Intelligent sludge discharge scheduling method, system, medium and equipment for sedimentation tank of water supply plant

Abstract

The invention relates to the field of sedimentation tank sludge discharge, and discloses a water supply plant sedimentation tank intelligent sludge discharge scheduling method, system, medium and equipment, and the method comprises the steps: obtaining label data used for training a sludge water initial solid content estimation model, and online instrument data used for training the sludge water initial solid content estimation model; according to the online instrument data, constructing feature data used for training a sludge water initial solid content estimation model so as to train the sludge water initial solid content estimation model to obtain a sludge water solid content change estimation model; calibrating the sludge water solid content change estimation model through the obtained sludge water solid content sample data in the sludge discharge process, and checking the sludge water solid content change estimation model based on the sludge amount; and setting a target sludge water average solid content, and comparing the estimated sludge water solid content obtained by the checked sludge water solid content change estimation model with the target sludge water average solid content so as to judge the opening and closing time of the sludge valve.

Description

Technical Field

[0001] The invention relates to the technical field of sedimentation tank sludge discharge, and in particular to an intelligent sludge discharge scheduling method, system, medium and equipment for a sedimentation tank in a water supply plant. Background Art

[0002] Sludge thickening and dehydration are important components of the energy consumption of water plant operation. Excessive sludge discharge will cause the water in the water plant sedimentation tank to be discharged, which may increase the energy consumption of the sludge thickening and dehydration process and the water plant's own water rate. Insufficient sludge discharge will cause excessive accumulation of sludge in the sedimentation tank, affecting the sedimentation process and increasing the risk of substandard effluent from the sedimentation tank. To achieve a moderate frequency and amount of sludge discharge, it is necessary to transform the sludge discharge mechanism from relying solely on manual experience to a quantitative scheduling framework.

[0003] Sensing the accumulation state of sludge in the sedimentation tank is the key to quantitatively scheduling the sludge discharge of the sedimentation tank. The main methods for sensing the state of sludge in the sedimentation tank in the past can be divided into two categories: automatic control methods based on online instruments and methods based on numerical simulation. The automatic control method based on online instruments requires a sludge level meter to monitor the sedimentation status in the sedimentation tank, or a sludge concentration meter to monitor the changes in the solid content of the discharged sludge. However, these two existing types of instruments still have problems such as inaccurate measurement and large data fluctuations, which greatly reduces the reliability of the automatic control system. The method based on numerical simulation mainly quantifies the particle settling velocity, and it is difficult to directly model the sludge solid content in each area of ​​the sedimentation tank, and the calibration and calculation process of these numerical models are often cumbersome and have limited accuracy. In contrast, data-driven models have strong predictive capabilities, but require a large amount of data for training, and their interpretability is limited.

[0004] Digital twin, an emerging theoretical and technical system, integrates data collection and model building, advocates the combination of mechanism knowledge and data-driven models, and can perform tasks such as monitoring, simulation, prediction and optimization. This provides a compromise and comprehensive solution for sedimentation tank sludge scheduling between sensor-based automatic control methods and numerical models. How to use conventional flow and water quality instruments to further organically combine mechanism knowledge and data-driven models to give full play to their respective strengths and achieve accurate sludge discharge in water plants needs to be solved. Summary of the invention

[0005] In view of the above problems, the object of the present invention is to provide a method, system, medium and equipment for intelligent sludge discharge scheduling of a sedimentation tank in a water supply plant, which can improve the water quality stability of the effluent from the sedimentation tank.

[0006] To achieve the above-mentioned purpose, in the first aspect, the technical solution adopted by the present invention is: an intelligent sludge discharge scheduling method for a sedimentation tank in a water supply plant, which comprises: obtaining label data for training a sludge water initial solid content estimation model and online instrument data for training the sludge water initial solid content estimation model, constructing feature data for training the sludge water initial solid content estimation model according to the online instrument data, so as to train the sludge water initial solid content estimation model and obtain a sludge water solid content change estimation model; calibrating the sludge water solid content change estimation model by obtaining sludge water solid content sample data during the sludge discharge process, and verifying the sludge water solid content change estimation model based on the mud amount; setting a target sludge water average solid content, and comparing the estimated sludge water solid content obtained by the verified sludge water solid content change estimation model with the target sludge water average solid content to determine the opening and closing time of the sludge discharge valve.

[0007] Furthermore, characteristic data for training the initial solid content estimation model of the sludge water is constructed based on the online instrument data, so as to train the initial solid content estimation model of the sludge water, including:

[0008] Set the initial hyperparameters and loss function of the model, and divide the test set;

[0009] Use K-Fold cross-validation strategy to divide the training set and validation set for training;

[0010] The hyperparameters with the lowest total loss function are selected as the optimal hyperparameters for training, and the results output by each model are averaged on the test set to obtain the final estimated results.

