Water quality treatment device control method, system and equipment and storage medium

Abstract

The invention provides a water quality treatment device control method, system and equipment and a storage medium, and the method comprises the following steps: carrying out data acquisition on a water quality treatment device based on a water quality treatment duration required by a flocculation process to obtain multiple groups of target water inlet data and multiple flocculant addition amounts from a first moment to a second moment, the duration from the first moment to the second moment is the water quality treatment duration; on the basis of the multiple sets of target water inlet data and the multiple flocculating agent adding amounts, the turbidity prediction model is used for conducting effluent turbidity prediction, and the predicted effluent turbidity is obtained; and based on the predicted effluent turbidity and the actual effluent turbidity of the water quality treatment device at the second moment, controlling the water quality treatment device to carry out feedback regulation on the adding amount of the flocculating agent. According to the embodiment of the invention, the accuracy and the stability of predicting the effluent turbidity are improved, the control accuracy of the adding amount of the flocculating agent is improved, the actual effluent turbidity can be maintained near the required effluent turbidity, and the effluent quality stability is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of water treatment, and in particular to a water treatment device control method, system, equipment and storage medium. Background Art

[0002] Flocculants can remove tiny suspended solids in water. In the water treatment process, the addition of flocculants is a crucial link. The amount of flocculants added is related to the amount of colloids and turbidity in the effluent, which affects the turbidity of the effluent. The flocculant addition method of the existing water treatment system is usually that the relevant personnel rely on experience to judge the water quality of the inlet water, and then determine the amount of flocculants to be added; or use a set effluent turbidity value as the target and adjust the flocculant dosage according to the actual effluent turbidity. However, the flocculant addition logic of these two methods is relatively simple, and it is impossible to accurately control the amount of flocculants added, which makes the effluent water quality vary greatly, resulting in poor stability of the effluent water quality. Summary of the invention

[0003] The present invention provides a water treatment device control method, system, equipment and storage medium to solve the problem that the existing flocculant addition method is relatively simple and the flocculant addition amount cannot be accurately controlled, resulting in poor stability of effluent water quality.

[0004] In a first aspect, an embodiment of the present application provides a method for controlling a water treatment device, comprising:

[0005] Based on the water quality treatment time required by the flocculation process, data is collected from the water quality treatment device to obtain multiple groups of target water inflow data and multiple flocculant dosages from the first moment to the second moment, and the time from the first moment to the second moment is the water quality treatment time;

[0006] Based on multiple sets of target influent data and multiple flocculant dosages, the turbidity of the effluent is predicted using a turbidity prediction model to obtain the predicted effluent turbidity. The turbidity prediction model is a neural network model trained based on multiple historical water quality treatment samples of the water quality treatment device at different historical moments and the water quality treatment time.

[0007] Based on the predicted effluent turbidity and the actual effluent turbidity of the water treatment device at the second moment, the water treatment device is controlled to perform feedback adjustment on the dosage of the flocculant.

[0008] In a second aspect, an embodiment of the present application provides a water treatment system, including a water treatment device and a control device, wherein the control device is used to:

[0009] Based on the water quality treatment time required by the flocculation process, data is collected from the water quality treatment device to obtain multiple groups of target water inflow data and multiple flocculant dosages from the first moment to the second moment, and the time from the first moment to the second moment is the water quality treatment time;

[0010] Based on multiple sets of target influent data and multiple flocculant dosages, the turbidity of the effluent is predicted using a turbidity prediction model to obtain the predicted effluent turbidity. The turbidity prediction model is a neural network model trained based on multiple historical water quality treatment samples of the water quality treatment device at different historical moments and the water quality treatment time.

[0011] Based on the predicted effluent turbidity and the actual effluent turbidity of the water treatment device at the second moment, the water treatment device is controlled to perform feedback adjustment on the dosage of the flocculant.

[0012] In a third aspect, an embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned water treatment device control method when executing the computer program.

[0013] In a fourth aspect, an embodiment of the present application provides a readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned water quality treatment device control method are implemented.

[0014] In a solution provided by the above-mentioned water quality treatment device control method, system, equipment and storage medium, data is collected from the water quality treatment device based on the water quality treatment time required by the flocculation process to obtain multiple sets of target water inlet data and multiple flocculant dosages at the first moment and the second moment, and the time from the first moment to the second moment is the water quality treatment time; then, based on the multiple sets of target water inlet data and multiple flocculant dosages, the turbidity of the effluent is predicted using a turbidity prediction model to obtain the predicted effluent turbidity, and the turbidity prediction model is a neural network model obtained by training based on multiple historical water quality treatment samples of the water quality treatment device at different historical moments and the water quality treatment time; finally, based on the predicted effluent turbidity and the actual effluent turbidity of the water quality treatment device at the second moment, the water quality treatment device is controlled to feedback adjust the dosage of the flocculant. In this embodiment, a high-precision turbidity prediction model is trained through historical water quality treatment samples and the water quality treatment time required for the flocculation process, and multiple groups of target water inlet data and flocculant dosage within the water quality treatment time at the second moment and before are used as input. The turbidity of the effluent is predicted by the turbidity prediction model, so that the predicted effluent turbidity at the second moment is close to the historically predicted effluent turbidity, thereby improving the accuracy and stability of the predicted effluent turbidity. Based on the predicted effluent turbidity at the second moment and the actual effluent turbidity, the flocculant dosage is feedback-adjusted, and the flocculant dosage can be quickly adjusted to the dosage that meets the water quality treatment requirements, thereby improving the accuracy of flocculant dosage control, ensuring that the actual effluent turbidity is maintained near the required effluent turbidity, and improving the stability of the effluent water quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative labor.

[0016] Figure 1 is a structural schematic diagram of a water treatment system in one embodiment of the present invention;

[0017] Figure 2 yes Figure 1 A schematic diagram of the structure of a medium-sized water treatment device;

[0018] Figure 3 It is a schematic flow chart of a method for controlling a water treatment device in one embodiment of the present invention;

[0019] Figure 4 yes Figure 3 A schematic diagram of an implementation process of step S30;

[0020] Figure 5 yes Figure 3 A schematic diagram of an implementation process of step S40;

[0021] Figure 6 is a schematic diagram of a process for obtaining multiple historical water quality treatment samples in one embodiment of the present invention;

[0022] Figure 7 yes Figure 1 A structural schematic diagram of the control device;

[0023] Figure 8 A schematic structural diagram of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0024] 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 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.

[0025] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or their collections. It should also be understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations.

[0026] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0027] References to "one embodiment" or "some embodiments" etc. described in the present specification mean that one or more embodiments of the present invention include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in some other embodiments", "in some other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0028] It should be understood that the order of execution of the steps in the following embodiments does not imply a precedence of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0029] In order to illustrate the technical solution of the present invention, specific embodiments are provided below for illustration.

[0030] The water treatment device control method provided by the embodiment of the present invention can be applied in Figure 1 In the water treatment system shown, the water treatment system includes a water treatment device and a control device, wherein the water treatment device communicates with the control device via a cable or a network.

[0031] like Figure 2As shown, the water quality treatment device includes a raw water pump, a raw water regulating tank, a flocculation reaction tank, a sedimentation tank, a filter tank, a clean water tank, a water supply pump, and a sludge treatment device connected to the sedimentation tank pipeline in sequence. The controller extracts water (influent) from the water source to the raw water regulating tank through the raw water pump, and then adds flocculant to the outlet of the raw water regulating tank, so that the influent undergoes flocculation reaction in the flocculation reaction tank, and then outputs it to the sedimentation tank for sedimentation, and outputs the upper treated water that has completed sedimentation to the filter tank for filtration. After the filtration is completed, the entire water quality treatment process is completed, and the clean water obtained by filtration is output to the clean water tank, and then the clean water tank is transported to the user through the water supply pump. Among them, after sedimentation in the sedimentation tank, the sludge is output to the sludge treatment device for sludge treatment, and the separated sludge is compressed and then removed centrally, and the treated water of the sludge at the separation may continue to flow back to the flocculation reaction tank for another flocculation reaction and sedimentation to achieve reuse. Among them, a water inlet collection device is set at the measuring point of the raw water regulating tank outlet to collect water inlet data, and a turbidity measuring device is set at the measuring point of the filter tank outlet or the clean water tank inlet to measure the outlet turbidity of the water quality treatment device.

[0032] It should be understood that, for the existing flocculant dosing method, relevant personnel usually rely on experience to judge the water quality of the inlet water, and then determine the flocculant dosage. The manual determination of the flocculant dosage method depends on human experience and technical level, and it is easy to have too little or too much dosage, resulting in unstable or substandard turbidity of the effluent (the effluent of the filter tank). Some water plants use programmable logic controllers (PLCs) to automatically control flocculants, but usually set a target effluent turbidity value, and adjust the flocculant dosage according to the actual effluent turbidity. The flocculant dosing logic is simple and rough. It can be seen that the flocculant addition logic of these two methods is relatively simple, and the flocculant dosage cannot be accurately controlled, resulting in large differences in effluent water quality and poor effluent water quality stability; and in PLC control, the actual effluent turbidity compared with the target effluent turbidity value is usually the effluent turbidity after adding flocculant and reacting for a certain period of time, which will cause the control of flocculant dosage to have a lag and low accuracy.

