Phosphorus removal intelligent dosing control method and system based on big data prediction model

Abstract

The invention discloses a dephosphorization intelligent dosing control method and system based on a big data prediction model, and belongs to the technical field of the water affair industry. The method comprises the steps that production operation and maintenance data of a sewage treatment plant are analyzed, a prediction model based on big data is established to predict the concentration of PO4 < 3->-P in the chemical phosphorus removal process in sewage treatment, the dosage of a phosphorus removal agent is optimized through an intelligent agent adding control system, and the purpose of timely and accurately adding the phosphorus removal agent according to needs is achieved; the problems of hysteresis of long detection time and data accuracy of an online detection instrument are solved.

Description

Technical Field

[0001] The present invention relates to a method and system for intelligent dosing control of phosphorus removal based on a big data prediction model, and belongs to the technical field of water industry. Background Art

[0002] The water quality of sewage from sewage treatment plants is complex and volatile. The composition of sewage varies greatly under different regions and climatic conditions, and is affected by the living habits of local residents and the development of enterprises. The main pollutants of general municipal sewage are nitrogen, phosphorus and organic matter, which are also the core indicators of treatment and assessment of sewage treatment plants. Among them, phosphorus pollutants rely on biological phosphorus removal on the one hand, and chemical removal by adding chemicals on the other hand to meet the strict emission standards of sewage treatment plants. With the increasingly stringent water quality emission standards in recent years, the total phosphorus emission standard has also been increased from 0.5mg / L to 0.3mg / L and even 0.2mg / L. Due to the limitation of biological phosphorus removal capacity, the dependence on chemical phosphorus removal in sewage treatment is getting higher and higher.

[0003] Chemical phosphorus removal is to precipitate or adsorb phosphorus compounds in wastewater by adding chemical reagents. The basic principle is to form insoluble phosphate precipitates by adding chemical reagents, and finally remove phosphorus from sewage by solid-liquid separation. Chemical phosphorus removal consumes reagents and will produce indirect carbon emissions. In addition to ensuring that the effluent quality meets the standards, sewage treatment also needs to reduce pollution and carbon emissions as much as possible. Therefore, it is necessary to determine the most appropriate amount of chemical reagents to be added based on the content of phosphorus pollutants in the sewage.

[0004] At present, most sewage treatment plants mainly rely on manual experience to add chemical agents. There are three main ways of adding chemicals: ① Constant addition over a period of time, ② Determine the addition amount based on the water volume, and ③ Determine the addition amount based on the total phosphorus concentration of the effluent. The above-mentioned dosing methods add a constant amount of chemicals over a period of time, which may cause water quality to exceed the standard due to insufficient addition, or cause waste of chemicals and increase carbon emissions due to excessive addition.

[0005] In order to solve the above problems, some sewage treatment plants have adopted intelligent phosphorus removal and dosing control systems, which mainly realize intelligent dosing based on the feedforward adjustment of the phosphate concentration of the inlet water of the chemical phosphorus removal unit and the feedback adjustment of the phosphate concentration of the effluent water; among them, the phosphate detection instrument is an important equipment of the intelligent phosphorus removal and dosing control system. By continuously monitoring the phosphate concentration in the effluent water, the effluent phosphate data is fed back to the intelligent phosphorus removal and dosing control system, and real-time data can be provided to operators to optimize the chemical phosphorus removal process; and due to the limitations of the current phosphate detection method, the phosphate online detection instrument is a chemical detection method, that is, the phosphate content in the water body is measured through a specific chemical reaction; and these chemical reaction processes generally last for a period of time, so the detection time of the phosphate online detection instrument is at least 5 to 10 minutes. In order to ensure the performance of the instrument, the detection cycle is generally not less than 30 minutes, which belongs to non-continuous monitoring, which will lead to the inability to realize real-time monitoring of the treatment process, and the phosphate data obtained has a lag, which makes the intelligent phosphorus removal and dosing control system also have a lag, and cannot timely add the corresponding agent to the "current" detection concentration.

[0006] In addition, the daily maintenance of phosphate online detection instruments has a great impact on the accuracy of data. Once the instrument maintenance is not in place, the reagents are not replenished or replaced in time, the accuracy of the detection data will be greatly reduced. And the instrument detection environment conditions have a great impact on the accuracy of the instrument, such as sludge concentration, temperature, etc., and the existing intelligent phosphorus removal and dosing control system usually does not take these factors into account. Summary of the invention

[0007] Aiming at the hysteresis and data accuracy problems of phosphate online detection instrument data in sewage treatment plants, the present invention provides a phosphorus removal intelligent dosing control method and system based on big data prediction model. On the basis of giving full play to the biological phosphorus removal capacity, a way to fully and effectively analyze and utilize the production and operation data of sewage treatment plants is established. Based on the big data prediction model, the PO4 3- -P concentration, optimize the dosage of phosphorus removal agent through the intelligent dosing control system, so as to achieve the purpose of timely and accurate dosing of phosphorus removal agent on demand, and propose management methods and systems for the collection, processing, analysis and display of historical data. When there is an abnormal situation, the system will send prompts and early warning signals to the "monitor" or "executor".