[0011] Furthermore, the construction of a prediction model for the change of the solid content of the sludge water includes: obtaining sample data of the solid content of the sludge water during the sludge discharge process, and constructing a prediction model for the change of the solid content of the sludge water according to the zoning law of crowded sedimentation;

[0012] The prediction model of the solid content change of sludge water is:

[0013]

[0014] Where t is the time after the mud discharge valve is opened, in seconds; S(t) is the solid content of the mud discharge water at time t, S(0) is the initial solid content of the mud discharge water, and a, b, m, and n are model parameters to be calibrated that are greater than 0.

[0015] Furthermore, the prediction model of the change of solid content in sludge water is calibrated, including:

[0016] A loss function is set, and sample data collected multiple times are substituted into the loss function respectively, with minimization of the loss function as the optimization goal, to solve the values ​​of multiple sets of fixed-value parameters and variable-value parameters of the model to be calibrated;

[0017] The fixed parameters of multiple models are averaged to obtain the final fixed parameters. A scatter plot of the variable parameters obtained from each model with respect to the initial solid content of the mud water is made to determine the functional relationship between the variable parameters and the initial solid content of the mud water. The function parameters are determined with the goal of minimizing the loss function.

[0018] Furthermore, the prediction model of the change of solid content in the discharged mud water is calibrated based on the mud volume, including:

[0019] Compare and verify the estimated dry mud generation with the dry mud volume shipped out of the water plant to determine the estimated dry mud generation;

[0020] The sludge discharge volume is estimated by the sludge water solid content change prediction model, and the sludge discharge volume is compared with the estimated dry mud production. If the error between the two is within the allowable range, the sludge water solid content change prediction model is verified to be successful.

[0021] Further, judging the opening and closing time of the mud discharge valve includes:

[0022] Set the target average solid content of sludge water;

[0023] Grid sampling candidate values ​​for the threshold of initial solid content of mud water that determines the opening of mud valve;

[0024] For each candidate value of the initial solid content threshold of the mud discharge water, determine the corresponding mud discharge valve closing time;

[0025] For each candidate solution, calculate the average solid content of the sludge water;

[0026] The initial solid content threshold of the mud discharge water corresponding to the result closest to the target average solid content of the mud discharge water is taken as the threshold for determining the opening of the mud discharge valve.

[0027] Furthermore, when the estimated initial solid content of the mud discharge water reaches a threshold value, the mud discharge valve is opened; when the mud discharge valve opening time reaches the mud discharge valve closing time, the mud discharge valve is closed.

[0028] In the second aspect, the technical solution adopted by the present invention is: an intelligent sludge discharge scheduling system for a sedimentation tank of a water supply plant, comprising: a model construction module, which obtains label data for training a sludge water initial solid content estimation model and online instrument data for training the sludge water initial solid content estimation model, and constructs feature data for training the sludge water initial solid content estimation model based on the online instrument data to train the sludge water initial solid content estimation model to obtain a sludge water solid content change estimation model; a model verification module, which calibrates the sludge water solid content change estimation model by obtaining sludge water solid content sample data during the sludge discharge process, and verifies the sludge water solid content change estimation model based on the mud amount; a sludge valve opening and closing time determination module, which sets a target sludge water average solid content, and compares the estimated sludge water solid content obtained by the verified sludge water solid content change estimation model with the target sludge water average solid content to determine the sludge valve opening and closing time.

[0029] In a third aspect, the technical solution adopted by the present invention is: a computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, and when the instructions are executed by a computing device, the computing device executes any one of the above methods.

[0030] In a fourth aspect, the technical solution adopted by the present invention is: a computing device, comprising: one or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any of the above methods.

[0031] The present invention adopts the above technical solution, which has the following advantages:

[0032] The present invention combines professional knowledge with data-driven modeling, and establishes a system perception mechanism that combines sensor monitoring with model-based prediction. By using conventional flow and water quality online monitoring instruments, the concentration change of sludge water can be estimated. It is particularly suitable for the current situation where sludge level meters and sludge concentration meters measure inaccurately and unstablely. By modeling the solid content of sludge water to guide the opening and closing of the sludge valve, it helps to improve the solid content of sludge water, save self-use water consumption and sludge concentration and dehydration energy consumption costs, improve the water quality stability of sedimentation tank effluent, and reduce economic costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 This is a flow chart of intelligent sludge discharge scheduling in the sedimentation tank of a water supply plant described in the embodiment of this application;

[0034] Figure 2 This is a flow chart of the method for estimating the initial solid content of the mud water described in the embodiment of the present application;

[0035] Figure 3 This is a flow chart of the training model for estimating the initial solid content of muddy water described in the embodiment of the present application;

[0036] Figure 4 This is a flow chart of the method for estimating the change of solid content in sludge water described in the embodiment of the present application;

[0037] Figure 5 Schematic diagram of the derivation process of the solid content change prediction model of the sludge water described in the embodiment of the present application;

[0038] Figure 6 This is a flow chart of the model for estimating the change in solid content of sludge water according to the embodiment of the present application;