[0033] To solve the above problems, in this embodiment, Figure 2As shown, a turbidity prediction model is obtained by training a neural network model based on multiple historical water quality treatment samples at different historical moments and the water quality treatment time required for the flocculation process. Among them, the water quality treatment time required for the flocculation process is the water quality treatment time calibrated by the water quality treatment device. In actual use, the control device collects data from the water quality treatment device based on the water quality treatment time, obtains multiple sets of target water inlet data and multiple flocculant dosages at the first moment and the second moment, and the time from the first moment to the second moment is the water quality treatment time; then, based on multiple sets of target water inlet data and multiple flocculant dosages, the turbidity prediction model is used to predict the effluent turbidity to obtain the predicted effluent turbidity at the second moment; finally, based on the predicted effluent turbidity and the actual effluent turbidity at the second moment in the water quality treatment device, the water quality treatment device is controlled to feedback and adjust the flocculant dosage.

[0034] The first moment is the moment before the first moment. Taking the second moment as the current moment as an example, in actual use, the control device collects data on the water quality treatment device based on the water quality treatment time, and obtains multiple groups of target water inlet data and multiple flocculant dosages from the previous first moment to the current moment; then based on the multiple groups of target water inlet data and multiple flocculant dosages, the turbidity prediction model is used to predict the effluent turbidity to obtain the predicted effluent turbidity at the current moment; finally, based on the predicted effluent turbidity at the current moment and the actual effluent turbidity at the current moment, the water quality treatment device is controlled to feedback adjust the flocculant dosage.

[0035] In this embodiment, a high-precision turbidity prediction model is trained through historical water quality treatment samples and the water quality treatment time required for the flocculation process, and multiple groups of target water inlet data and flocculant dosage within the water quality treatment time at the second moment and before are used as input. The turbidity of the effluent is predicted by the turbidity prediction model, so that the predicted effluent turbidity at the second moment is close to the effluent turbidity predicted at its most recent historical moment, thereby improving the accuracy and stability of the predicted effluent turbidity. Based on the actual effluent turbidity at the second moment and the actual effluent turbidity, the flocculant dosage is feedback-adjusted, and the flocculant dosage can be quickly adjusted to the dosage that meets the water quality treatment requirements, thereby improving the accuracy of flocculant dosage control, ensuring that the actual effluent turbidity is maintained near the target effluent turbidity, and improving the stability of the effluent water quality.

[0036] In addition, the target water inlet data from the water quality treatment time before the second moment to the second moment is used to predict the effluent turbidity. The predicted effluent turbidity will take the water quality treatment time into consideration. Therefore, when the flocculant dosage is feedback-adjusted, the actual effluent turbidity and the predicted effluent turbidity at the same moment can be used for feedback-adjustment. The flocculant dosage can be adjusted in real time instead of using the actual effluent turbidity for a period of time after the second moment. This solves the problem of lag in the existing flocculant dosage adjustment method. The adjustment method in this embodiment is more timely and accurate, which further improves the accuracy of the flocculant dosage, thereby further improving the stability of the effluent water quality.

[0037] In this embodiment, the water treatment system includes a water treatment device and a control device. The specific structure of the water treatment device is only for illustrative purposes. In other embodiments, the water treatment system also includes other necessary devices; the water treatment device may also include other necessary devices, such as flocculant dosing equipment, i.e., automated dosing equipment. The control device may be a controller, such as an electronic control unit, or may be, but not limited to, various personal computers, laptops, smart phones, tablet computers, and portable wearable devices.

[0038] In one embodiment, if Figure 3 As shown, a water quality treatment device control method is provided, and the method is applied in Figure 1 The control device in FIG. 1 is taken as an example to illustrate, and the steps are as follows:

[0039] S10: Pre-training to obtain a turbidity prediction model.

[0040] In the process of treating tap water (wastewater) by a water treatment device, a turbidity prediction model needs to be trained in advance so that the model can be used to predict the turbidity of the effluent water later.

[0041] The turbidity prediction model is a neural network model trained based on multiple historical water quality treatment samples of the water quality treatment device at different historical moments and the water quality treatment time required for the flocculation process. The water quality treatment time required for the flocculation process is the time required in the water quality treatment device from the input of the flocculant to the completion of the water quality filtration in the filter tank. The water quality treatment time is usually a pre-calibrated value of the water quality treatment device. The prediction accuracy of the turbidity prediction model can be improved by training the model with multiple real historical water quality treatment samples of the water quality treatment device and the water quality treatment time required for the flocculation process.

[0042] S20: Collecting data from the water quality treatment device based on the water quality treatment time to obtain multiple groups of target water inflow data and multiple flocculant dosages at the first moment and the second moment.

[0043] When the water treatment device performs water treatment, the control device needs to collect water inflow data in the water treatment device based on the water treatment time, obtain target water inflow data at the first moment and the second moment, that is, obtain multiple sets of target water inflow data at different moments, and collect the flocculant dosage actually used by the water treatment device to obtain the flocculant dosage from the first moment to the second moment, that is, obtain multiple flocculant dosages at different moments. Among them, the time from the first moment to the second moment is the water treatment time.

[0044] It should be understood that, during the operation of the water treatment system, the second moment in this embodiment is each control node of the control device, that is, the current moment of the control device during the operation.

[0045] For example, the control device can collect the target water inflow data in the water quality treatment device through the water inflow collection device to obtain the target water inflow data at different times, and then use the water quality treatment time as the time window and the second moment (current moment) as the starting position, and use the sliding time window algorithm to intercept the target water inflow data at different times before the second moment, and quickly intercept the data within a time window (water quality treatment time), that is, all the target water inflow data from the first moment to the second moment, and then take the target water inflow data at the same moment as a group to obtain multiple groups of target water inflow data, which is simple and convenient. Similarly, use the water quality treatment time as the time window and the second moment as the starting position, and use the sliding time window algorithm to intercept the flocculant dosage at different times before the second moment, and obtain multiple flocculant dosages from the first moment to the second moment, which is simple and convenient.

[0046] S30: Based on multiple groups of target influent data and multiple flocculant dosages, the turbidity of the effluent is predicted using a turbidity prediction model to obtain a predicted effluent turbidity.

[0047] Then, the control device needs to predict the effluent turbidity using the turbidity prediction model based on multiple sets of target influent data and multiple flocculant dosages within the water quality treatment time before the second moment, and obtain the predicted effluent turbidity corresponding to the second moment.

[0048] For example, the target influent data and flocculant dosage at different times can be directly input into the turbidity prediction model to predict the effluent turbidity, and the predicted effluent turbidity corresponding to the second time can be obtained, which is simple and convenient.

[0049] S40: Based on the predicted effluent turbidity and the actual effluent turbidity of the water treatment device at the second moment, the water treatment device is controlled to perform feedback adjustment on the dosage of the flocculant.

[0050] After controlling the water treatment device to input flocculant, the control device needs to control the water treatment device to feedback adjust the amount of flocculant added at the second moment (or the second moment and its subsequent moments) based on the predicted effluent turbidity and the actual effluent turbidity of the water treatment device at the second moment.

[0051] For example, the control device can collect the real-time effluent turbidity of the water quality treatment device, and then use the predicted effluent turbidity as the target and the actual effluent turbidity at the second moment as feedback to control the dosing equipment of the water quality treatment device to feedback and adjust the dosage of the flocculant. Specifically, when the actual effluent turbidity is greater than the predicted effluent turbidity, the dosage of the flocculant is increased; when the actual effluent turbidity is less than the predicted effluent turbidity, the dosage of the flocculant is reduced; when the actual effluent turbidity is equal to the predicted effluent turbidity, no adjustment is required to ensure the stability of the effluent turbidity.

[0052] In this embodiment, a high-precision turbidity prediction model is trained through historical water quality treatment samples and the water quality treatment time required for the flocculation process, and multiple groups of target water inlet data and flocculant dosage within the water quality treatment time at the second moment and before are used as input. The turbidity of the effluent is predicted by the turbidity prediction model, so that the predicted effluent turbidity is close to the historically predicted effluent turbidity, thereby improving the accuracy and stability of the predicted effluent turbidity. Based on the target predicted effluent turbidity at the second moment and the actual effluent turbidity, the flocculant dosage is feedback-adjusted, and the flocculant dosage can be quickly adjusted to the dosage that meets the water quality treatment requirements, thereby improving the accuracy of flocculant dosage control, ensuring that the actual effluent turbidity is maintained near the target effluent turbidity, and improving the stability of the effluent water quality. In addition, by using the actual effluent turbidity and the predicted effluent turbidity at the same time for feedback adjustment, real-time flocculant dosage adjustment can be performed, which solves the problem of lag in the existing flocculant dosage adjustment method. The adjustment method in this embodiment is more timely and accurate, further improving the accuracy of flocculant dosage, thereby further improving the stability of effluent water quality.

[0053] In one embodiment, if Figure 4 As shown, in step S30, based on multiple groups of target influent data and multiple flocculant dosages, the turbidity of the effluent is predicted using a turbidity prediction model to obtain the predicted effluent turbidity, which specifically includes the following steps:

[0054] S31: Based on the target water inflow data of the water treatment device at the second moment, the dosage of flocculant is predicted using a dosage prediction model to obtain the predicted dosage at the second moment.