[0008] A phosphorus removal intelligent dosing control method and system based on a big data prediction model includes a data acquisition module, a data management module, a big data prediction model, an intelligent dosing control model, an early warning alarm module and a user module.

[0009] The data acquisition module is used to collect historical and real-time data such as water quality, water volume, temperature, MLSS, MLVSS, reflux ratio, HRT, SRT, phosphorus release rate and potential, dosage, agent concentration, etc. of the sewage treatment plant's inlet, pretreatment, biological treatment, secondary sedimentation tank, deep treatment, effluent and other process units, including the collection of online and offline data.

[0010] Among them, historical data is the operation and maintenance data of the sewage treatment plant over the years, and real-time data is composed of online data obtained by online detection instruments in real time and the most recently entered offline data. Online data is data detected in real time by online detection instruments and transmitted through mainstream communication protocols; offline data is data that cannot be obtained in real time through instruments and needs to be entered manually. Manual entry of offline data uses a data import template that combines the work record requirements of laboratory personnel and the system data input requirements to enter historical data and new data.

[0011] Furthermore, water quality data include COD, BOD5, TP, PO4 3- -P, DO, ORP, etc. The biological treatment and deep treatment process units are the main units for phosphorus removal control.

[0012] The data governance module is used to screen, organize, analyze, transform, store and manage the collected big data. This includes the deviation analysis of online detection data. The deviation analysis of online detection instruments is performed based on the acquired historical data. First, DO, ORP, MLSS, COD, PO4 3- -P or TP and other indicators are compared and analyzed with the offline data, and the online detection data is recorded as the pollutant concentration measurement value C m , the offline data result is recorded as the reference value C p , the difference between the analytical value and the reference value ε=C m / C p . Analyze the distribution law of ε under different temperature, sludge concentration and other conditions, and obtain the calibration coefficient ε` under the corresponding conditions of online data. Therefore, the data C after the online detection instrument is calibrated mo =ε`·C m .

[0013] Furthermore, the correlation and trend between the historical data after treatment are analyzed, and a big data prediction model is established through value mining. The real-time collected influent water volume, COD, BOD5, TP, biological pool water temperature, MLSS, MLVSS, reflow ratio, HRT, SRT, phosphorus release rate and potential, and reagent concentration are used as feedforward data to evaluate the current phosphorus concentration TP after biological phosphorus removal based on the big data prediction model. Ino (PO Ino), dosing time, and sampling time after the phosphorus removal process to detect and evaluate the "real" phosphorus removal efficiency. The phosphorus concentration data after biological phosphorus removal is transmitted to the intelligent dosing control model. The intelligent dosing control model is based on the current phosphorus removal agent dosage of the system and the dosage ratio R of different initial phosphorus concentrations entering the chemical phosphorus removal unit after biological phosphorus removal. x Calculate the dosage of the reagent Q y , and the dosage of the agent is Q y The data are transmitted to the dosing execution unit and the big data prediction model. The execution unit allocates the dosage of each pump for chemical dosing according to the operating status and quantity of the pump. The big data prediction model predicts the phosphorus concentration of the effluent after chemical phosphorus removal.

[0014] The phosphorus concentration TP of the chemical phosphorus removal process after biological phosphorus removal predicted by the big data prediction model Ino (PO Ino ) and the effluent phosphorus concentration test value TP of biological phosphorus removal Inmo (PO Inmo ), if the relative error of the predicted phosphorus concentration after biological phosphorus removal and entering the chemical phosphorus removal process is greater than 10%, that is, |TP Inmo -TP Ino | / TP Inmo >10%(|PO Inmo -PO Ino | / PO Inmo >10%), the built-in parameters of the big data prediction model are optimized.

[0015] Among them, TP Ino (PO Ino ) is a big data prediction model to predict the effluent TP (PO4 3- -P) concentration (i.e., the predicted chemical phosphorus removal unit influent TP (PO4 3- -P) concentration); TP Inmo (PO Inmo ) is the effluent TP after biological phosphorus removal (PO4 3- -P) concentration detection value (i.e., chemical phosphorus removal unit influent TP (PO4 3- -P) concentration detection value, corrected).