[0039] Figure 7 This is a flow chart of the model calibration method based on mud volume described in the embodiments of the present application;

[0040] Figure 8 This is a flow chart of the method for judging the opening and closing time of the mud discharge valve described in the embodiment of the present application;

[0041] Fig. 9 This is a schematic diagram of the principle of the mud discharge valve closing time judgment criterion described in the embodiment of the present application. DETAILED DESCRIPTION

[0042] In order to organically combine the mechanism knowledge with the data-driven model to give full play to their respective strengths and realize the precise sludge discharge of the water plant, the present invention provides a method, system, medium and equipment for intelligent sludge discharge scheduling of the sedimentation tank of a water supply plant, including: constructing and training a model for estimating the initial solid content of sludge water; constructing a model for estimating the change of solid content of sludge water; calibrating the model for estimating the change of solid content of sludge water; checking the two constructed models based on the amount of mud; judging the opening and closing time of the sludge valve. The present invention helps to improve the solid content of sludge water, save self-use water consumption and the energy consumption cost of sludge concentration and dehydration, improve the water quality stability of the effluent from the sedimentation tank, and reduce economic costs.

[0043] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0044] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0045] In one embodiment of the present invention, a method for intelligent sludge discharge scheduling of a sedimentation tank in a water supply plant is provided to improve the solid content of sludge water discharged from the sedimentation tank. Figure 1 As shown, the method comprises the following steps:

[0046] 1) Obtain label data for training the initial solid content estimation model of mud water and online instrument data for training the initial solid content estimation model of mud water, and construct feature data for training the initial solid content estimation model of mud water based on the online instrument data to train the initial solid content estimation model of mud water and obtain the mud water solid content change estimation model.

[0047] 2) The sludge water solid content change prediction model is calibrated by obtaining sample data of the sludge water solid content during the sludge discharge process, and the sludge water solid content change prediction model is verified based on the mud volume.

[0048] 3) Set the target average solid content of the mud water, and compare the estimated solid content of the mud water obtained by the mud water solid content change prediction model with the target average solid content of the mud water to determine the opening and closing time of the mud valve.

[0049] In the above step 1), if Figure 2 As shown in the figure, the training of the initial solid content estimation model of the mud water includes the following steps:

[0050] 1.1.1) Obtain label data for training the initial solid content estimation model of the mud water. The label data is the solid content of the mud water when the mud valve is just opened every time the mud is discharged from the sedimentation tank, that is, the initial solid content of the mud water.

[0051] The specific method for collecting label data is to quickly collect the initial sludge water after the sludge discharge valve is opened every time the sedimentation tank is discharged, and determine its solid content.

[0052] In order to obtain the initial solid content of sludge water at different sludge accumulation stages, it is preferred in this embodiment to moderately change the sludge discharge frequency, such as two discharges a day, one discharge a day, one discharge every two days, etc., which is not limited in this embodiment.

[0053] Preferably, the amount of data collected is not less than 100, which is not limited in this embodiment.

[0054] Preferably, it is advisable to collect the initial solid content data of the muddy water for one year to cover all seasons and take into account the seasonal influence. This embodiment does not limit this.

[0055] 1.1.2) Obtain online instrument data for training the initial solid content estimation model of sludge water. The online instrument data includes the turbidity of the incoming water, the coagulant addition flow rate, the inlet flow rate of the sedimentation tank, the opening and closing time of the sedimentation tank sludge valve, and the turbidity of the effluent from the sedimentation tank collected when the coagulation sedimentation tank of the water supply plant is in operation.

[0056] Preferably, this embodiment additionally collects and uses data on inlet water temperature, pH, conductivity, and residual chlorine.

[0057] It can be understood that the additional preferred online instrument data in this embodiment is data used for machine learning model training, which can be arbitrarily selected according to actual instrument conditions and requirements on site, and this embodiment does not limit this.

[0058] 1.1.3) Construct feature data for model training, including the total amount of treated water between each two sludge discharges (m 3 ), total amount of reagent added (L), average reagent addition rate (L / m 3 ), average inlet turbidity (NTU), total suspended solids (SS) removal (t), and the running time since the last sludge discharge. Match these data with the label data one by one.

[0059] The calculation formula for the average agent dosage rate is:

[0060]

[0061] in, is the average reagent dosage rate, T is the total time between two sludge discharges, is the total amount of reagent added between two sludge discharges (L), is the total amount of water treated between two sludge discharges (m 3 ).

[0062] The calculation formula for average incoming water turbidity is:

[0063]

[0064] in, is the average water turbidity, turbidity in,t is the turbidity of the incoming water, Q t is the inflow flow rate of the sedimentation tank (m 3 ).

[0065] The total amount of SS removal can be estimated using a turbidity monitoring instrument:

[0066]

[0067] Where, ΔSS is the total amount of SS removed (t), turbidity ef,t is the effluent turbidity of the sedimentation tank (NTU), and k is the conversion coefficient between SS and turbidity, which is obtained through on-site measurement.

