1, characterized in that, The correlation coefficient is as follows: Among them, and represent the values of two variables; and represent the overall means of the two variables.

3. A method for early warning of reverse osmosis membrane fouling according to claim 1, characterized in that, The operating parameters of the reverse osmosis device include the conductivity, pH value, ORP value, temperature, flow rate and pressure of the concentrated water and the produced water, and one or more combinations of the operating frequency feedback and current of the high-pressure pump and the differential pressure at the inlet and outlet of the reverse osmosis device.

4. A method for early warning of reverse osmosis membrane fouling according to claim 1, characterized in that, The improvement of the early warning model for reverse osmosis membrane fouling includes: Resetting the hyperparameters; If the accuracy rate and AUC of the reverse osmosis membrane fouling warning model after resetting the hyperparameters are still not higher than the preset threshold, other machine learning methods are used to construct the reverse osmosis membrane fouling warning model until a reverse osmosis membrane fouling warning model that meets the on-site deployment requirements is obtained.

5. A reverse osmosis membrane fouling warning method as described in claim 1, characterized in that if fouling begins to occur in the reverse osmosis device to be measured, the warning result is to issue a warning signal, and the operating conditions of the reverse osmosis device are checked and adjusted in a timely manner according to the warning result; after adjusting the operating conditions, it is judged whether there is still fouling in the reverse osmosis device to be measured through the constructed reverse osmosis membrane fouling warning model; if there is still fouling, the warning result continues to issue a warning signal; if not, the warning result issues a normal signal.

6. A reverse osmosis membrane fouling warning system, characterized in that it includes: a model construction module, which is used to construct a reverse osmosis membrane fouling warning model based on the historical measured data of the reverse osmosis device to be measured by using machine learning methods; a data acquisition module, which is used to acquire the measured data of the reverse osmosis device to be measured and select characteristic variables for denoising and preprocessing; a warning module, which is used to input the denoised and preprocessed characteristic variables into the constructed reverse osmosis membrane fouling warning model to obtain the warning result of the reverse osmosis device to be measured; The method of constructing a reverse osmosis membrane fouling warning model based on the historical measured data of the reverse osmosis device by using machine learning methods includes: acquiring the historical measured data of the reverse osmosis device to be measured for feature engineering and feature selection, determining the correlation coefficients between the operating parameters of the reverse osmosis device in the historical measured data, and selecting one or more operating parameters that have the greatest impact on the reverse osmosis membrane fouling problem as characteristic variables according to the determined correlation coefficients; denoising and preprocessing the selected characteristic variables; using machine learning methods to construct a reverse osmosis membrane fouling warning model according to the preprocessed characteristic variables, and determining the evaluation indexes of the reverse osmosis membrane fouling warning model, including the accuracy rate index and the AUC index; training, validating and testing the constructed reverse osmosis membrane fouling warning model according to the preprocessed characteristic variables and the determined evaluation indexes to obtain the constructed reverse osmosis membrane fouling warning model; The method of training, validating and testing the constructed reverse osmosis membrane fouling warning model according to the preprocessed characteristic variables and the determined evaluation indexes to obtain the constructed reverse osmosis membrane fouling warning model includes: dividing the preprocessed characteristic variables into a training set, a validation set and a test set, and setting the hyperparameters of the constructed reverse osmosis membrane fouling warning model; based on the set hyperparameters, training, cross-validating and testing the constructed reverse osmosis membrane fouling warning model through the training set, the validation set and the test set, calculating the accuracy rates of the constructed reverse osmosis membrane fouling warning model on the training set, the validation set and the test set, drawing the ROC curve of the reverse osmosis membrane fouling warning model on the test set, calculating the AUC of the test set, and outputting the warning result; If both the accuracy rate and the AUC are higher than the pre-set thresholds, the constructed reverse osmosis membrane fouling warning model is obtained; otherwise, the reverse osmosis membrane fouling warning model is improved until a reverse osmosis membrane fouling warning model that meets the on-site deployment requirements is obtained.

7. A processing device, characterized in that, it includes computer program instructions, wherein when the computer program instructions are executed by the processing device, they are used to implement the steps corresponding to the reverse osmosis membrane fouling warning method described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, computer program instructions are stored on the computer-readable storage medium, wherein when the computer program instructions are executed by the processor, they are used to implement the steps corresponding to the reverse osmosis membrane fouling warning method described in any one of claims 1-5.

Images