[0016] Prediction of effluent phosphorus concentration TP after chemical phosphorus removal eo (PO eo ), TP eo (PO eo ) and the preset value TP c (PO c ) for comparison. If the predicted value of effluent phosphorus concentration after chemical phosphorus removal exceeds 1.2 times the preset value, that is, TP eo >1.2TP c(PO eo >1.2PO c ), a water quality warning is issued. Otherwise, it is compared with the effluent test value TP after chemical phosphorus removal. emo (PO emo ) for comparison, if the relative error of the predicted value of the effluent phosphorus concentration after chemical phosphorus removal is greater than 10%, that is, |TP emo -TP eo | / TP emo >10%(|PO emo -PO eo | / PO emo >10%, the big data prediction model is optimized.

[0017] Among them, TP eo (PO eo ) is the effluent TP (PO4 3- -P) concentration; TP c (PO c ) is the effluent TP after chemical phosphorus removal (PO4 3- The preset concentration value of TP is 0.1-0.4 mg / L, which is determined according to the standards of each sewage treatment plant and after analysis of historical data; emo (PO emo ) is the effluent TP after chemical phosphorus removal (PO4 3- -P) concentration detection value (corrected).

[0018] Detect the phosphorus concentration TP of the wastewater after chemical phosphorus removal at the effluent of the chemical phosphorus removal unit emo (PO emo ), if the phosphorus concentration of the effluent after chemical phosphorus removal exceeds 1.2 times the preset value, that is, TP emo >1.2TP c (PO emo >1.2PO c ), the intelligent dosing control model optimizes and adjusts the dosing amount Q y The maximum value is immediately executed and an emergency alarm is issued. The staff will check and decide whether to activate the emergency plan. Otherwise, the model feedback coefficient is adjusted.

[0019] The effluent TP(PO4 3- -P), secondary sedimentation tank effluent TP (PO4 3- -P), deep phosphorus removal unit effluent TP (PO4 3- -P) is used as feedback data to make feedback adjustments to the big data prediction model and intelligent dosing control model to continuously optimize the model.

[0020] Calculation formula for the dosage of the reagent:

[0021] (1) When TP is the monitoring target:

[0022] Q y =[K1(TP In -TP c )+K2(TP e -TP c )]*10-6*(Q*1000 / 2)*R x ;

[0023] (2) PO4 3- -P is the monitoring target:

[0024] Q y =[K1(PO In -PO c )+K2(PO e -PO c )]*10-6*(Q*1000 / 2)*R x .

[0025] Where:

[0026] Q y : dosage of phosphorus removal agent; K1: front feedback coefficient; K2: rear feedback coefficient; R x : The dosage of phosphorus removal agent and the dosage ratio of different initial phosphorus concentrations entering the chemical phosphorus removal unit after biological phosphorus removal are determined according to the field bench test simulation test; TP In is the TP concentration of the influent of the chemical phosphorus removal process, TP e is the TP concentration of the effluent from the chemical phosphorus removal process, TP c It is the preset value of TP concentration of effluent after chemical phosphorus removal; PO In PO4 for chemical phosphorus removal process 3- -P concentration, PO e PO4 is the effluent from the chemical phosphorus removal process 3- -P concentration, PO c is the preset value of TP concentration of effluent after chemical phosphorus removal, and Q is the wastewater treatment flow rate of the chemical phosphorus removal unit.

[0027] The early warning and alarm module and the user module are used to warn and alarm abnormal situations in the control process, and provide users with reference data and operation functions. In addition to the above-mentioned early warning and alarm, when the inlet TP concentration is abnormal (exceeding the historical highest TP MAX), reduced biological phosphorus removal capacity, and abnormally increased phosphorus concentration in the inlet of the chemical phosphorus removal unit will all be warned and alarmed. After the staff investigates the cause, they can choose to retain or delete the abnormal value and filter the data entered in the database. The user enters the instrument's usage time and operation and maintenance content and cycle. The system automatically generates operation and maintenance plan reminders based on the most recent maintenance date i and maintenance cycle T to ensure the normal operation of the instrument.

[0028] Furthermore, diversified charts are used to demonstrate the differences between the analysis and calculation results of the big data prediction model and the actual effects, including differences in pollutant concentrations (the differences are compared in model calculation values, online detection values / manual test values), differences in agent dosage (model calculation values, actual dosage values), etc.

[0029] In summary, the present invention includes at least one of the following beneficial effects:

[0030] 1. Solve the problems of lag in online detection instruments and accuracy of detection data, and realize the integration, analysis and utilization of a large amount of production and operation data such as offline data of sewage treatment plants. The data types in the sewage treatment process are numerous and complex, and there is a strong correlation between various types of data. By establishing a big database for data governance and establishing a big data prediction model, the analysis results can be effectively used to discover contradictions.