[0068] Preferably, this embodiment constructs additional characteristic data by means of additionally collected online instrument data, including the average incoming water temperature, average incoming water pH, average incoming water conductivity, and average incoming water residual chlorine between each two sludge discharges. The calculation formula for these additionally constructed characteristic data is:

[0069]

[0070] in, is the additional feature to be constructed, x t These are the corresponding instrument readings (inlet water temperature, pH, conductivity, and residual chlorine data).

[0071] It can be understood that the additional feature data constructed in this embodiment is the input data of the machine learning model, which is obtained by additionally selecting online instrument data, and this embodiment does not limit this.

[0072] 1.1.4) Train the model for estimating the initial solid content of sludge water. The input sample data includes the total amount of treated water between each two sludge discharges (m 3 ), total amount of reagent added (L), average reagent addition rate (L / m 3 ), average inlet water turbidity (NTU), total SS removal (t), operating time since the last sludge discharge, average inlet water temperature, average inlet water pH, average inlet water conductivity and average inlet water residual chlorine.

[0073] In this embodiment, the initial solid content estimation model for mud water may be a model for estimating the initial solid content of mud water after the mud discharge valve is opened based on input data, for example, it may be a gradient boosting tree model, which is not limited in this embodiment.

[0074] In the above step 1), if Figure 3 As shown, characteristic data for training the initial solid content estimation model of the sludge water is constructed based on the online instrument data to train the initial solid content estimation model of the sludge water, including the following steps:

[0075] 1.2.1) Set the initial hyperparameters and loss function of the model and divide the test set;

[0076] In this embodiment, for example, it may be the default hyperparameters and mean square error loss function of the model, which is not limited in this embodiment.

[0077] All data are sorted in chronological order, and the last 1 / 5 of the data is divided into a test set. It is understandable that the proportion of the test set division can be determined according to the amount and quality of the data, and this embodiment does not limit this.

[0078] 1.2.2) Use K-Fold cross-validation strategy to divide the training set and validation set for training;

[0079] Specifically, the first 4 / 5 of the data are sorted by time and divided into 4 parts on average. The K-Fold cross-validation strategy is used to perform 4 trainings. Each time, one of the data is used as the validation set, and the other data is used as the training set. Four trained models are obtained, and the sum of the loss functions of these four models on the validation set is calculated.

[0080] 1.2.3) Set different hyperparameters for training, select the hyperparameters with the lowest total loss function as the optimal hyperparameters for training, average the results output by each model on the test set, and obtain the final estimated results. The four models at this time are the finalized models for estimating the change in solid content of sludge water.

[0081] The test set data were input into the four models respectively to obtain the initial solid content of the muddy water estimated by the four models. The average of the results output by the four models was used to obtain the final estimated result.

[0082] It is understandable that the K-Fold cross-validation strategy is mainly used to make full use of data training when the amount of data is small and reduce the risk of overfitting. The amount and proportion of the validation set division can be determined according to the amount and quality of the data, and this embodiment does not limit this.

[0083] In the above step 1), sample data of the solid content of the sludge water during the sludge discharge process is obtained. Each time the sedimentation tank is discharged, the sludge water is collected at 0 seconds, 30 seconds from the start of the sludge water outflow, and at the end of the sludge discharge, and the solid content thereof is measured respectively. Sample data is collected at least three times when the sedimentation tank is discharged.

[0084] It is understandable that, in one sample data collection, the frequency of collecting the solid content data of the sludge water is used for model parameter calibration, and it is advisable to collect the data at as high a frequency as possible under the conditions permitted by manpower and equipment, and this embodiment does not limit this.

[0085] It is understandable that sample data should be collected as many times as possible within the limits of manpower and equipment.

[0086] In this embodiment, the construction of the prediction model for the change of solid content in mud water is specifically as follows: sample data of the solid content in mud water during the mud discharge process is obtained, and the prediction model for the change of solid content in mud water is constructed according to the zoning law of crowded sedimentation.

[0087] The prediction model of the solid content change of sludge water is:

[0088]

[0089] Where t is the time after the mud discharge valve is opened, in seconds; S(t) is the solid content of the mud discharge water at time t, S(0) is the initial solid content of the mud discharge water, and a, b, m, and n are model parameters to be calibrated that are greater than 0.

[0090] The model is derived from the partition law of crowded sedimentation. Figure 5 As shown, the derivation process of the sludge water solid content change prediction model provided by the embodiment of the present disclosure is derived based on the crowded sedimentation zoning principle.

[0091] like Figure 5 As shown in the figure, during the crowded sedimentation process, three areas will be formed: isoconcentration area, variable concentration area and compaction area. The sludge solid content in the compaction area is higher, and the solid content in the isoconcentration area is lower. The variable concentration area is between the compaction area and the isoconcentration area and is a transition area. During the sludge discharge process, parts of the compaction area, variable concentration area and isoconcentration area will be discharged in sequence. Therefore, the solid content of the sludge water will go through three stages: continuous high solid content - rapid reduction of solid content - continuous low solid content.