[0055] It is important to understand that when collecting data on the water treatment device based on the water treatment duration, since no flocculant was actually added at the second moment (current moment), it is impossible to collect the flocculant dosage at that moment, and only the flocculant dosage at the historical moment can be collected. Therefore, it is necessary to use the target water inflow data at the second moment to predict the flocculant dosage, and use the prediction result as the predicted dosage at the second moment to improve the data input of the subsequent turbidity prediction model.

[0056] Among them, the control device needs to obtain a pre-trained dosage prediction model, which is a prediction model obtained by deep learning of the extreme gradient boosting tree classifier based on historical water inlet data and historical flocculant dosage in multiple historical water quality treatment samples.

[0057] After obtaining multiple sets of target water inflow data of the water quality treatment device, the control device needs to determine the target water inflow data at the second moment from the multiple sets of target water inflow data, and then use the dosage prediction model to predict the flocculant dosage to obtain the predicted dosage.

[0058] S32: Based on multiple groups of target influent data, multiple flocculant dosages and the predicted dosage at the second moment, the turbidity of the effluent is predicted using a turbidity prediction model to obtain a predicted effluent turbidity.

[0059] After obtaining the predicted dosage, the control device uses the turbidity prediction model to predict the effluent turbidity based on multiple groups of target water inflow data, multiple flocculant dosages and the predicted dosage at the second moment, and obtains the predicted effluent turbidity at the second moment. That is, the predicted dosage is used as the flocculant dosage at the second moment, thereby obtaining multiple flocculant dosages from the first moment to the second moment, and then the target water inflow data and flocculant dosage at the same moment are taken as a data pair, and the data pairs at different moments are input into the turbidity prediction model together to predict the effluent turbidity, and obtain the predicted effluent turbidity.

[0060] In one embodiment, before using the turbidity prediction model to predict the effluent turbidity, it is necessary to determine whether there is an abnormal flocculant dosage among the multiple flocculant dosages from the preset historical moment to the second moment. When the flocculant dosage is a null value, a 0 value or an abnormal value (i.e., a mutation value or a value greater than the calibration value), it indicates that the flocculant dosage is an abnormal flocculant dosage. When there is an abnormal flocculant dosage among the multiple flocculant dosages, based on the target water inflow data at the moment of the abnormal flocculant dosage, the flocculant dosage is predicted using the dosage prediction model to obtain the corresponding predicted dosage, and the abnormal flocculant dosage is replaced by the corresponding predicted dosage, thereby obtaining multiple flocculant dosages after treatment. Finally, based on multiple groups of target water inflow data, multiple flocculant dosages after treatment, and the predicted dosage at the second moment, the turbidity of the effluent is predicted using the turbidity prediction model to obtain the predicted effluent turbidity.

[0061] In this embodiment, based on the target water inflow data at the second moment in the water treatment device, the dosage prediction model is used to predict the flocculant dosage to obtain the predicted dosage, and then based on multiple sets of target water inflow data, multiple flocculant dosages and the predicted dosage at the second moment, the turbidity prediction model is used to predict the effluent turbidity to obtain the predicted effluent turbidity. The process of using the turbidity prediction model to predict the effluent turbidity is refined, and the predicted dosage predicted by the dosage prediction model and the multiple flocculant dosages within the previous water treatment time are jointly input into the turbidity prediction model, so that the predicted effluent turbidity at the second moment can be predicted more accurately.

[0062] In one embodiment, step S31, i.e., based on the target water inflow data at the second moment in the water treatment device, the dosage of flocculant is predicted using the dosage prediction model to obtain the predicted dosage at the second moment, specifically comprising the following steps:

[0063] S311: Standardize the target water inflow data at the second moment to obtain standard water inflow data at the second moment.

[0064] It is important to understand that when training a dosage prediction model, the training data needs to be standardized to unify data of different magnitudes into the same magnitude, so as to avoid the impact of different metrics and value ranges between features on data analysis and improve the accuracy of the training model. In other words, in order to ensure that the model can smoothly perform data recognition, the input data of the dosage prediction model needs to be standardized.

[0065] That is, it is necessary to perform standardization on the target influent data of the water quality treatment device at the second moment to obtain the standardized target influent data at the second moment, i.e., the standard influent data. The target influent data may include multiple influent parameters such as influent flow rate, influent pH, influent temperature, influent dissolved oxygen, influent turbidity, and chemical oxygen demand, and the standard influent data includes multiple influent parameters such as standardized influent flow rate, influent pH, influent temperature, influent dissolved oxygen, influent turbidity, and chemical oxygen demand.

[0066] Specifically, it is necessary to obtain the preset mean and preset standard deviation of each inlet parameter in the target inlet data, wherein the preset mean and preset standard deviation of each inlet parameter are the mean and standard deviation of each inlet parameter obtained by performing mean processing and standard deviation calculation on each inlet parameter of all historical inlet data in multiple historical water quality treatment samples when training the dosage prediction model. Then, based on the preset mean and preset standard deviation of each inlet parameter, standard processing is performed on each inlet parameter in the target inlet data to obtain standard inlet data including multiple standardized inlet parameters.

[0067] Among them, the standardization is performed by the following formula:

[0068] z = (x-mean) / std;

[0069] Among them, z represents the standardized inlet parameter, that is, the inlet parameter in the standard inlet data; x represents the inlet parameter of the target inlet data; mean represents the preset mean of the inlet parameter; std represents the preset standard deviation of the inlet parameter.

[0070] In other embodiments, the preset mean value and the preset standard deviation of the water inlet parameter may also be a pre-calibrated constant.

[0071] S312: Input the standard water inflow data at the second moment into the dosage prediction model for prediction to obtain an initial predicted dosage.

[0072] After obtaining the standard water inflow data at the second moment, the standard water inflow data at the second moment is input into the dosage prediction model to predict the flocculant dosage, and obtain the initial predicted dosage at the second moment.

[0073] S313: De-normalize the initial predicted dosage to obtain the predicted dosage.

[0074] After obtaining the initial predicted dosage at the second moment, it is necessary to perform a denormalization process on the initial predicted dosage at the second moment to obtain the predicted dosage at the second moment. Specifically, it is necessary to obtain a preset dosage mean and dosage standard deviation, which are the mean and standard deviation obtained by performing mean processing and standard deviation calculation on all historical dosages of flocculants in multiple historical water quality treatment samples when training the dosage prediction model. Then, based on the preset dosage mean and dosage standard deviation, the initial predicted dosage is denormalized to obtain the predicted dosage at the second moment.

[0075] Among them, the predicted dosage at the second moment is calculated by the following formula, that is, the calculation formula for the denormalization process is as follows:

[0076] x = z * std + mean;

[0077] Among them, x represents the predicted dosage at the second moment; z represents the initial predicted dosage at the second moment; mean represents the preset mean dosage; std represents the preset standard deviation of dosage.

[0078] When training the dosage prediction model, the dosage mean and dosage standard deviation of the training data are de-standardized to make the processing standards of the model input and output consistent, further improving the accuracy of the predicted dosage.

[0079] In this embodiment, the target water inlet data at the second moment is standardized to obtain the standard water inlet data at the second moment, and then the standard water inlet data at the second moment is input into the dosage prediction model to predict the flocculant dosage to obtain the initial predicted dosage at the second moment, and the initial predicted dosage at the second moment is de-standardized to obtain the predicted dosage at the second moment; it clarifies the specific process of using the dosage prediction model to predict the predicted dosage at the second moment, and first standardizes the model input to improve the prediction accuracy, and then de-standardizes the model output to restore the predicted dosage to the real data representation, thereby improving the accuracy of the predicted dosage.

[0080] In one embodiment, in step S32, based on multiple sets of target influent data, multiple flocculant dosages and the predicted dosage at the second moment, the turbidity of the effluent is predicted using a turbidity prediction model to obtain the predicted effluent turbidity, which specifically includes the following steps:

[0081] S321: performing standardization processing on each data in each target water inflow data to obtain multiple groups of standard water inflow data, and performing standardization processing on the predicted dosage at the second moment and each flocculant dosage to obtain multiple standard flocculant dosages.

[0082] The control device needs to perform standardization processing on each data (i.e. each inlet parameter) in each set of target inlet data to obtain multiple sets of standard inlet data. Each set of target inlet data includes multiple inlet parameters such as inlet flow rate, inlet pH, inlet temperature, inlet dissolved oxygen, inlet turbidity (and chemical oxygen demand), etc., and each set of standard inlet data includes standardized inlet flow rate, inlet pH, inlet temperature, inlet dissolved oxygen, inlet turbidity (and chemical oxygen demand), etc.

[0083] Specifically, it is necessary to obtain the preset mean and preset standard deviation of each influent parameter in the target influent data, wherein the preset mean and preset standard deviation of each influent parameter are the mean and standard deviation of each influent parameter obtained by performing mean processing and standard deviation calculation on each influent parameter of all historical influent data in multiple historical water quality treatment samples when training the turbidity prediction model. Then, based on the preset mean and preset standard deviation of each influent parameter, standard processing is performed on each influent parameter in each group of target influent data to obtain standard influent data including multiple standardized influent parameters, and all target influent data are traversed to obtain multiple groups of standard influent data.

