Intelligent detection and early warning system for sludge activity and early warning method based on intelligent detection and early warning system

Abstract

The invention relates to the technical field of sewage treatment, in particular to a sludge activity intelligent detection and early warning system based on specific oxygen uptake rate (SOUR) dynamic analysis, multi-parameter reverse traceability and expert decision linkage and an early warning method based on the sludge activity intelligent detection and early warning system. The system is composed of a sensing and monitoring system, a judgment and decision-making system and an expert database early warning system, and real-time detection and accurate early warning of sludge activity abnormity are achieved through dynamic baseline model construction, abnormity traceability engine operation and expert rule base matching. The SOUR grading dynamic baseline technology is innovatively adopted, and the problem that the false alarm rate of traditional fixed threshold value early warning is high is effectively solved; through a multi-parameter reverse traceability mechanism, accurate positioning of abnormal causes is realized; and in combination with an expert decision-making system, intelligent regulation and control suggestions can be automatically generated. Practice shows that the method not only can ensure that the water quality stably reaches the standard for a long time and improve the fine management level of sewage treatment, but also can remarkably reduce the energy consumption and the operation cost of the system, and has important engineering application value and popularization prospect.

Description

TECHNICAL FIELD

[0001] The present application relates to the sewage treatment technical field, specifically to a sludge activity intelligent detection and early warning system and an early warning method based thereon. BACKGROUND

[0002] The activated sludge method is the most widely used treatment method in the current water treatment process, and sludge activity as a core index reflecting the operation state of the biochemical treatment system, its monitoring and control is crucial to the sewage treatment effect. The traditional sludge activity monitoring mainly relies on manual methods such as SV30 and microscopic examination, which has problems such as response lag, strong subjectivity and inability to realize real-time early warning. Specific oxygen uptake rate (SOUR) as a key parameter directly reflecting the metabolic activity of microorganisms has been widely concerned in analyzing, evaluating and predicting the system operation state.

[0003] In recent years, specific oxygen uptake rate as an important parameter for detecting sludge biological activity has been widely valued by domestic and foreign experts in analyzing, evaluating and predicting the system operation state and treatment capacity. However, the traditional SOUR early warning adopts a fixed threshold, and when the influent load fluctuates or the water temperature changes, the false positive rate is high; single parameter threshold alarm cannot identify complex abnormal causes, and the deterioration of effluent quality and sludge activity anomaly are often treated separately, lacking effective reverse tracing ability; the traditional control system relies on manual experience, and it is difficult to realize fine control. Therefore, it is urgent to establish a dynamic coupling mechanism of SOUR and water quality / process parameters to realize rapid tracing and accurate control of abnormality.

[0004] In view of this, the present application provides a sludge activity intelligent detection and early warning system and an early warning method based thereon, which constructs a closed-loop optimization mechanism of "SOUR abnormality judgment→effluent quality feedback→control range real-time calibration→abnormality cause judgment", aiming to break through the bottleneck of difficult real-time diagnosis of sludge activity abnormality, eliminate the blind area of multi-parameter abnormality cause identification, replace the inefficient control mode driven by manual experience, and provide an effective solution for early warning, accurate control and energy saving of the biochemical system. SUMMARY

[0005] The purpose of the present application is to provide a sludge activity intelligent detection and early warning system and an early warning method based thereon, to solve the bottleneck of difficult real-time diagnosis of sludge activity abnormality, eliminate the blind area of multi-parameter abnormality cause identification, and lack of reverse tracing ability.