[0092] Based on the principles of accurately reflecting physical laws and minimizing model complexity, the Sigmoid function is flipped, translated and scaled to obtain a model for estimating changes in the solid content of mud water.

[0093] According to the above principles, in the prediction model of the solid content change of the sludge water, b is related to the moment when the solid content of the sludge water decreases significantly, and a is related to the rate of decrease of the solid content of the sludge water. The more sludge accumulates, the higher the solid content of the compaction zone. On the one hand, this will make S(0) higher; on the other hand, it will also make the range of the compaction zone larger, that is, the larger b is; it will also make the range of the variable concentration zone smaller, that is, the solid content of the sludge water decreases faster, that is, the larger a is. Therefore, a and b are both variable parameters related to S(0). Since S(0) is the initial solid content of the sludge water, it will change. Among them, step 1) is to estimate the initial solid content of each sludge water discharge. According to the physical meaning of a and b, a and b will vary with S(0), so they are not constant parameters.

[0094] m and n are constant parameters. Since S(t)→n when t→+∞, n represents the solid content in the isoconcentration zone, that is, the solid content after the sludge is discharged to the non-sludge accumulation zone. Among them, m has no actual physical meaning. When applied to the same sedimentation tank, m will not change significantly. Therefore, for this model, m is defined as a constant parameter, and m does not change with the change of S(0); according to the physical meaning of n, n has nothing to do with S(0). Therefore, n is a constant parameter, and n does not change with the change of S(0).

[0095] In the above step 2), if Figure 6 As shown in the figure, the prediction model of the solid content change of the sludge water is calibrated, including the following steps:

[0096] 2.1.1) If the number of times p of collecting sample data on the solid content of the mud water is 3, calculate the average solid content of the mud water at the end of each mud discharge, and use this as the value of n; if p>3, jump directly to step 2.1.2).

[0097] 2.1.2) Set the loss function, substitute the sample data collected p times into the loss function, take minimizing the loss function as the optimization goal, and solve the values ​​of the fixed and variable parameters of the p groups of models to be calibrated;

[0098] Preferably, this embodiment selects the root mean square error between the sample data and the model output as the loss function, but this embodiment does not limit the type of loss function. The root mean square error is:

[0099]

[0100] In the formula, RMSE is the root mean square error, q is the number of times sample data is collected each time, S i is the measured solid content of the sample data, It is the solid content estimated by the solid content change prediction model of sludge water.

[0101] 2.1.3) Take the average of the multiple model parameters obtained to obtain the final constant parameter m.

[0102] Specifically, the parameters m obtained from the p models are averaged to obtain the final fixed value parameter m. If p>3, the parameters n obtained from the p models are also averaged to obtain the final fixed value parameter n.

[0103] The variable parameters a and b obtained from the p models are plotted as scatter plots, with S(0) as the horizontal axis and a and b as the vertical axes. The scatter distribution is observed to determine the functional relationship between a and S(0), b and S(0), and the function parameters are determined with the goal of minimizing the loss function.

[0104] In the above step 2), if Figure 7 As shown in the figure, the prediction model of the change of solid content in the sludge water is calibrated based on the sludge volume, including the following steps:

[0105] 2.2.1) Compare and verify the estimated amount of dry mud generated with the amount of dry mud shipped out of the water plant to determine the estimated amount of dry mud generated;

[0106] In this embodiment, the estimation formula of dry mud generation is as follows:

[0107] m 产生=(k1C0+k2D)×k0Q0×10 -6

[0108] In the formula, m 产生 is the amount of dry mud produced (t); C0 is the raw water turbidity (NTU); D is the coagulant dosage rate (mg / L); Q0 is the amount of treated water (m 3 ); k0 is the water use rate of the water plant; k1 is the conversion coefficient of turbidity to SS, which is obtained through on-site measurement; k2 is the coefficient of the chemical conversion to dry mud, which depends on the type of chemical.

[0109] The calculation formula for the amount of dry mud transported out of the water plant is:

[0110] m′ 产生 =(1-w)×ρ×j×V×10 -3

[0111] In the formula, m′ 产生 is the amount of dry mud transported (t); w is the moisture content of the transported mud cake, obtained through actual measurement; ρ is the density of the transported mud cake (kg / m 3 ), obtained through actual measurement; j is the number of vehicles used to transport the mud cake; V is the volume of mud cake loaded on each vehicle used for transportation (m 3 ).

[0112] It is understandable that, since the amount of mud cake loaded on each truck is prone to significant fluctuations during actual production operations, the density and moisture content of the mud cake are also likely to be uneven, so m′ 产生 Only used to verify the dry mud production formula, and m 产生 The allowable error range is determined according to the on-site operating conditions and requirements, and this embodiment does not limit this.

[0113] 2.2.2) Estimate the amount of sludge discharged by using the sludge water solid content change prediction model, and compare the sludge discharge amount with the estimated dry mud production. If the error between the two is within the allowable range, the sludge water solid content change prediction model is verified to be successful.