[0084] At the same time, the predicted dosage at the second moment and the dosage of each flocculant are standardized to obtain multiple standard flocculant dosages. Specifically, it is necessary to obtain a preset dosage mean and dosage standard deviation, which are the mean and standard deviation obtained by performing mean processing and standard deviation calculation on all historical flocculant dosages in multiple historical water quality treatment samples when training the turbidity prediction model. Then, based on the preset dosage mean and dosage standard deviation, the predicted dosage at the second moment and the dosage of each flocculant are standardized to obtain multiple standard flocculant dosages.

[0085] The calculation formulas for each influent parameter and each standard flocculant dosage in each set of standard influent data can refer to the calculation formula for the standardization process above, and will not be repeated here.

[0086] S322: Input multiple sets of standard influent data and multiple standard flocculant dosages into a turbidity prediction model to predict effluent turbidity and obtain an initial predicted turbidity.

[0087] Then, multiple groups of standard influent data and multiple standard flocculant dosages are input into the turbidity prediction model to predict the effluent turbidity, and the initial predicted turbidity at the second moment is obtained.

[0088] S323: Denormalize the initial predicted turbidity to obtain the predicted effluent turbidity.

[0089] After obtaining the initial predicted turbidity at the second moment, the initial predicted turbidity at the second moment is denormalized to obtain the predicted effluent turbidity at the second moment. It is necessary to obtain a preset turbidity mean and turbidity standard deviation, which are the mean and standard deviation obtained by performing mean processing and standard deviation calculation on all historical effluent turbidities in multiple historical water quality treatment samples when training the turbidity prediction model.

[0090] Among them, the predicted effluent turbidity is calculated by the following formula:

[0091] x = z * std + mean;

[0092] Among them, x represents the predicted effluent turbidity; z represents the initial predicted turbidity; mean represents the preset turbidity mean; std represents the preset turbidity standard deviation.

[0093] The mean and standard deviation of the dosage of the training data were used to perform de-standardization when training the turbidity prediction model, so that the processing standards of the model input and output were consistent, further improving the accuracy of predicting the effluent turbidity.

[0094] In this embodiment, each data in each target water inflow data is standardized to obtain multiple groups of standard water inflow data, and the predicted dosage and each flocculant dosage are standardized to obtain multiple standard flocculant dosages, and then the multiple groups of standard water inflow data and the multiple standard flocculant dosages are input into the turbidity prediction model to predict the effluent turbidity to obtain the initial predicted turbidity, and then the initial predicted turbidity is de-standardized to obtain the predicted effluent turbidity; it clarifies the specific process of obtaining the predicted effluent turbidity, first standardizing the model input can improve the prediction accuracy, and then de-standardizing the model output can restore the predicted effluent turbidity to the real data representation, thereby improving the accuracy of the predicted effluent turbidity.

[0095] In one embodiment, if Figure 5 As shown, in step S40, based on the predicted effluent turbidity and the actual effluent turbidity of the water treatment device at the second moment, the water treatment device is controlled to perform feedback adjustment on the dosage of the flocculant, which specifically includes the following steps:

[0096] S41: When the actual effluent turbidity at the second moment meets the water quality treatment requirement, the predicted dosage is adjusted based on the deviation between the predicted effluent turbidity and the actual effluent turbidity.

[0097] In this embodiment, after the predicted effluent turbidity and the actual effluent turbidity of the water quality treatment device at the second moment are obtained, it is necessary to determine whether the actual effluent turbidity at the second moment meets the water quality treatment requirement. For example, the water quality treatment requirement requires that the effluent turbidity of the water quality treatment device be within a turbidity range. If the actual effluent turbidity at the second moment is within the turbidity range, it is determined that the actual effluent turbidity at the second moment meets the water quality treatment requirement; if the actual effluent turbidity at the second moment is not within the turbidity range, it is determined that the actual effluent turbidity at the second moment does not meet the water quality treatment requirement.

[0098] When determining that the actual effluent turbidity at the second moment meets the water quality treatment requirements, it is necessary to obtain the predicted dosage at the second moment. The predicted dosage is based on the target inlet data at the second moment, and the dosage prediction model is used to predict the flocculant dosage to obtain the dosage; then, based on the deviation between the predicted effluent turbidity and the actual effluent turbidity, the predicted dosage at the second moment is adjusted to obtain the adjusted predicted dosage.

[0099] When it is determined that the actual effluent turbidity at the second moment does not meet the water quality treatment requirements, it is necessary to obtain a pre-calibrated target effluent turbidity, which is within the above turbidity range required by the water quality treatment device. The target effluent turbidity can be the minimum value or median value within the turbidity range. Then, based on the deviation between the predicted effluent turbidity and the target effluent turbidity, the predicted dosage at the second moment is adjusted to obtain the adjusted predicted dosage.

[0100] Among them, the predicted dosage is adjusted to determine whether the absolute value of the deviation between the predicted effluent turbidity and the actual effluent turbidity at the second moment (target effluent turbidity) is greater than a preset value. If it is greater than the preset value, when the predicted effluent turbidity is greater than the actual effluent turbidity at the second moment (target effluent turbidity), a certain dosage is added on the basis of the predicted dosage to obtain the adjusted predicted dosage, wherein the increased dosage increases as the absolute value of the deviation between the two increases; when the predicted effluent turbidity is less than the actual effluent turbidity at the second moment (target effluent turbidity), a certain dosage is reduced on the basis of the predicted dosage to obtain the adjusted predicted dosage, wherein the reduced dosage decreases as the absolute value of the deviation between the two increases; so that the adjusted predicted dosage is closer to the flocculant dosage corresponding to the actual effluent turbidity (target effluent turbidity) at the second moment, which ensures the stability of the effluent turbidity and reduces the flocculant dosage cost.

[0101] S43: Based on the adjusted predicted dosage, multiple groups of target influent data and multiple flocculant dosages, the turbidity prediction model is used to re-predict the effluent turbidity until the absolute value of the deviation is less than the preset value, and the adjusted predicted dosage is output as the target dosage.

[0102] After the predicted dosage at the second moment is adjusted, the turbidity prediction model is used to re-predict the effluent turbidity based on the adjusted predicted dosage, multiple groups of target influent data and multiple flocculant dosages to obtain a new predicted effluent turbidity, and then the deviation between the new predicted effluent turbidity and the actual effluent turbidity (or target effluent turbidity) is judged. When the absolute value of the deviation between the two is greater than the preset value (such as 0, 0.1, 0.2), steps S42-S43 are repeated, that is, it is necessary to continue to adjust the dosage based on the adjusted predicted dosage to obtain the adjusted predicted dosage, and then the turbidity prediction model is used to re-predict the effluent turbidity until the absolute value of the deviation between the two is less than or equal to the preset value, then the adjustment is stopped, and the latest adjusted predicted dosage is output as the target dosage.

[0103] S44: Taking the target dosage as the actual dosage of flocculant at the second moment, and controlling the water treatment device to add flocculant.

[0104] After obtaining the target dosage, the control device needs to use the target dosage as the actual flocculant dosage at the second moment, and control the water quality treatment device to input the flocculant.

[0105] In this embodiment, when the actual effluent turbidity at the second moment meets the water quality treatment demand, the predicted dosage is adjusted based on the deviation between the predicted effluent turbidity and the actual effluent turbidity; based on the adjusted predicted dosage, multiple groups of target inlet data and multiple flocculant dosages, the turbidity prediction model is used to re-predict the effluent turbidity until the absolute value of the deviation is less than the preset value, the adjusted predicted dosage is output as the target dosage, and the target dosage is used as the actual flocculant dosage at the second moment to control the water quality treatment device to input flocculant. According to the deviation between the predicted effluent turbidity and the actual effluent turbidity, the predicted dosage is adjusted, and it is iteratively adjusted until the absolute value of the deviation between the two is less than the preset value, and the adjusted predicted dosage is output as the target dosage, and the water quality treatment device is controlled to input flocculant, which can ensure that the current flocculant dosage is consistent or similar to the dosage corresponding to the actual effluent turbidity at the second moment, that is, the subsequent effluent turbidity can be made consistent or similar to the current effluent turbidity, ultimately achieving the purpose of saving flocculant and maintaining stable effluent turbidity.

[0106] In one embodiment, in order to ensure the effect of subsequent model training and improve the prediction accuracy of the turbidity prediction model and the dosage prediction model, it is necessary to pre-process the historical data collected at each measuring point in the water quality treatment device, so as to obtain multiple historical water quality treatment samples of the water quality treatment device at different historical moments, and then use the multiple historical water quality treatment samples at different historical moments for training to obtain the turbidity prediction model and the dosage prediction model. Specifically, Figure 6 As shown, multiple historical water quality treatment samples were obtained in the following ways:

[0107] S01: Obtain the historical data collected at each measuring point in the water quality treatment device, and perform standardization processing to obtain the influent water quality data, historical flocculant dosage and historical effluent turbidity at different historical moments.

[0108] Among them, it is necessary to first set up multiple measuring points in the water quality treatment device, so as to collect water quality data at various locations in the water quality treatment device through water quality collection devices at the measuring points. Figure 2 As shown, a measuring point can be set at the inlet or outlet of the raw water regulating tank, and the water inlet data can be collected through the water quality collection device; a measuring point can be set at the inlet of the flocculation reaction tank, and the actual flocculant dosage can be collected through the flocculant measuring device, which is more accurate, and the flocculant input controlled by the control device can also be directly obtained; and a measuring point can be set at the filter tank outlet or the clean water tank inlet, and the actual effluent turbidity of the water quality treatment device can be collected through the turbidity measuring device. During the operation of the water quality treatment device, the data of the above-mentioned measuring points need to be collected and maintained in real time to facilitate subsequent calls, such as model training, model update and data analysis.