[0031] The big data prediction model proposed in this application can judge the biological phosphorus removal capacity and predict the phosphorus concentration of the inlet and outlet water of the chemical phosphorus removal unit based on the current real-time influent water volume, TP concentration, COD (BOD5) concentration, biological pool water temperature, ORP, MLSS (MLVSS), DO, HRT, SRT, phosphorus release rate and potential, phosphorus removal agent concentration, etc. The predicted phosphorus concentration of the inlet water of the chemical phosphorus removal unit provides the dosage calculation data for the intelligent dosing control model in a timely manner, and the HRT and SRT calculations provide the agent addition time, which solves the lag of the long detection time of the online detection instrument. In addition, data is the core of intelligent control; the core treatment unit of the sewage treatment plant is the biological pool, and the key to the good operation of the biological pool lies in mastering the performance of the activated sludge in the biological pool, such as: phosphorus release and phosphorus removal performance, denitrification performance, nitrification performance, sludge sedimentation performance, etc. The above performance is not the data that can be obtained in real time by online detection equipment, and manual sampling is required to carry out testing and simulation experiments. This system has an offline data input channel. By improving the collection and management methods of basic data, it realizes the full utilization of offline data and automatically calibrates the online detection data to improve the accuracy of the data. It provides accurate and comprehensive data samples for the large database and compares and calibrates the online data.

[0032] 2. Solve the problem of untimely operation and maintenance of online detection instruments.

[0033] The degree and timeliness of the operation and maintenance of the instrument have a great impact on the accuracy of the online detection data. If the operation and maintenance are not timely, the accuracy of the online instrument detection data will be greatly reduced. Therefore, the present invention formulates an operation and maintenance plan according to the maintenance cycle and maintenance content of the instrument, automatically generates a plan, and reminds the staff in time to reduce the data error caused by the instrument failure.

[0034] 3. The intelligent control effect can be analyzed in real time.

[0035] It allows system users to intuitively understand the effectiveness of the system, provides data reference for decision makers, and saves a lot of manpower analysis work. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0037] Figure 1 It is a flow chart of the intelligent dosing control method for phosphorus removal based on the big data prediction model in the embodiment of the present application;

[0038] Figure 2 It is a maintenance plan flow chart;

[0039] Figure 3 It is a commonly used process flow chart for sewage treatment plants. DETAILED DESCRIPTION

[0040] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0041] like Figure 3 As shown, the main treatment processes of municipal sewage treatment plants including nitrogen and phosphorus removal include but are not limited to:

[0042] ① Biological pool for biological denitrification and phosphorus removal + secondary sedimentation tank for mud-water separation + denitrification deep bed filter for deep denitrification. The reagent dosing points include the effluent from the biological pool, the effluent from the secondary sedimentation tank, and the inlet of the denitrification deep bed filter, corresponding to Figure 3 The dosing point (a), dosing point (b) and dosing point (c) in FIG.

[0043] ② Biological pool with biological denitrification and phosphorus removal as the main method + secondary sedimentation tank for mud and water separation + high-efficiency sedimentation tank for deep phosphorus removal + denitrification deep bed filter for deep nitrogen removal. The reagent addition points include the effluent from the biological pool and the inlet from the high-efficiency sedimentation tank, corresponding to Figure 3 The dosing point (a) and dosing point (d) in FIG.

[0044] ③ Biological pool with biological denitrification and phosphorus removal as the main method + secondary sedimentation tank for mud and water separation + flotation tank for deep phosphorus removal + denitrification deep bed filter for deep nitrogen removal. The reagent addition points include the effluent of the biological pool and the inlet of the flotation tank, corresponding to Figure 3 The dosing point (a) and dosing point (e) in FIG.

[0045] Taking polyaluminium chloride PAC as a chemical phosphorus removal agent as an example, the agent addition point is Figure 3 Any one or more combinations of the five addition points (a), (b), (c), (d) and (e).

[0046] Embodiment 1

[0047] This embodiment provides a method for intelligent dosing control of phosphorus removal based on a big data prediction model. The flowchart of the method is as follows: Figure 1 As shown, the method includes:

[0048] Step 1: Obtain historical data and current real-time data of operation and maintenance of the sewage treatment plant over the years;

[0049] Among them, the current real-time data is composed of the online data obtained in real time by the online detection instrument and the most recent offline data entered, including water quality data, water volume, temperature, MLSS (MLVSS), reflux ratio, HRT, SRT, phosphorus release rate and potential, and reagent concentration of the sewage treatment plant's inlet, pretreatment, biological pool, secondary sedimentation tank, deep treatment, and effluent process units; water quality data includes COD, BOD5, TP, PO4 3- -P, DO, ORP, etc.; in the online detection data, the water quality data of the pretreatment unit refers to the water quality data after the water passes through the coarse screen, fine screen and grit chamber, and the water quality data of the deep phosphorus removal unit refers to the water quality data of the denitrification deep bed filter, high-efficiency sedimentation tank or flotation tank.