[0114] Specifically, if the error between the amount of mud discharged and the estimated amount of dry mud produced is within the allowable range, then

[0115]

[0116] In the formula, m 排 is the amount of dry mud discharged (t), D is the number of mud discharges within the time range covered by the dry mud generation estimation, d represents the number of mud discharges, T 排，d is the duration of opening the mud discharge valve during the dth mud discharge (s), S d (t) is the prediction model of the change of solid content in mud water corresponding to the dth mud discharge, ρ 水 is the density of water (1×103 kg / m 3 ), V 排，d (t) is the amount of mud water discharged at time t during the dth mud discharge (m 3 ), calculated by the water level change of the downstream sludge regulating pool:

[0117]

[0118] In the formula, S 底 is the bottom area of ​​the sludge regulating pond (m 2 ), Δh d is the change in liquid level in the mud regulating tank before and after mud discharge (m), Q 出，d (t) is the effluent flow rate from the sludge regulating tank to the downstream sludge thickening tank (m 3 / s).

[0119] It is understandable that during the actual production operation, the mud in the pool may remain or be washed out. 产生 and m 排 The allowable error range is determined according to the on-site operation conditions and requirements, and this embodiment does not limit this.

[0120] In the above step 3), if Figure 8 As shown, judging the opening and closing time of the mud discharge valve includes the following steps:

[0121] 3.1) Setting a target average solid content of sludge water; the target value is determined by the water plant process, operation status and demand, and is not limited in this embodiment.

[0122] 3.2) Grid sampling candidate values ​​for the threshold value of the initial solid content S(0) of the mud discharge water that determines the opening of the mud discharge valve;

[0123] In this embodiment, 30 possible initial solid contents of the sludge water are selected from 1.1 to 4.0 times the target solid content with an interval of 0.1.

[0124] 3.3) For each candidate value of the initial solid content S(0) threshold of the mud discharge water, determine the corresponding mud discharge valve closing time;

[0125] In this embodiment, for each S(0), the fourth-order derivative of S(t) is calculated, and the largest solution among the t corresponding to 0 is taken. At this time, t is the closing time of the mud discharge valve.

[0126]

[0127] In the formula, t 关 That is, the closing time of the mud discharge valve.

[0128] like Fig. 9As shown, the principle of the criterion for judging the closing time of the mud discharge valve is schematically shown. When the first-order derivative of S(t) reaches the minimum value, the solid content drops sharply, and the solid content of the mud water decreases at the fastest rate. After that, the rate of decrease of the solid content of the mud water slows down, and this slowing trend continues to intensify until the second-order derivative of S(t) reaches the maximum value, which is the core point of the variable concentration zone. After that, the trend of slowing down the rate of decrease of the solid content of the mud water gradually eases. When this easing trend changes from accelerating to slowing down, that is, the second-order derivative of S(t) reaches the inflection point, corresponding to the maximum t that makes the fourth-order derivative of S(t) 0, this is the end point of the variable concentration zone. After this point, the change in the rate of decrease of the solid content of the mud water slows down significantly. It can be considered that the mud discharge has discharged the compaction area and the variable concentration area, leaving only the upper isoconcentration area.

[0129] 3.4) For each candidate solution, calculate the average solid content of the sludge water;

[0130] In this embodiment, for each S(0), its corresponding S(t) and the closing time of the mud discharge valve, the average solid content of the mud discharge water is calculated.

[0131]

[0132] In the formula, is the average solid content of the sludge water, T 排 It is the opening time of the mud discharge valve.

[0133] 3.5) The initial solid content threshold of the mud water corresponding to the result closest to the target average solid content S(0) of the mud water is taken as the threshold for determining the opening of the mud valve.

[0134] During the operation of the water plant, the initial solid content estimation model of the sludge water is run online. When the estimated initial solid content of the sludge water reaches the threshold, the sludge valve is opened; when the sludge valve opening time reaches the sludge valve closing time, the sludge valve is closed.

[0135] In one embodiment of the present invention, a system for intelligent sludge discharge scheduling of a sedimentation tank in a water supply plant is provided, comprising:

[0136] A model building module, which obtains label data for training a model for estimating the initial solid content of mud water and online instrument data for training the model for estimating the initial solid content of mud water, and constructs feature data for training the model for estimating the initial solid content of mud water according to the online instrument data, so as to train the model for estimating the initial solid content of mud water and obtain a model for estimating the change of the solid content of mud water;

[0137] The model verification module calibrates the prediction model of the change of the solid content of the sludge water by obtaining the sample data of the solid content of the sludge water during the sludge discharge process, and verifies the prediction model of the change of the solid content of the sludge water based on the amount of sludge;

[0138] The mud discharge valve opening and closing time determination module sets the target average solid content of the mud discharge water, and compares the estimated mud discharge water solid content obtained by the verified mud discharge water solid content change prediction model with the target average solid content of the mud discharge water to determine the mud discharge valve opening and closing time.