[0109] When model training or model updating is required, the historical data collected from each measuring point in the water quality treatment device are obtained, and the historical data collected from each measuring point include the water inlet measuring point data, flocculant dosage and effluent turbidity at different historical moments. Then, the historical data collected at different historical moments are cleaned, and the 0 value, null value and abnormal value in each data are removed to obtain the historical data collected at different historical moments after deep cleaning, that is, the water inlet measuring point data, flocculant dosage and effluent turbidity at different historical moments are obtained. Among them, the abnormal value can be detected and determined by algorithms such as box plot and local outlier factor (LOF).

[0110] Then, the inlet measurement point data, flocculant dosage and effluent turbidity at different historical moments after cleaning were standardized to obtain the inlet water quality data, historical flocculant dosage and historical effluent turbidity at different historical moments. Standardization can unify data of different magnitudes into the same magnitude to avoid the impact of different measurements and value ranges between features on data analysis, and improve the convenience of subsequent data processing and the effect of model training.

[0111] Specifically, each parameter in the water inlet measurement point data at different historical moments is standardized to obtain the water quality data of the water inlet at different historical moments, and the flocculant dosage at different historical moments is standardized to obtain the historical flocculant dosage at different historical moments, and the effluent turbidity at different historical moments is standardized to obtain the historical effluent turbidity at different historical moments. The standardization process is described above and will not be repeated here.

[0112] S02: Screening target inlet water parameters for inlet water quality data at different historical moments, and obtaining historical inlet water data at different historical moments.

[0113] Then, it is necessary to screen the target influent parameters of the influent water quality data at different historical moments, and obtain the historical influent data at different historical moments by screening. The target influent parameters are the influent parameters that affect the flocculant dosage and / or the effluent turbidity.

[0114] During data collection, a large number of influent parameters are collected in the influent water quality data. Some parameters (target influent parameters) may be strongly correlated with the flocculant dosage and / or the effluent turbidity, while some influent parameters may have little effect on the flocculant dosage and / or the effluent turbidity. In order to avoid model overfitting and improve the model training effect, it is necessary to eliminate some unnecessary influent parameters (influent parameters other than the target influent parameters) in the influent water quality data and retain the target influent parameters in the influent water quality data, so as to screen out the historical influent data at different historical moments.

[0115] Among them, the target water inlet parameter can be a water inlet parameter calibrated according to historical experience, such as water inlet flow, water inlet temperature, water inlet turbidity, etc. In other embodiments, in order to improve data accuracy, the water inlet parameters in the water quality data can also be subjected to correlation analysis with the flocculant dosage and the effluent turbidity, so as to determine the target water inlet parameter that can affect the flocculant dosage and / or the effluent turbidity, and then screen the target water inlet parameter, retain the target water inlet parameter of the water quality data, and eliminate other water inlet parameters, so as to obtain the historical water inlet data at different historical moments.

[0116] For example, the inlet water quality data include inlet flow rate, inlet pH value, inlet temperature, inlet dissolved oxygen, inlet turbidity, chemical oxygen demand (COD), inlet total phosphorus (total phosphorus content in inlet water), inlet total nitrogen (total nitrogen content in inlet water), and inlet residual chlorine (residual chlorine in inlet water).

[0117] Based on the influent water quality data, historical flocculant dosage and historical effluent turbidity at different historical moments, the correlation analysis is performed on each influent parameter and the flocculant dosage to obtain the correlation between each influent parameter and the flocculant dosage, and the correlation analysis is performed on each influent parameter and the effluent turbidity to obtain the correlation between each influent parameter and the effluent turbidity, that is, the correlation analysis data is obtained, that is, the correlation analysis data includes the correlation between each influent parameter and the flocculant dosage in the influent water quality data, and / or the correlation between each influent parameter and the effluent turbidity. Among them, the value range of the correlation is [0, 1], and the larger the value, the stronger the correlation.

[0118] According to the historical data verification analysis, the correlation of the parameters that have a greater impact on the model must be 0.7 or above. According to the correlation between each influent parameter and the amount of flocculant added, the influent parameters with a correlation greater than or equal to 0.7 include influent flow, influent pH, influent temperature, influent dissolved oxygen, influent turbidity and chemical oxygen demand COD; that is, the target parameters can include influent flow, influent pH, influent temperature, influent dissolved oxygen, influent turbidity and chemical oxygen demand COD. Correspondingly, according to the correlation between each influent parameter and the effluent turbidity, the influent parameters with a correlation greater than or equal to 0.7 include influent flow, influent pH, influent temperature, influent dissolved oxygen, influent turbidity and chemical oxygen demand COD; that is, the target parameters can include influent flow, influent pH, influent temperature, influent dissolved oxygen and influent turbidity.

[0119] According to the correlation between each influent parameter and the flocculant dosage, and the correlation between each influent parameter and the effluent turbidity, it can be seen that the correlation between the three influent parameters, namely, the influent total phosphorus, the influent total nitrogen, and the influent residual chlorine, and the flocculant dosage and the effluent turbidity are all less than 0.2, and the correlation between the chemical oxygen demand COD and the effluent turbidity is less than 0.3.

[0120] If the dosage prediction model and the turbidity prediction model are trained using the same sample data, the influent total phosphorus, influent total nitrogen, influent residual chlorine, and chemical oxygen demand (COD) can be removed from the influent water quality data to obtain historical influent data, that is, historical influent data at different historical moments include influent flow, influent pH, influent temperature, influent dissolved oxygen, influent turbidity and other data.

[0121] If you want to improve the prediction accuracy of the dosage prediction model, the dosage prediction model and the turbidity prediction model can be trained using different sample data. The total phosphorus, total nitrogen, residual chlorine, and chemical oxygen demand (COD) of the influent water quality data are removed to obtain the first historical influent data at different historical moments, so that the first historical influent data can be used to train the turbidity prediction model later. The total phosphorus, total nitrogen, and residual chlorine of the influent water can be removed from the influent water quality data to obtain the second historical influent data at different historical moments, so that the second historical influent data can be used to train the dosage prediction model later.

[0122] S03: Combine the historical water inlet data, historical flocculant dosage and historical effluent turbidity at the same historical moment into a historical water quality treatment sample to obtain multiple historical water quality treatment samples.

[0123] After obtaining the historical water inflow data at different historical moments, the historical water inflow data, the historical flocculant dosage and the historical effluent turbidity at the same historical moment are combined into a historical water quality treatment sample to obtain multiple historical water quality treatment samples.

[0124] In one embodiment, the historical water inflow data includes first historical water inflow data and second historical water inflow data. After obtaining the historical water inflow data at different historical moments, the first historical water inflow data, the historical flocculant dosage, and the historical effluent turbidity at the same historical moment are combined into a historical water quality treatment sample to obtain a plurality of first historical water quality treatment samples for training the turbidity prediction model; the first historical water inflow data, the historical flocculant dosage, and the historical effluent turbidity at the same historical moment are combined into a historical water quality treatment sample to obtain a plurality of second historical water quality treatment samples for training the dosage prediction model. The historical water inflow data in the first historical water quality treatment sample includes the inflow flow rate, the inflow pH, the inflow temperature, the inflow dissolved oxygen content, and the inflow turbidity; the historical water inflow data in the second historical water quality treatment sample includes the inflow flow rate, the inflow pH, the inflow temperature, the inflow dissolved oxygen content, the inflow turbidity, and the chemical oxygen demand COD. Different model training data are determined according to different correlation analysis structures, which improves the accuracy of historical water quality treatment samples, and thus more accurate dosage prediction models and turbidity prediction models can be trained separately.

[0125] In this embodiment, the historical data collected from each measuring point in the water quality treatment device is obtained, and data cleaning is performed to filter out 0 values, null values, and abnormal values, so as to obtain the influent water quality data, historical flocculant dosage, and historical effluent turbidity at different historical moments; the influent water quality data at different historical moments are screened for target influent parameters, and the historical influent data at different historical moments are screened, and the target influent parameters are influent parameters that affect the flocculant dosage and / or effluent turbidity; the historical influent data, historical flocculant dosage, and historical effluent turbidity at the same historical moment are combined into a historical water quality treatment sample to obtain multiple historical water quality treatment samples. The pretreatment process of multiple historical water quality treatment samples is clarified, which can ensure the accuracy of the historical water quality treatment samples, thereby improving the model training effect, and further improving the prediction accuracy of the turbidity prediction model and the dosage prediction model.

[0126] In one embodiment, step S02, i.e., performing target water inflow parameter screening on the water quality data of the inflow at different historical moments, and screening to obtain the historical water inflow data at different historical moments, specifically includes the following steps:

[0127] S021: Conduct data correlation analysis on the influent water quality data, flocculant dosage and effluent turbidity value at different historical moments, and screen the influent parameters in the initial influent water quality data based on the correlation analysis data to obtain the target influent parameters whose correlation meets the requirements.