[0050] Among them, online data refers to data detected in real time on-site by online detection instruments and transmitted through mainstream communication protocols; offline data refers to data that cannot be obtained in real time through instruments and needs to be entered manually.

[0051] The historical data of operation and maintenance of sewage treatment plants over the years are composed of online data obtained in real time by online detection instruments and entered offline data. In addition to the water quality data, water volume and temperature of process units such as sewage treatment plant inlet, pretreatment, biological pool, secondary sedimentation tank, deep treatment and effluent, it also includes MLSS (Mixed Liquor Suspended Solids), MLVSS (Mixed Liquor Volatile Suspended Solids), ORP (Oxidation Reduction Potential), reflux ratio, HRT (Hydraulic Retention Time), SRT (Sludge Retention Time), phosphorus release rate and potential, dosage, agent concentration, etc.

[0052] Step 2: Analyze the deviation of online detection data

[0053] First, DO, ORP, MLSS, COD, PO4 3- -P or TP and other indicators of the online data and offline data for comparison and analysis, the online detection data of each indicator is recorded as C m , the offline data is recorded as the reference value C p , the difference between the analytical value and the reference value ε=C m / C p . Analyze the distribution law of ε under different temperature, sludge concentration and other conditions, and obtain the calibration coefficient ε` under the corresponding conditions of online data. Therefore, the data C after the online detection instrument is calibrated mo =ε`·C m Furthermore, if the online data is abnormal due to failure, standard solution calibration, etc., this abnormal data is removed during model training, and the offline data entered at the corresponding time is used as the model input data.

[0054] Step 3: Establish a big data prediction model based on the correlation and trend of historical data of operation and maintenance of the sewage treatment plant over the years;

[0055] Based on the operation and maintenance data of the sewage treatment plant over the years (influent water volume, COD, BOD5, TP, etc., biological pool MLSS, MLVSS, ORP, DO, reflow ratio, SRT, HRT, phosphorus release rate and potential, etc., chemical phosphorus removal unit inlet and outlet TP (PO4 3--P) concentration, agent dosage, phosphorus removal agent dosage and chemical phosphorus removal unit influent phosphorus concentration dosage ratio, etc.) are used as feedforward or feedback data of the big data prediction model, and a data sample training library for the big data prediction model is established. According to the process operation characteristics of the sewage treatment plant, data sorting and analysis rules are formulated. Through intelligent algorithms, the correlation between biological phosphorus removal capacity and different influent BOD5 / COD, BOD5 / TP or COD / TP, biological pool MLSS or MLVSS, HRT, SRT, phosphorus release rate and potential, reflux ratio, water temperature and other conditions is established to establish a big data prediction model.

[0056] The big data prediction model is input with influent volume, COD, BOD5, TP, biological pool MLSS, MLVSS, ORP, DO, reflow ratio, SRT, HRT, phosphorus release rate and potential, etc. The model outputs the phosphorus concentration of the inlet and outlet water of the chemical phosphorus removal unit.

[0057] Step 4: Based on the big data prediction model, intelligent dosing control for phosphorus removal is performed according to real-time collected data;

[0058] The real-time collected influent volume, COD, BOD5, TP, biological pool water temperature, MLSS, MLVSS, reflow ratio, HRT, SRT, phosphorus release rate and potential, and reagent concentration are used as feedforward data to input into the big data prediction model. The current TP (PO4 3- -P) concentration (i.e. TP (PO4 3- -P) concentration), dosing time, and sampling time of water quality after phosphorus removal process. 3- -P) concentration (i.e. TP (PO4 3- -P) concentration) is sent to the intelligent dosing control model to calculate the dosage Q y , and then the dosage of the agent is Q y The data is transmitted to the dosing execution unit and the big data prediction model. The execution unit allocates the dosage of each dosing pump according to the operation status of the dosing pump, such as start, stop, failure or maintenance, and the number of dosing pumps, and controls the dosing pump to dosing the agent. The big data prediction model predicts the effluent TP (PO4 3- -P) concentration. The intelligent dosing control model and big data prediction model timely optimize and adjust the model's feedback coefficient, dosing coefficient, built-in parameters, etc. according to the effluent phosphorus concentration feedback of the phosphorus removal unit.

[0059] Step 5, analyzing the difference between the predicted value and the detected value and optimizing the model parameters;

[0060] The effluent phosphorus concentration TP after biological phosphorus removal predicted by the big data prediction model Ino (POIno ) and the effluent phosphorus concentration TP after chemical phosphorus removal eo (PO eo ), the phosphorus concentration of the effluent after biological phosphorus removal TP Inmo (PO Inmo ) and the phosphorus concentration of the effluent after chemical phosphorus removal TP emo (PO emo ), preset phosphorus concentration TP of effluent c (PO c ) for analysis and comparison:

[0061] If |TP Inmo -TP Ino | / TP Inmo >10%(|PO Inmo -PO Ino | / PO Inmo >10%), the built-in parameters of the big data prediction model are optimized;

[0062] If |TP emo -TP eo | / TP emo >10%(|PO emo -PO eo | / PO emo >10%), the built-in parameters of the big data prediction model are optimized.