[0139] In the above embodiment, characteristic data for training the initial solid content estimation model of the muddy water is constructed based on the online instrument data to train the initial solid content estimation model of the muddy water, including:

[0140] Set the initial hyperparameters and loss function of the model, and divide the test set;

[0141] Use K-Fold cross-validation strategy to divide the training set and validation set for training;

[0142] The hyperparameters with the lowest total loss function are selected as the optimal hyperparameters for training, and the results output by each model are averaged on the test set to obtain the final estimated results.

[0143] In the above embodiment, the construction of the prediction model for the change of solid content of the sludge water includes: obtaining sample data of the solid content of the sludge water during the sludge discharge process, and constructing the prediction model for the change of solid content of the sludge water according to the zoning law of crowded sedimentation;

[0144] The prediction model of the solid content change of sludge water is:

[0145]

[0146] Where t is the time after the mud discharge valve is opened, in seconds; S(t) is the solid content of the mud discharge water at time t, S(0) is the initial solid content of the mud discharge water, and a, b, m, and n are model parameters to be calibrated that are greater than 0.

[0147] In the above embodiment, the solid content change prediction model of the sludge water is calibrated, including:

[0148] A loss function is set, and the sample data collected multiple times are substituted into the loss function respectively, with minimizing the loss function as the optimization goal, and the values ​​of multiple sets of fixed parameters and variable parameters of the models to be calibrated are solved; the fixed parameters of multiple models are averaged to obtain the final fixed parameters; a scatter plot of the variable parameters obtained from each model with respect to the initial solid content of the mud water is made to determine the functional relationship between the variable parameters and the initial solid content of the mud water, and the function parameters are determined with the goal of minimizing the loss function.

[0149] In the above embodiment, the prediction model of the solid content change of the discharged mud water is verified based on the mud volume, including:

[0150] Compare and verify the estimated dry mud generation with the dry mud volume shipped out of the water plant to determine the estimated dry mud generation;

[0151] The sludge discharge volume is estimated by the sludge water solid content change prediction model, and the sludge discharge volume is compared with the estimated dry mud production. If the error between the two is within the allowable range, the sludge water solid content change prediction model is verified to be successful.

[0152] In the above embodiment, determining the opening and closing time of the mud discharge valve includes:

[0153] Set the target average solid content of sludge water;

[0154] Grid sampling candidate values ​​for the threshold of initial solid content of mud water that determines the opening of mud valve;

[0155] For each candidate value of the initial solid content threshold of the mud discharge water, determine the corresponding mud discharge valve closing time;

[0156] For each candidate solution, calculate the average solid content of the sludge water;

[0157] The initial solid content threshold of the mud discharge water corresponding to the result closest to the target average solid content of the mud discharge water is taken as the threshold for determining the opening of the mud discharge valve.

[0158] Among them, when the estimated initial solid content of the mud discharge water reaches the threshold, the mud discharge valve is opened; when the mud discharge valve opening time reaches the mud discharge valve closing time, the mud discharge valve is closed.

[0159] The system provided in this embodiment is used to execute the above-mentioned method embodiments. Please refer to the above-mentioned embodiments for specific processes and detailed contents, which will not be repeated here.

[0160] In one embodiment of the present invention, a computing device is provided, which may be a terminal, and may include: a processor, a communication interface, a memory, a display screen, and an input device. Among them, the processor, the communication interface, and the memory communicate with each other through a communication bus. The processor is used to provide computing and control capabilities. The memory includes a non-volatile storage medium and an internal memory, and the non-volatile storage medium stores an operating system and a computer program. When the computer program is executed by the processor, the method in each of the above embodiments is implemented; the internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a management network, NFC (near field communication) or other technologies. The display screen may be a liquid crystal display screen or an electronic ink display screen, and the input device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computing device, or an external keyboard, touchpad or mouse, etc. The processor may call the logic instructions in the memory.

[0161] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0162] In one embodiment of the present invention, a computer program product is provided, wherein the computer program product includes a computer program stored on a non-transitory computer-readable storage medium, wherein the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the methods provided by the above-mentioned method embodiments.

[0163] In one embodiment of the present invention, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium stores server instructions, and the computer instructions enable a computer to execute the methods provided in the above embodiments.

[0164] The above embodiment provides a computer-readable storage medium, whose implementation principle and technical effect are similar to those of the above method embodiment, and will not be repeated here.

[0165] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0166] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0167] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0168] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these 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 invention.

Claims

1. A method for intelligent sludge discharge scheduling in a sedimentation tank of a water supply plant, characterized in that: include: Obtaining label data for training a model for estimating the initial solid content of muddy water and online instrument data for training the model for estimating the initial solid content of muddy water, constructing feature data for training the model for estimating the initial solid content of muddy water based on the online instrument data, so as to train the model for estimating the initial solid content of muddy water and obtain a model for estimating the change of solid content of muddy water; The sludge water solid content change prediction model is calibrated by obtaining the sample data of the sludge water solid content during the sludge discharge process, and the sludge water solid content change prediction model is checked based on the sludge volume; The target average solid content of the mud water is set, and the estimated solid content of the mud water obtained by the calibrated mud water solid content change prediction model is compared with the target average solid content of the mud water to determine the opening and closing time of the mud valve.