[0128] Among them, based on the influent water quality data at different historical moments, the historical flocculant dosage and the historical effluent turbidity, the correlation analysis is performed on each influent parameter and the flocculant dosage to obtain the correlation analysis data, wherein the correlation analysis data includes the correlation between each influent parameter and the flocculant dosage in the influent water quality data, and / or the correlation between each influent parameter and the effluent turbidity. Then, based on the correlation analysis data, the influent parameters in the initial influent water quality data are screened to obtain the target influent parameters whose correlation meets the requirements.

[0129] The specific steps include:

[0130] S0221: Determine the first correlation coefficient between each influent parameter and the flocculant dosage based on the historical flocculant dosage and influent water quality data at different historical moments.

[0131] According to the flocculant dosage and influent water quality data at different historical moments, the correlation between each influent parameter in the influent water quality data and the flocculant dosage is calculated to determine the correlation between each influent parameter and the flocculant dosage, that is, to determine the first correlation coefficient between each influent parameter and the flocculant dosage. The value range of the correlation is [0, 1], and the larger the value, the stronger the correlation.

[0132] S0222: Determine the second correlation coefficient between each inlet parameter and the effluent turbidity based on the historical effluent turbidity and inlet water quality data at different historical moments.

[0133] According to the historical effluent turbidity and influent water quality data at different historical moments, the correlation between each influent parameter in the influent water quality data and the effluent turbidity is calculated to determine the correlation between each influent parameter and the effluent turbidity, that is, to determine the second correlation coefficient between each influent parameter and the flocculant dosage.

[0134] S0223: Based on the first correlation coefficient and the second correlation coefficient, all water inlet parameters are screened to obtain a target water inlet parameter whose correlation coefficient is greater than a preset threshold.

[0135] Then, based on the first correlation coefficient and the second correlation coefficient, all water inlet parameters are screened to obtain target water inlet parameters with a correlation coefficient (correlation) greater than a preset threshold value (0.5 or 0.6 or 0.7). Among them, the water inlet parameters with correlation coefficients greater than the preset threshold value in the first correlation coefficient and the second correlation coefficient can be determined, thereby obtaining the target water inlet parameter, that is, the target water inlet parameter is the water inlet parameter with correlation coefficients greater than the preset threshold value in the first correlation coefficient and the second correlation coefficient. Taking the intersection of the two as the target water inlet parameter makes subsequent data processing simpler.

[0136] For example, the influent water quality data may include influent flow, influent pH value, influent temperature, influent dissolved oxygen, influent turbidity, chemical oxygen demand (COD), influent total phosphorus (total phosphorus content in influent), influent total nitrogen (total nitrogen content in influent), influent residual chlorine (residual chlorine in influent). According to the historical data verification analysis, the correlation of the parameters that have a greater impact on the model must be 0.7 or above. According to the correlation between each influent parameter and the amount of flocculant added (i.e., the first correlation coefficient), the influent parameters with a correlation greater than or equal to 0.7 include influent flow, influent pH value, influent temperature, influent dissolved oxygen, influent turbidity, and chemical oxygen demand (COD). Correspondingly, according to the correlation between each inlet parameter and the effluent turbidity (i.e., the second correlation coefficient), the inlet parameters with a correlation greater than or equal to 0.7 include inlet flow, inlet pH, inlet temperature, inlet dissolved oxygen, inlet turbidity, and chemical oxygen demand (COD). According to the correlation between each inlet parameter and the amount of flocculant added (i.e., the first correlation coefficient) and the correlation between each inlet parameter and the effluent turbidity (i.e., the second correlation coefficient), it can be seen that the correlation between the three inlet parameters, i.e., inlet total phosphorus, inlet total nitrogen, and inlet residual chlorine, and the flocculant addition and effluent turbidity, is less than 0.2, and the correlation between chemical oxygen demand and effluent turbidity is less than 0.3.

[0137] That is, the first correlation coefficient and the second correlation coefficient, the correlation coefficients of the inlet water parameters are both greater than the preset threshold values, including: inlet water flow rate, inlet water pH, inlet water temperature, inlet water dissolved oxygen content, inlet water turbidity, that is, the target inlet water parameters include inlet water flow rate, inlet water pH, inlet water temperature, inlet water dissolved oxygen content, and inlet water turbidity.

[0138] In other embodiments, the target water inlet parameter may include a first target water inlet parameter and a second target water inlet parameter. That is, the water inlet parameter whose correlation coefficient (correlation) in the first correlation coefficient is greater than the first preset threshold value can be determined to obtain the first target water inlet parameter; the water inlet parameter whose correlation coefficient (correlation) in the second correlation coefficient is greater than the second preset threshold value can be determined to obtain the second target water inlet parameter. Among them, the first preset threshold value and the second preset threshold value can be systematic or different. For example, the first target water inlet parameter includes the inlet flow rate, the inlet pH, the inlet temperature, the inlet dissolved oxygen content, the inlet turbidity and the chemical oxygen demand COD; the second target water inlet parameter includes the inlet flow rate, the inlet pH, the inlet temperature, the inlet dissolved oxygen content and the inlet turbidity.

[0139] S023: Parameter screening is performed on the historical water inflow data at different historical moments based on the target water inflow parameter to obtain the historical water inflow data at different historical moments.

[0140] Then, the historical water inflow data at different historical moments are screened based on the target water inflow parameters to obtain the historical water inflow data at different historical moments.

[0141] For example, if the dosage prediction model and the turbidity prediction model are trained using the same sample data, the inlet parameters with correlation coefficients greater than the preset threshold are selected as the target inlet parameters, including inlet flow, inlet pH, inlet temperature, inlet dissolved oxygen, and inlet turbidity; the inlet total phosphorus, inlet total nitrogen, inlet residual chlorine, and chemical oxygen demand can be removed from the inlet water quality data at different historical moments, and the inlet flow, inlet pH, inlet temperature, inlet dissolved oxygen, and inlet turbidity are retained, thereby obtaining historical inlet data at different historical moments. That is, the historical inlet data at different historical moments include inlet flow, inlet pH, inlet temperature, inlet dissolved oxygen, inlet turbidity, and other data.

[0142] In other embodiments, the target water inflow parameter may include a first target water inflow parameter and a second target water inflow parameter. When the parameters of the historical water inflow data at different historical moments are screened based on the target water inflow parameter, the parameters of the historical water inflow data at different historical moments are screened based on the first target water inflow parameter, the total phosphorus content, the total nitrogen content, and the residual chlorine content of the inflow are eliminated, and the inflow flow rate, the pH value of the inflow, the temperature of the inflow, the dissolved oxygen content, the turbidity of the inflow, and the chemical oxygen demand are retained to obtain the first historical water inflow data at different historical moments; at the same time, the parameters of the historical water inflow data at different historical moments are screened based on the second target water inflow parameter, the total phosphorus content, the total nitrogen content, the residual chlorine content, and the chemical oxygen demand of the inflow are eliminated, and the inflow flow rate, the pH value of the inflow, the temperature of the inflow, the dissolved oxygen content, and the turbidity of the inflow are retained to obtain the second historical water inflow data at different historical moments.

[0143] That is, the historical water inflow data includes first historical water inflow data and second historical water inflow data with different data. Subsequently, the dosage prediction model can be trained based on the first historical water inflow data, and the turbidity prediction model can be trained based on the second historical water inflow data, and different sample data can be used for model training to improve the prediction accuracy of the dosage prediction model.

[0144] In this embodiment, by performing data correlation analysis on the influent water quality data, flocculant dosage and effluent turbidity value, and screening the influent parameters in the influent water quality data based on the correlation analysis data, the target influent parameters whose correlation meets the requirements are obtained, and then the historical influent data at different historical moments are screened based on the target influent parameters to obtain the historical influent data at different historical moments, so as to obtain more accurate model training data and improve the model training effect.

[0145] In one embodiment, before step S31 or S41, that is, before the dosage prediction model is used to predict the dosage of flocculant and the preset dosage is obtained, a neural network model needs to be trained based on multiple historical water quality treatment samples at different historical moments to obtain a dosage prediction model, so that the dosage prediction model can be used to predict the dosage of flocculant later. Before step S31, the turbidity prediction model is trained in the following way:

[0146] S301: extracting data from a plurality of historical water quality treatment samples at different historical moments to obtain training sample groups at different historical moments, wherein the training sample groups include historical water inflow data and historical flocculant dosage at the same historical moment.

[0147] After preprocessing the historical data collected at each measuring point in the water quality treatment device, multiple historical water quality treatment samples at different historical moments are obtained. Each historical water quality treatment sample includes historical influent data, historical flocculant dosage, and historical effluent turbidity. Then, data extraction is performed on multiple historical water quality treatment samples at different historical moments to obtain training sample groups at different historical moments, and the training sample groups include historical influent data and historical flocculant dosage at the same historical moment.

[0148] In one embodiment, the training sample group includes the first historical water inflow data and the historical flocculant dosage at the same historical moment, so as to improve the data diversity and accuracy of the training sample group, thereby improving the model training effect. That is, the historical water inflow data in the historical water quality treatment sample includes the water inflow flow, water pH, water temperature, water dissolved oxygen content, water turbidity and chemical oxygen demand. In other embodiments, the historical water inflow data in the historical water quality treatment sample may also include water inflow flow, water pH, water temperature, water dissolved oxygen content and water turbidity.