[0006] To achieve the above purpose, the present application adopts the following technical scheme: a sludge activity intelligent detection and early warning system, comprising a perception monitoring system, a judgment and decision system and an expert library early warning system; The perception monitoring system is responsible for obtaining and calculating water quality, sludge and process parameters, obtaining SOUR value data and associated parameter data associated with the SOUR value data, and constructing a water quality monitoring feature data set of the SOUR value data and the associated parameter data; The decision-making system is built-in with a dynamic baseline model and an abnormality tracing engine; The built-in dynamic baseline model is constructed with a deep learning library TensorFlow, and the built-in dynamic baseline model takes SOUR value as the main parameter, obtains historical SOUR value data and classifies the SOUR value data, determines the threshold range of SOUR value data at different classifications and the threshold range of associated parameter data at corresponding SOUR classification; The built-in dynamic baseline model includes an input layer for inputting a historical water quality monitoring feature data set, the historical water quality monitoring feature data set includes a training set, a validation set and a test set, a training model and an evaluation model are used to evaluate the performance of the model; According to the change of the influent parameter and the requirement of the required pollutant removal rate, the classification of the SOUR value to be executed at present is determined, the current SOUR value data obtained by the perception monitoring system is compared with the threshold range of its current execution classification, and it is judged whether the current SOUR value data is within the threshold range of its current execution classification; When the current SOUR value data deviates from the threshold range in its current execution classification, the abnormality tracing engine is started to reverse the detection of the associated parameter data to lock the key inducement; The expert library early warning system is used to trigger sludge activity abnormality alarm, synchronously output key inducement and generate control suggestion by matching expert library.

[0007] Further, the associated parameter data includes real-time monitoring data of water temperature, influent COD, influent ammonia nitrogen, sludge concentration MLSS, OUR, biological system effluent ammonia nitrogen, orthophosphorus, nitrate nitrogen, influent flow, external reflux flow, internal reflux flow, aeration air volume, residual sludge discharge flow, and detection and calculation data of sludge organic fraction f, sludge load F / M, sludge age SRT, internal reflux ratio RN, external reflux ratio RW, gas-water ratio, specific oxygen uptake rate SOUR and pollutant removal rate.

[0008] Further, the data obtained before the calculation of the perception monitoring system needs to be preprocessed, and the processed data is compressed and stored in the database of the system, providing a data basis for the calculation of the perception monitoring system; The preprocessing includes denoising processing, outlier detection and correction, data interpolation and smoothing, calibration and calibration, unit conversion and filtering processing.

[0009] Further, the built-in dynamic baseline model first classifies the historical SOUR value data according to a certain span, then calculates the skewness of the SOUR value in each primary classification to determine the distribution characteristics in the classification, determines the classification threshold according to the distribution characteristics, and classifies the SOUR value data; the current SOUR value is collected and recorded, and the classification threshold is automatically optimized according to the classification method, forming a built-in dynamic baseline model that adapts to different processing requirements; when the real-time SOUR is in the corresponding classification, the relationship between the mode and the mean value in the classification is determined by the same logic, and the standard threshold of each related parameter is determined.

[0010] Further, the historical SOUR value data includes the mean value of the SOUR value data of each day in the past 365 days.

[0011] Further, the abnormal source engine relies on the built-in dynamic baseline model to perform the following steps: S1, according to the change of the influent parameter and the required pollutant removal rate, the classification Li of the current SOUR value in the built-in dynamic baseline model is obtained; S2, the current SOUR value data is processed and compared with the built-in dynamic baseline model; S3, according to Li, the threshold range Pmin~Pmax of each associated parameter data Pi in the classification Li and the mean value Pm in the classification are fed back; S4, automatically calculate the amount of each associated parameter data Pi deviating from its respective threshold range, that is, the deviation |(Pi-Pm) / Pm|×100%, wherein Pi is the real-time parameter value, and Pm is the mean value of the parameter in the classification; S5, list the associated parameter data Pi with the largest deviation as the key inducement and sort it according to the deviation size, and output the Top3 key inducement information to the expert library early warning system.

[0012] Further, in S2, the perception monitoring system obtains SOUR value data every hour, and compares the current SOUR value data with the average value of the previous 71 consecutive data as the new current SOUR value data with the built-in dynamic baseline model.