[0063] Step 6: abnormal situation warning and alarm;

[0064] Phosphorus concentration TP of effluent after chemical phosphorus removal predicted by big data prediction model eo (PO eo ), the phosphorus concentration of the effluent after chemical phosphorus removal TP emo (PO emo ), preset phosphorus concentration TP of effluent c (PO c ) for analysis and comparison:

[0065] If TP eo >1.2TP c (PO eo >1.2PO c ), then water quality warning will be issued;

[0066] If TP emo >1.2TP c (PO emo >1.2PO c ), the intelligent dosing control model is optimized and the maximum dosing amount is immediately executed, and an emergency prompt alarm is issued.

[0067] In addition to the above warnings and alarms, when the inlet TP concentration is abnormal (exceeds the historical maximum TP MAX ), reduced biological phosphorus removal capacity, and abnormally increased phosphorus concentration in the chemical phosphorus removal influent will all trigger early warnings and alarms.

[0068] Embodiment 2

[0069] This embodiment provides an intelligent dosing control system for phosphorus removal based on a big data prediction model. The system is implemented based on the method proposed in Example 1, and includes: a data acquisition module, a data governance module, a big data prediction model, an intelligent dosing control model, an early warning alarm module and a user module.

[0070] The data collection module collects historical data and real-time data from the sewage treatment plant as a database for subsequent data management and building a big data prediction model;

[0071] Historical data include water quality, water volume, and temperature data of process units such as influent, pretreatment, biological treatment, secondary sedimentation tank, deep treatment, and effluent. Real-time data include MLSS, MLVSS, reflux ratio, HRT, SRT, phosphorus release rate and potential, dosage, and reagent concentration. Water quality data include COD, BOD5, TP, PO4 3- -P, DO, ORP, etc.

[0072] The data governance module ensures the accuracy of the output of the big data prediction model by screening, sorting, analyzing, converting, storing and managing the data obtained by the data collection module;

[0073] DO, ORP, MLSS, COD, PO4 3- -P or TP and other indicators online data and offline data are compared and analyzed to obtain the difference between the measured value and the reference value ε = C m / C p , and analyze the ε distribution law under different temperature, sludge concentration and other conditions to obtain the calibration coefficient under the corresponding conditions of the online data ε `, get the data C after the online detection instrument is calibrated mo =ε`·C m ; In the case of abnormal online data, the offline data entered at the corresponding time is used as the output data of the big data prediction model.

[0074] The big data prediction model conducts correlation and trend analysis based on the data obtained from the data governance module to build a big data prediction model;

[0075] The real-time collected influent volume, COD, BOD5, TP, biological pool water temperature, MLSS, MLVSS, reflow ratio, HRT, SRT, phosphorus release rate and potential, and reagent concentration are used as feedforward data to input into the big data prediction model, and the current phosphorus concentration TP after biological phosphorus removal is predicted based on the big data prediction model. Ino or PO Ino , dosing time and sampling time after the phosphorus removal process are used to evaluate the "real" phosphorus removal efficiency of the system.

[0076] The intelligent dosing control model receives the output results from the big data prediction model and calculates the dosage Q of the agent based on the output data. y And transmit it to the dosing execution unit and the big data prediction model;

[0077] When TP concentration is used as the monitoring target, the dosage of the reagent Q y The expression is:

[0078] Q y =[K1(TP In -TP c )+K2(TP e -TP c )]*10 -6 *(Q*1000 / 2)*R x

[0079] PO4 3 --When P is the monitoring target, the dosage of the agent is Q y The expression is:

[0080] Q y =[K1(PO In -PO c )+K2(PO e -PO c )]*10 -6 *(Q*1000 / 2)*R x

[0081] Among them, K1: front feedback coefficient; K2: rear feedback coefficient; R x : The dosage of phosphorus removal agent and the dosage ratio of different initial phosphorus concentrations entering the chemical phosphorus removal unit after biological phosphorus removal; TP In is the TP concentration of the influent of the chemical phosphorus removal process, TP c The concentration of TP in the effluent after chemical phosphorus removal is set at 0.1-0.4 mg / L. e is the TP concentration of the effluent from the chemical phosphorus removal process, PO In PO4 for chemical phosphorus removal process 3- -P concentration, PO c PO4 is the effluent after chemical phosphorus removal3- -P concentration preset value, ranging from 0.1 to 0.4 mg / L, PO e PO4 is the effluent from the chemical phosphorus removal process 3- -P concentration, Q is the wastewater treatment flow rate of the chemical phosphorus removal unit;

[0082] The execution unit allocates the dosage of each pump according to the operating status and quantity of the pumps for agent addition, and the big data prediction model optimizes the model based on the feedback data;

[0083] If |TP Inmo -TP Ino | / TP Inmo >10%(|PO Inmo -PO Ino | / PO Inmo >10%), the built-in parameters of the big data prediction model are optimized;

[0084] If |TP emo -TP eo | / TP emo >10%(|PO emo -PO eo | / PO emo >10%), the built-in parameters of the big data prediction model are optimized.