[0149] S302: inputting the historical water inflow data in the training sample group into the extreme gradient boosting tree classifier to predict the flocculant dosage, obtaining the predicted flocculant amount, and determining the model loss value based on the predicted flocculant amount and the historical flocculant dosage in the training sample group.

[0150] After obtaining the training sample groups at different historical moments, the historical influent data of a training sample group is input into the extreme gradient boosting tree classifier to predict the flocculant dosage, and the predicted flocculant amount is obtained. Based on the predicted flocculant amount and the historical flocculant dosage in the training sample group, the model loss value is determined.

[0151] That is, based on the predicted amount of flocculant and the historical dosage of flocculant in the training sample group, the model loss value is determined, including: determining the difference between the predicted amount of flocculant and the corresponding historical dosage of flocculant as the prediction deviation; when the prediction deviation is less than or equal to the hyperparameters set by the extreme gradient boosting tree classifier, the model loss value is calculated using the mean square error function; when the prediction deviation is greater than the hyperparameters set by the extreme gradient boosting tree classifier, the model loss value is calculated using the mean absolute error.

[0152] That is, the model loss value is calculated as follows:

[0153]

[0154] Among them, x is the predicted amount of flocculant; y is the corresponding historical dosage of flocculant, that is, the historical dosage of flocculant in the training sample group; h(x,y) is the prediction deviation, that is, the difference between x and y; δ is the hyperparameter of the extreme gradient boosting tree classifier, and the value of δ is set according to needs; L δ (h(x,y)) is the model loss value.

[0155] In other embodiments, the model loss value may be calculated in other ways, for example, directly calculating the edit distance between the predicted amount of flocculant and the historical amount of flocculant added in the training sample group as the model loss value.

[0156] S303: When the model loss value does not meet the convergence requirements, the parameters of the extreme gradient boosting tree classifier are updated and iterated based on the remaining historical water inflow data until the model loss value meets the convergence requirements, and the converged parameters of the extreme gradient boosting tree classifier are output to obtain a dosage prediction model.

[0157] After obtaining the model loss value, when the model loss value does not meet the convergence requirements, the parameters of the extreme gradient boosting tree classifier are updated and iterated based on the remaining historical water inlet data, that is, S302 to S303 are repeatedly executed until the model loss value meets the convergence requirements, and the converged parameters of the extreme gradient boosting tree classifier are output to obtain the dosage prediction model.

[0158] When the model loss value is less than or equal to the preset loss value, or the number of training times reaches the preset number of times, it is determined that the model loss value meets the convergence requirement. Conversely, when the model loss value is greater than the preset loss value, and the number of training times does not reach the preset number of times, it is determined that the model loss value does not meet the convergence requirement.

[0159] In this embodiment, data is extracted from multiple historical water quality treatment samples at different historical moments to obtain training sample groups at different historical moments, and then the historical water inflow data in the training sample group is input into the extreme gradient boosting tree classifier to predict the flocculant dosage to obtain the predicted flocculant dosage, and the model loss value is determined based on the predicted flocculant dosage and the historical flocculant dosage in the training sample group; when the model loss value does not meet the convergence requirements, the parameters of the extreme gradient boosting tree classifier are updated and iterated based on the remaining historical water inflow data until the model loss value meets the convergence requirements, and the parameters of the converged extreme gradient boosting tree classifier are output to obtain the dosage prediction model. The training process of the dosage prediction model is clarified, and training is performed based on the historical water inflow data and the corresponding historical flocculant dosage, thereby improving the accuracy of the dosage prediction model.

[0160] In one embodiment, before step S30, that is, before using the turbidity prediction model to predict the effluent turbidity, it is necessary to train a neural network model based on multiple historical water quality treatment samples at different historical moments to obtain a turbidity prediction model so that the turbidity prediction model can be used to predict the effluent turbidity later. Before step S30, that is, before using the turbidity prediction model to predict the effluent turbidity, the turbidity prediction model is trained in the following manner:

[0161] SA11: Obtain multiple historical water quality treatment samples at different historical moments.

[0162] After preprocessing the historical data collected at each measuring point in the water quality treatment device, multiple historical water quality treatment samples at different historical moments are obtained, and the historical water quality treatment samples include historical influent data, historical flocculant dosage, and historical effluent turbidity. The historical influent data includes influent flow, influent pH, influent temperature, influent dissolved oxygen, and influent turbidity.

[0163] S12: Taking the water quality treatment time as the time window, a sliding time window algorithm is used to perform sliding sampling on the historical water inflow data and the historical flocculant dosage at different historical moments to obtain multiple sample input data groups.

[0164] Then, with the time window of the water quality treatment time required by the flocculation reaction process, the sliding time window algorithm is used to perform sliding sampling on the historical water inflow data and the historical flocculant dosage at different historical moments, and obtain multiple sample input data groups sorted in sequence. Among them, each sample input data group includes the historical water inflow data, the historical flocculant dosage and the historical effluent turbidity within a preset time period; the starting time of the multiple sample input data groups increases in sequence, and the duration corresponding to each preset time period is the water quality treatment duration.

[0165] For example, the plurality of sample input data groups include n sample input data groups: the first sample input data group, the second sample input data group, ... the nth sample input data group, then the first sample input data group includes historical water inflow data and historical flocculant dosage from historical moment h to historical moment h+t; the second sample input data group includes historical water inflow data and historical flocculant dosage from historical moment h+1 to historical moment (h+1)+t... the nth sample input data group includes historical water inflow data and historical flocculant dosage from historical moment h+n-1 to historical moment (h+n-1)+t. t is the water quality treatment time required for the flocculation reaction process.

[0166] S13: Inputting the sample input data group into a preset network to predict the turbidity of the effluent water, obtaining a predicted value of the turbidity of the effluent water, and determining the total loss value based on the predicted value of the turbidity of the effluent water and the historical turbidity of the effluent water at the end time in the sample input data group;

[0167] After obtaining multiple sample input data groups, one sample input data group is input into a preset network to predict the effluent turbidity to obtain an effluent turbidity prediction value, and the total loss value is determined based on the effluent turbidity prediction value and the historical effluent turbidity at the end time in the sample input data group.

[0168] Among them, considering that the methods of calculating the total loss through the mean absolute error (MAE) function and the mean-square error (MSE) function have both advantages and disadvantages, such as MAE is not sensitive to outliers, but is very unstable at extreme points, and the gradient near the extreme points is very large, while the gradient of MSE gradually decreases near the extreme points to obtain accurate extreme values, but it is sensitive to outliers. Therefore, a robust loss function (Huber Loss) with parameters is used to calculate the total loss value of the preset network. The Huber Loss function can concentrate the advantages of MAE and MSE, which can enhance the robustness of MSE to outliers, and use MAE to reduce the impact of outliers, avoid the model's excessive sensitivity to extreme data, reduce the negative impact of outliers on model training, and improve the prediction accuracy of the model.

[0169] That is, based on the predicted effluent turbidity and the historical effluent turbidity at the end time in the sample input data group, the total loss value is determined, including: determining the difference between the predicted effluent turbidity and the historical effluent turbidity at the end time as the prediction deviation; when the prediction deviation is less than or equal to the hyperparameters set by the preset network, the total loss value is calculated using the mean square error function; when the prediction deviation is greater than the hyperparameters set by the preset network, the total loss value is calculated using the mean absolute error.

[0170] That is, the total loss value is calculated as follows:

[0171]

[0172] Among them, x is the predicted value of the effluent turbidity; y is the actual effluent turbidity, that is, the historical effluent turbidity at the end time of the sample input data group; h(x,y) is the prediction deviation, that is, the difference between x and y; δ is the hyperparameter of the preset network, and the value of δ is set according to needs; L δ (h(x,y)) is the total loss value.

[0173] In other embodiments, the total loss value may be calculated in other ways, for example, by directly calculating the edit distance between the predicted effluent turbidity and the historical effluent turbidity at the end time in the sample input data set as the total loss value.

[0174] S05: When the total loss value does not meet the convergence requirements, iteratively train the parameters of the preset network based on the remaining sample input data group until the total loss value meets the convergence requirements, output the parameters of the converged preset network, and obtain the turbidity prediction model.

[0175] After obtaining the total loss value, when the total loss value does not meet the convergence requirements, the parameters of the preset network are iteratively trained based on the remaining sample input data group, that is, the above steps S12 to S13 are repeated until the total loss value meets the convergence requirements, and the parameters of the preset network after convergence are output to obtain the turbidity prediction model.

[0176] When the total loss value is less than or equal to the preset loss value, or the number of training times reaches the preset number of times, it is determined that the total loss value meets the convergence requirement. On the contrary, when the total loss value is greater than the preset loss value and the number of training times does not reach the preset number of times, it is determined that the total loss value does not meet the convergence requirement.

[0177] In this embodiment, multiple historical water quality treatment samples at different historical moments are obtained, and the historical water quality treatment samples include historical water inlet data, historical flocculant dosage, and historical effluent turbidity; the water quality treatment time required for the flocculation reaction process is used as the time window, and the sliding time window algorithm is used to perform sliding sampling on the historical water inlet data and historical flocculant dosage at different historical moments to obtain multiple sample input data groups; the sample input data group is input into the preset network to predict the effluent turbidity, and the effluent turbidity prediction value is obtained, and the total loss value is determined based on the effluent turbidity prediction value and the historical effluent turbidity at the end time in the sample input data group; when the total loss value does not meet the convergence requirements, the parameters of the preset network are iteratively trained based on the remaining sample input data groups until the total loss value meets the convergence requirements, and the parameters of the preset network after convergence are output to obtain the turbidity prediction model. The training process of the turbidity prediction model is clarified, and multiple sample input data groups are used as input for training, which improves the accuracy of the turbidity prediction model.