[0013] Further, the expert library early warning system includes an expert knowledge graph, which contains water quality-activity composite rules; the water quality-activity composite rules include: Rule R1: ammonia nitrogen removal rate decreases, effluent ammonia nitrogen increases and SOUR is normal, output "increase aeration, increase sludge concentration, control SOUR to Li level, adjust Pi value to Pmin~Pmax"; Rule R2: ammonia nitrogen removal rate decreases, effluent ammonia nitrogen increases and SOUR decreases, output "increase aeration, increase sludge concentration, add carbon source, adjust Pi value to Pmin~Pmax"; Rule R3: total phosphorus removal rate decreases, effluent total phosphorus increases, SOUR is normal and biological pool effluent orthophosphorus is normal, output "check secondary sedimentation tank sludge, increase phosphorus removal agent dosage, increase dosage point"; Rule R4: total phosphorus removal rate decreases, effluent total phosphorus increases, SOUR is normal and biological pool effluent orthophosphorus increases, output "check anaerobic tank phosphorus release, ensure anaerobic tank DO <0.2 mg / L, check secondary sedimentation tank sludge, control SOUR to Li level, increase phosphorus removal agent dosage, increase dosage point to 2-3, adjust Pi value to Pmin~Pmax"; Rule R5: total phosphorus removal rate decreases, effluent total phosphorus increases, SOUR decreases and biological pool effluent orthophosphorus increases, output "check anaerobic tank phosphorus release, ensure anaerobic tank DO <0.2 mg / L, check secondary sedimentation tank sludge, increase aeration, add carbon source, increase phosphorus removal agent dosage, increase dosage point to 2-3, adjust Pi value to Pmin~Pmax"; Rule R6: total nitrogen removal rate decreases, effluent total nitrogen increases, SOUR is normal and biological pool nitrate increases, output "check internal reflux DO, control internal reflux DO <0.5 mg / L, control SOUR to Li level, add carbon source to anoxic tank, adjust Pi value to Pmin~Pmax"; Rule R7: total nitrogen removal rate decreases, effluent total nitrogen increases, SOUR decreases and biological pool nitrate increases, output "check internal reflux DO, control internal reflux DO <0.5 mg / L, add carbon source to anoxic tank, add nitrifying bacteria, adjust Pi value to Pmin~Pmax"; Rule R8: two or more of ammonia nitrogen, total phosphorus and total nitrogen removal rates decrease, effluent increases, SOUR drops sharply, output "check if there are toxic substances in the influent, adjust the influent to 80% of the design load, add carbon source, control SOUR, adjust Pi value to Pmin~Pmax".

[0014] The early warning method of the sludge activity intelligent detection and early warning system comprises the following steps: a. Real-time acquisition of SOUR value data and associated parameter data, the sampling frequency of SOUR value data and associated parameter data is 1 time / 60 min; wherein the sludge organic fraction f is manually detected, and the update frequency is 1 time / month; b. According to the change of the influent parameters and the required pollutant removal rate, determine the current required SOUR value classification to be executed; c. After data processing of the current SOUR value, compare it with the baseline model, perform baseline comparison once an hour, and judge whether the current SOUR value data deviates from the threshold range of its current execution classification; d. When the current SOUR value data deviates beyond the threshold range of its current execution classification, start multi-parameter reverse tracing, and troubleshoot key causes; e. Match the expert library early warning system, issue an alarm, and output expert suggestions to the operation and maintenance personnel terminal.

[0015] The beneficial effects of the present application are: 1. The present application innovatively adopts SOUR grading dynamic baseline technology, effectively solving the problem of high false alarm rate of traditional fixed threshold early warning; through the multi-parameter reverse tracing mechanism, the precise positioning of abnormal causes is realized; combined with the expert decision system, intelligent control suggestions can be automatically generated; 2. The present application not only can guarantee long-term stable water quality and improve the fine management level of sewage treatment, but also can significantly reduce system energy consumption and operation cost, and has important engineering application value and popularization prospect. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 is a system structure diagram of the present application; Figure 2 is an abnormal source tracing engine workflow diagram of the present application; Figure 3 is a decision tree reasoning engine workflow diagram of the present application. DETAILED DESCRIPTION

[0017] The technical solutions of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application.