[0085] The early warning alarm module uses the phosphorus concentration TP of the effluent after chemical phosphorus removal predicted by the big data prediction model eo or PO eo , Phosphorus concentration test value of effluent after chemical phosphorus removal TP emo or PO emo , Preset phosphorus concentration of effluent TP c or PO c Perform analysis and comparison, and issue early warning alarms based on the comparison results:

[0086] If TP eo >1.2TP c (PO eo >1.2PO c ), then water quality warning will be issued;

[0087] If TP emo >1.2TP c (PO emo >1.2PO c ), the intelligent dosing control model is optimized and the maximum dosing amount is immediately executed, and an emergency prompt alarm is issued;

[0088] In addition, when the influent TP concentration exceeds the historical maximum TP MAX, reduced biological phosphorus removal capacity, and abnormally increased phosphorus concentration in the chemical phosphorus removal influent will all trigger early warnings and alarms.

[0089] The user module includes a data import template. Users use the data import template to enter the instrument's usage time, operation and maintenance, and maintenance plan. The maintenance plan flowchart is as follows: Figure 2 As shown in the figure, the system automatically generates operation and maintenance plan reminders based on the most recent maintenance date i and maintenance cycle T, and dispatches instrument maintenance or maintenance tasks to staff, including instrument calibration, manual monitoring comparison, reagent replacement, pretreatment part cleaning, etc. The differences between the analysis and calculation results of the big data prediction model and the actual effects are displayed in a variety of ways, such as charts, including differences in pollutant concentrations (the comparison of differences is based on model calculation values, online detection values / manual test values), differences in reagent dosage, etc.

[0090] Although the present invention has been disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Anyone familiar with this technology can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be based on the definition of the claims.

Claims

1. A phosphorus removal intelligent dosing control method based on a big data prediction model, characterized in that: The method comprises: Step 1: Obtain the historical data of the operation and maintenance of the sewage treatment plant over the years and the current real-time data; the historical data includes the online data obtained in real time by the online detection instrument and the offline data entered manually; the current real-time data includes the online data obtained in real time by the online detection instrument and the most recently entered offline data; Step 2: Perform deviation analysis based on the online data and offline data obtained in step 1; Step 3: Analyze the correlation and trend between the historical data obtained in step 1 and establish a big data prediction model; Step 4: Input the real-time collected data as feedforward data into the big data prediction model, and perform intelligent dosing control for phosphorus removal according to the prediction results of the big data prediction model; Step 5: Analyze the difference between the prediction results of the big data prediction model and the actual detection value, and optimize the parameters of the big data prediction model; Step 6: Compare the predicted value of the big data prediction model with the preset value, and issue an alarm when the predicted value exceeds the preset value.

2. The method according to claim 1, characterized in that The step 2 comprises: The online detection data is recorded as C m , the offline data is recorded as the reference value C p , the difference between the analytical value and the reference value ε=C m / C p ; Analyze the ε distribution law under different conditions and obtain the calibration coefficient ε` under the corresponding conditions of the online data; then the online instrument detection data after calibration is: C mo =e`·C m In addition, when the online data is abnormal, the offline data entered at the corresponding time is used as the model input data.

3. The method according to claim 2, characterized in that The step 3 comprises: The operation and maintenance data of the sewage treatment plant over the years are used as the feedforward or feedback data of the big data prediction model, and a data sample training library for the big data prediction model is established. Data collation and analysis rules are formulated according to the process operation characteristics of the sewage treatment plant. Through intelligent algorithms, the correlation between biological phosphorus removal capacity and different influent BOD5 / COD, BOD5 / TP or COD / TP, biological pool MLSS or MLVSS, HRT, SRT, phosphorus release rate and potential, reflux ratio, water temperature and other conditions is established to establish a big data prediction model.