[0178] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0179] In one embodiment, a control device is provided, which corresponds one-to-one to the control method of the water treatment device in the above embodiment. Figure 7 As shown, the parameter aggregation device includes a collection module 701, a prediction module 702 and a control module 703. The detailed description of each functional module is as follows:

[0180] The acquisition module 701 is used to collect data from the water quality treatment device based on the water quality treatment time required by the flocculation process, and obtain multiple groups of target water inflow data and multiple flocculant dosages from the first moment to the second moment, where the time from the first moment to the second moment is the water quality treatment time;

[0181] Prediction module 702, used to predict effluent turbidity based on multiple groups of target influent data and multiple flocculant dosages using a turbidity prediction model to obtain predicted effluent turbidity, wherein the turbidity prediction model is a neural network model trained based on multiple historical water quality treatment samples of the water quality treatment device at different historical moments and the water quality treatment duration;

[0182] The control module 703 is used to control the water treatment device to perform feedback adjustment on the dosage of the flocculant based on the predicted effluent turbidity and the actual effluent turbidity of the water treatment device at the second moment.

[0183] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the present application. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0184] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0185] The present application also provides an electronic device, such as Figure 8 As shown, the electronic device includes: at least one processor, a memory, and a computer program stored in the memory and executable on the at least one processor, wherein the processor implements the steps in any of the above-mentioned method embodiments when executing the computer program, or implements the functions of the modules / units in the above-mentioned device embodiments when executing the computer program.

[0186] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present application. The one or more modules / units may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program in the electronic device.

[0187] Those skilled in the art will understand that Figure 8 These are merely examples of the electronic device and do not constitute a limitation of the electronic device. The electronic device may include more or fewer components than those shown in the figure, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0188] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0189] The memory may be an internal storage unit of the electronic device, such as a hard disk or memory of the electronic device. The memory may also be an external storage device of the electronic device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device. Furthermore, the memory may also include both an internal storage unit of the electronic device and an external storage device.

[0190] An embodiment of the present application further provides a readable storage medium, wherein the readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0191] An embodiment of the present application provides a computer program product. When the computer program product runs on a mobile terminal, the mobile terminal can implement the steps in the above-mentioned method embodiments when executing the computer program product.

[0192] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the camera / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), electric carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electric carrier signals and telecommunication signals.

[0193] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0194] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0195] In the embodiments provided in the present application, it should be understood that the disclosed devices / equipment and methods can be implemented in other ways. For example, the device / equipment embodiments described above are merely schematic, for example, the division of the modules or units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0196] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0197] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A water treatment device control method, characterized in that: include: Based on the water quality treatment time required by the flocculation process, data is collected from the water quality treatment device to obtain multiple groups of target water inflow data and multiple flocculant dosages from the first moment to the second moment, and the time from the first moment to the second moment is the water quality treatment time; Based on multiple groups of the target influent data and multiple flocculant dosages, the effluent turbidity is predicted using a turbidity prediction model to obtain predicted effluent turbidity, wherein the turbidity prediction model is a neural network model obtained by training based on multiple historical water quality treatment samples of the water quality treatment device at different historical moments and the water quality treatment duration; Based on the predicted effluent turbidity and the actual effluent turbidity of the water treatment device at the second moment, the water treatment device is controlled to perform feedback adjustment on the dosage of the flocculant.

2. The water treatment device control method according to claim 1, characterized in that: The step of controlling the water treatment device to perform feedback adjustment on the dosage of the flocculant based on the predicted effluent turbidity and the actual effluent turbidity of the water treatment device at the second moment includes: When the actual effluent turbidity at the second moment meets the water quality treatment requirement, the predicted dosage is adjusted based on the deviation between the predicted effluent turbidity and the actual effluent turbidity, and the predicted dosage is obtained by predicting the flocculant dosage based on the target influent data at the second moment by using a dosage prediction model; Based on the adjusted predicted dosage, multiple groups of target water inlet data and multiple flocculant dosages, the turbidity prediction model is used to re-predict the effluent turbidity until the absolute value of the deviation is less than a preset value, and the adjusted predicted dosage is output as the target dosage; The target dosage is used as the actual dosage of the flocculant at the second moment, and the water treatment device is controlled to input the flocculant.

3. The water treatment device control method according to claim 1, characterized in that: The method of predicting the effluent turbidity based on the multiple groups of target influent data and the multiple flocculant dosages using a turbidity prediction model to obtain the predicted effluent turbidity includes: Based on the target influent data at the second moment, a dosage prediction model is used to predict the flocculant dosage to obtain a predicted dosage at the second moment, wherein the dosage prediction model is a prediction model obtained by deep learning based on a plurality of historical water quality treatment samples; Based on the multiple groups of the target influent data, the multiple flocculant dosages and the predicted dosage at the second moment, the turbidity prediction model is used to predict the effluent turbidity to obtain the predicted effluent turbidity.

4. The water treatment device control method according to claim 3, characterized in that: The method of predicting the effluent turbidity by using the turbidity prediction model based on the multiple groups of the target influent data, the multiple flocculant dosages and the predicted dosage at the second moment to obtain the predicted effluent turbidity includes: Standardizing each data in each target water inflow data to obtain multiple groups of standard water inflow data, and standardizing the predicted dosage and each flocculant dosage to obtain multiple standard flocculant dosages; Inputting a plurality of sets of the standard influent data and a plurality of the standard flocculant dosages into the turbidity prediction model to predict the effluent turbidity, thereby obtaining an initial predicted turbidity; The initial predicted turbidity is subjected to denormalization processing to obtain the predicted effluent turbidity.

5. The water treatment device control method according to claim 1, characterized in that: The plurality of historical water quality treatment samples are obtained in the following manner: Obtaining historical data collected from each measuring point in the water quality treatment device, and performing standardization processing to obtain influent water quality data, historical flocculant dosage, and historical effluent turbidity at different historical moments; Performing target inlet water parameter screening on the inlet water quality data at different historical moments to obtain historical inlet water data at different historical moments, wherein the target inlet water parameter is an inlet water parameter that affects the flocculant dosage and / or the outlet water turbidity; The historical water inlet data, the historical flocculant dosage and the historical effluent turbidity at the same historical moment are combined into one historical water quality treatment sample to obtain a plurality of historical water quality treatment samples.

6. The water treatment device control method according to claim 5, characterized in that: The target inlet water parameter screening is performed on the inlet water quality data at different historical moments to obtain the historical inlet water data at different historical moments, including: Performing data correlation analysis on the influent water quality data, the historical flocculant dosage and the historical effluent turbidity, and screening the influent parameters in the influent water quality data based on the correlation analysis data to obtain the target influent parameters whose correlation meets the requirements; The inlet water quality data at different historical moments are screened based on the target inlet water parameter to obtain the historical inlet water data at different historical moments.

7. The water treatment device control method according to any one of claims 1 to 6, characterized in that: The turbidity prediction model is trained in the following way: Acquire a plurality of the historical water quality treatment samples at different historical moments, wherein the historical water quality treatment samples include historical water inlet data, historical flocculant dosage, and historical effluent turbidity; Taking the water quality treatment duration as the time window, a sliding time window algorithm is used to perform sliding sampling on the historical water inflow data and the historical flocculant dosage at different historical moments, to obtain a plurality of sample input data groups, each of which includes the historical water inflow data, the historical flocculant dosage and the historical effluent turbidity within a preset time period, and the starting times of the plurality of sample input data groups increase in sequence, and the duration corresponding to each preset time period is the water quality treatment duration; Inputting the sample input data set into a preset network to predict the turbidity of the effluent water to obtain a predicted value of the turbidity of the effluent water, and determining a total loss value based on the predicted value of the turbidity of the effluent water and the historical turbidity of the effluent water at the end time in the sample input data set; When the total loss value does not meet the convergence requirement, the parameters of the preset network are iteratively trained based on the remaining sample input data group until the total loss value meets the convergence requirement, and the converged parameters of the preset network are output to obtain the turbidity prediction model.

8. A water treatment system, characterized in that: It includes a water treatment device and a control device, wherein the control device is used for: Based on the water quality treatment time required by the flocculation process, data is collected from the water quality treatment device to obtain multiple groups of target water inflow data and multiple flocculant dosages from the first moment to the second moment, and the time from the first moment to the second moment is the water quality treatment time; Based on multiple groups of the target influent data and multiple flocculant dosages, the effluent turbidity is predicted using a turbidity prediction model to obtain predicted effluent turbidity, wherein the turbidity prediction model is a neural network model obtained by training based on multiple historical water quality treatment samples of the water quality treatment device at different historical moments and the water quality treatment duration; Based on the predicted effluent turbidity and the actual effluent turbidity of the water treatment device at the second moment, the water treatment device is controlled to perform feedback adjustment on the dosage of the flocculant.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the water treatment device control method according to any one of claims 1 to 7 are implemented.

10. A readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the water treatment device control method according to any one of claims 1 to 7 are implemented.

Images