[0018] Embodiments of the present application: The present embodiment provides a sludge activity intelligent detection and early warning system, as shown in Figure 1 , including a perception monitoring system, a decision-making system, and an expert library early warning system.

[0019] The perception monitoring system transmits real-time monitoring data and calculation parameters to the decision-making system, the decision-making system analyzes and processes the data, and then sends abnormal signals and key causes to the expert library early warning system, and the expert library early warning system generates early warning information and control suggestions and sends them to the terminal.

[0020] The perception monitoring system includes monitoring and calculation of water quality, sludge quality, and process parameters.

[0021] The perception monitoring system includes real-time monitoring of water temperature, influent COD, influent NH3-N (ammonia nitrogen), sludge concentration (MLSS), oxygen consumption rate (OUR), biological system effluent NH3-N, orthophosphate, nitrate, external reflux flow, internal reflux flow, residual sludge discharge flow, and calculation of sludge load (F / M), sludge age (SRT), internal reflux ratio, external reflux ratio, air-water ratio, specific oxygen consumption rate (SOUR), and manual input of sludge organic fraction f after artificial detection.

[0022] Wastewater treatment plants are usually equipped with water quality, sludge quality, process parameter, etc. wastewater analyzers to ensure comprehensive monitoring of the entire treatment process.

[0023] Different analysis techniques are used for water quality parameter detection according to different parameters such as water quality, sludge quality, and process parameters in wastewater treatment.

[0024] Wastewater analyzers use different analysis techniques to monitor and analyze different water quality, sludge quality, and process parameters, which can more accurately understand the composition and characteristics of wastewater. Here are some common water quality parameters and their corresponding analysis techniques: Water temperature: Function: Used to detect the temperature of wastewater in the biological tank.

[0025] Analysis technique: Thermistor method, thermocouple method.

[0026] Working principle: The thermistor method uses the property that the resistance value of a high-precision platinum resistor (such as Pt100) changes with temperature, and accurately calculates the temperature by measuring the change in its resistance value. The thermocouple method uses the Seebeck effect of the loop composed of two different metals to measure the temperature difference between the two ends.

[0027] Chemical oxygen demand (COD): Function: Used to detect the quality of influent water in wastewater treatment plants.

[0028] Analysis technique: High temperature and high pressure digestion-photometric method, molecular absorption spectrophotometric method.

[0029] Working principle: The high temperature and high pressure digestion-photometric method oxidizes organic matter in the sample under high temperature and high pressure conditions, and then measures the amount of oxidation product to determine COD. The molecular absorption spectrophotometric method measures the absorption of reduced chromium acid salt in the sample to determine COD.

[0030] Ammonia nitrogen (NH3-N): Function: Used to detect the quality of influent water and biological system effluent water in wastewater treatment plants.

[0031] Analysis technique: High performance liquid chromatography (HPLC), photometric method.

[0032] Working principle: HPLC detects ammonia nitrogen by partitioning and separating the solution in the chromatographic column. Photometric method is to use the reaction of ammonia nitrogen and reagent to produce a colorimetric reaction, and measure the light absorption to determine the concentration of ammonia nitrogen.

[0033] Sludge concentration (MLSS): Function: Used to detect the sludge concentration of the biological system in the wastewater treatment plant.

[0034] Analysis technique: Optical scattering method (transmission light, scattered light), microwave method, traditional gravimetric method.

[0035] Working principle: Optical scattering method is the first choice for online monitoring; the sensor emits a light beam to the water sample, and the suspended sludge particles will absorb and scatter light, by measuring the attenuation of transmitted light intensity (transmission method) or the change of scattered light intensity (scattering method), the sludge concentration can be calculated. Microwave method uses the degree of attenuation of microwave energy through sludge mixture to measure the concentration, which is less affected by bubbles and color. Gravimetric method is the laboratory reference method, which directly measures the mass of solid in unit volume of water sample by filtration, drying and weighing.

[0036] f is the organic fraction of sludge (f is constant in a certain period of time, such as monthly detection update): Function: Used to detect the content of organic matter in sludge.