4. The method according to claim 3, characterized in that The step 4 comprises: The real-time collected influent volume, COD, BOD5, TP, biological pool water temperature, MLSS, MLVSS, reflow ratio, HRT, SRT, phosphorus release rate and potential, reagent concentration, etc. are used as feedforward data to input into the big data prediction model. The big data prediction model outputs the current TP or PO4 entering the chemical phosphorus removal unit after biological phosphorus removal. 3- -P concentration, dosing time, sampling time of water quality after phosphorus removal process, and calculate the dosage Q of the agent y ; When TP concentration is used as the monitoring target, the dosage of the reagent Q y The expression is: Q y =[K1(TP In -TP c )+K2(TP e -TP c )]*10 -6 *(Q*1000 / 2)*R x PO4 3- -P is the monitoring target, the dosage of the agent is Q y The expression is: Q y =[K1(PO In -PO c )+K2(PO e -PO c )]*10 -6 *(Q*1000 / 2)*R x Among them, K1: front feedback coefficient; K2: rear feedback coefficient; R x : The dosage of phosphorus removal agent and the dosage ratio of different initial phosphorus concentrations entering the chemical phosphorus removal unit after biological phosphorus removal; TP In is the TP concentration of the influent of the chemical phosphorus removal process, TP c The concentration of TP in the effluent after chemical phosphorus removal is set at 0.1-0.4 mg / L. e is the TP concentration of the effluent from the chemical phosphorus removal process, PO In PO4 for chemical phosphorus removal process 3- -P concentration, PO c PO4 is the effluent after chemical phosphorus removal 3- -P concentration preset value, ranging from 0.1 to 0.4 mg / L, PO e PO4 is the effluent from the chemical phosphorus removal process 3- -P concentration, Q is the wastewater treatment flow rate of the chemical phosphorus removal unit; The dosage of the agent is Q y The data is transmitted to the dosing execution unit and the big data prediction model. The execution unit allocates the dosage of each dosing pump according to the operating status of the dosing pump, such as start, stop, failure or maintenance, and the number of dosing pumps, and controls the dosing pumps to add the agent.

5. The method according to claim 4, characterized in that The step 5 comprises: The effluent phosphorus concentration TP after biological phosphorus removal predicted by the big data prediction model Ino or PO Ino and the effluent phosphorus concentration TP after chemical phosphorus removal eo or PO eo , Phosphorus concentration test value of effluent after biological phosphorus removal TP Inmo or PO Inmo和 Phosphorus concentration test value TP of effluent after chemical phosphorus removal emo or PO emo , Preset phosphorus concentration of effluent TP c (PO c ) for analysis and comparison: If |TP Inmo -TP Ino | / TP Inmo >10%(|PO Inmo -PO Ino | / PO Inmo >10%), the built-in parameters of the big data prediction model are optimized; If |TP emo -TP eo | / TP emo >10%(|PO emo -PO eo | / PO emo >10%), the built-in parameters of the big data prediction model are optimized.

6. The method according to claim 5, characterized in that The step 6 comprises: The phosphorus concentration TP of the effluent after chemical phosphorus removal predicted by the big data prediction model eo or PO eo , Phosphorus concentration test value of effluent after chemical phosphorus removal TP emo or PO emo , Preset phosphorus concentration of effluent TP c or PO c For analysis and comparison: If TP eo >1.2TP c (PO eo >1.2PO c ), then water quality warning will be issued; If TP emo >1.2TP c (PO emo >1.2PO c ), the intelligent dosing control model is optimized and the maximum dosing amount is immediately executed, and an emergency prompt alarm is issued; In addition, when the influent TP concentration exceeds the historical maximum TP MAX , reduced biological phosphorus removal capacity, and abnormally increased phosphorus concentration in the chemical phosphorus removal influent will all trigger early warnings and alarms.

7. The method according to claim 6, characterized in that The historical data in step 1 include water quality data, water volume, and temperature of the sewage treatment plant inlet, pretreatment, biological pool, secondary sedimentation tank, deep treatment, and effluent process units, as well as MLSS, MLVSS, reflux ratio, HRT, SRT, phosphorus release rate and potential, dosage, and agent concentration; The water quality data include COD, BOD5, TP, PO4 3- -P, DO, ORP.

8. A phosphorus removal intelligent dosing control system based on a big data prediction model, characterized in that: The system is implemented based on the method described in any one of claims 1 to 7, and the system includes a data acquisition module, a data management module, a big data prediction model, an intelligent dosing control model, an early warning alarm module and a user module.

9. The system according to claim 8, characterized in that The user module includes a data import template. The user enters the instrument's usage time and operation and maintenance plan through the data import template. The system automatically generates operation and maintenance plan reminders based on the most recent maintenance date i and the maintenance cycle T, and dispatches instrument maintenance or maintenance tasks to staff, including instrument calibration, manual monitoring and comparison, reagent replacement, and pre-treatment cleaning content.

10. The system according to claim 9, characterized in that The system uses a diversified approach to demonstrate the differences between the analysis and calculation results of the big data prediction model and the actual effects, including differences in pollutant concentrations and differences in agent dosages.

Images