2. The intelligent sludge discharge scheduling method for a sedimentation tank in a water supply plant according to claim 1, characterized in that: Based on the online instrument data, characteristic data for training the initial solid content estimation model of the sludge water is constructed to train the initial solid content estimation model of the sludge water, including: Set the initial hyperparameters and loss function of the model, and divide the test set; Use K-Fold cross-validation strategy to divide the training set and validation set for training; The hyperparameters with the lowest total loss function are selected as the optimal hyperparameters for training, and the results output by each model are averaged on the test set to obtain the final estimated results.

3. The intelligent sludge discharge scheduling method for a sedimentation tank in a water supply plant according to claim 1, characterized in that: The construction of a prediction model for the change of solid content in sludge water includes: obtaining sample data of solid content in sludge water during the sludge discharge process, and constructing a prediction model for the change of solid content in sludge water according to the zoning law of crowded sedimentation; The prediction model of the solid content change of sludge water is: Where t is the time after the mud discharge valve is opened, in seconds; S(t) is the solid content of the mud discharge water at time t, S(0) is the initial solid content of the mud discharge water, and a, b, m, and n are model parameters to be calibrated that are greater than 0.

4. The intelligent sludge discharge scheduling method for a sedimentation tank in a water supply plant according to claim 1, characterized in that: Calibrate the prediction model of the solid content change of the sludge water, including: A loss function is set, and sample data collected multiple times are substituted into the loss function respectively, with minimization of the loss function as the optimization goal, to solve the values ​​of multiple sets of fixed-value parameters and variable-value parameters of the model to be calibrated; The fixed parameters of multiple models are averaged to obtain the final fixed parameters. A scatter plot of the variable parameters obtained from each model with respect to the initial solid content of the mud water is made to determine the functional relationship between the variable parameters and the initial solid content of the mud water. The function parameters are determined with the goal of minimizing the loss function.

5. The intelligent sludge discharge scheduling method for a sedimentation tank in a water supply plant according to claim 1, characterized in that: The prediction model of the change of solid content in the sludge water is calibrated based on the amount of sludge, including: Compare and verify the estimated dry mud generation with the dry mud volume shipped out of the water plant to determine the estimated dry mud generation; The sludge discharge volume is estimated by the sludge water solid content change prediction model, and the sludge discharge volume is compared with the estimated dry mud production. If the error between the two is within the allowable range, the sludge water solid content change prediction model is verified to be successful.

6. The intelligent sludge discharge scheduling method for a sedimentation tank in a water supply plant according to claim 1, characterized in that: Determine the opening and closing time of the mud discharge valve, including: Set the target average solid content of sludge water; Grid sampling candidate values ​​for the threshold of initial solid content of mud water that determines the opening of mud valve; For each candidate value of the initial solid content threshold of the mud discharge water, determine the corresponding mud discharge valve closing time; For each candidate solution, calculate the average solid content of the sludge water; The initial solid content threshold of the mud discharge water corresponding to the result closest to the target average solid content of the mud discharge water is taken as the threshold for determining the opening of the mud discharge valve.

7. The intelligent sludge discharge scheduling method for a sedimentation tank in a water supply plant according to claim 6, characterized in that: When the estimated initial solid content of the mud discharge water reaches the threshold, the mud discharge valve is opened; when the mud discharge valve opening time reaches the mud discharge valve closing time, the mud discharge valve is closed.

8. An intelligent sludge discharge scheduling system for a sedimentation tank in a water supply plant, characterized in that: include: A model building module, which obtains label data for training a model for estimating the initial solid content of mud water and online instrument data for training the model for estimating the initial solid content of mud water, and constructs feature data for training the model for estimating the initial solid content of mud water according to the online instrument data, so as to train the model for estimating the initial solid content of mud water and obtain a model for estimating the change of the solid content of mud water; The model verification module calibrates the prediction model of the change of the solid content of the sludge water by obtaining the sample data of the solid content of the sludge water during the sludge discharge process, and verifies the prediction model of the change of the solid content of the sludge water based on the amount of sludge; The mud discharge valve opening and closing time determination module sets the target average solid content of the mud discharge water, and compares the estimated mud discharge water solid content obtained by the verified mud discharge water solid content change prediction model with the target average solid content of the mud discharge water to determine the mud discharge valve opening and closing time.

9. A computer-readable storage medium storing one or more programs, characterized in that: The one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform any one of the methods of claims 1 to 7.

10. A computing device, characterized in that include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured as the one or more processors Execute, the One or more programs include a method for executing any one of the methods described in claims 1 to 7 instruction.

Images