[0037] Analysis technique: Gravimetric method (high temperature ignition method).

[0038] Working principle: Dry the sludge to constant weight and burn it in a muffle furnace; calculate the content of organic matter in sludge by weight loss.

[0039] Oxygen uptake rate (OUR): Function: Used to detect the activity of sludge in the biological system of the wastewater treatment plant.

[0040] Analysis technique: Dynamic respiration measurement method.

[0041] Working principle: Using oxygen uptake rate OUR respirometer, using dissolved oxygen (DO) probe to monitor the process of oxygen consumption by activated sludge microorganisms when degrading organic matter in real time, by recording the decline rate (slope) of dissolved oxygen concentration per unit time, the oxygen uptake rate of microorganisms can be calculated (usually unit is mgO2 / L·h). OUR is a key parameter for evaluating sludge activity and biochemical reaction process.

[0042] Orthophosphate: Function: Used to detect the effluent quality of the biological system in the wastewater treatment plant.

[0043] Analysis technique: Molecular absorption spectrophotometry, inductively coupled plasma mass spectrometry (ICP-MS).

[0044] Working principle: Molecular absorption spectrophotometry measures the concentration of phosphate by measuring its absorption at a specific wavelength. ICP-MS determines the concentration of phosphate by converting the phosphorus atoms in the sample into ions and then measuring the ions with a mass spectrometer.

[0045] Nitrogen: Function: Used to detect the effluent quality of the biological system of the sewage plant.

[0046] Analysis technique: UV absorption spectroscopy, ion selective electrode method, ion chromatography.

[0047] Working principle: UV absorption spectroscopy is the most commonly used online technology. Nitrate ions (NO3 - ) have a characteristic absorption peak in the ultraviolet spectral region (such as 210 nm), and the absorbance follows the Lambert-Beer law, so the concentration of nitrate can be directly determined by measuring the absorbance; this method is fast and does not require reagents. The ion selective electrode method uses a membrane electrode that responds specifically to nitrate ions, and the concentration is converted by measuring the potential difference it produces. Ion chromatography is a laboratory standard method that uses ion exchange columns to separate various ions in the water sample and then uses a conductivity detector for quantitative analysis, with high precision and the ability to measure multiple ions simultaneously.

[0048] Flow: Function: Used to detect the inflow, external return flow, internal return flow, aeration air volume, and residual sludge discharge flow of the biological system of the sewage plant.

[0049] Analysis technique: Open channel flow meter (ultrasonic level meter + weir / tank), full pipe flow meter (electromagnetic flow meter, ultrasonic flow meter).

[0050] Working principle: Open channel flow measurement: Most commonly used in the sewage channels within the plant.

[0051] By installing a standard weir plate (such as a Parshall trough or a triangular weir) in the open channel, using its inherent "liquid level-flow" relationship, first use a non-contact ultrasonic level meter to accurately measure the liquid level height upstream of the tank, and then automatically calculate the instantaneous flow rate through the built-in formula. Full pipe flow measurement: Electromagnetic flow meters are the absolute main force for full pipe measurement in sewage plants, which are based on Faraday's law of electromagnetic induction. When a conductive fluid (sewage) flows through a magnetic field, it generates an induced electromotive force proportional to the average flow rate, thereby accurately measuring the flow rate. Ultrasonic flow meters use the time difference method, which measures the time difference between the forward and reverse propagation of ultrasonic pulses to calculate the flow rate of the fluid, and then obtains the flow rate.

[0052] In the intelligent early warning system for sludge activity, the operation data generated by the analyzer may contain some data with non-stationary characteristics, i.e., non-stationary data, which need to be preprocessed before being calculated by the perception monitoring system.

[0053] Preprocessing is to intervene in the raw data generated by analysis, and store the processed data in the database of the system after compression.

[0054] In the grading early warning system of the control room in the sewage plant, the preprocessing of the raw data generated by the sewage analyzer is to improve the quality, accuracy and availability of the data. The following are some common preprocessing steps and their effects (including but not limited to these, which can be added or selected according to actual conditions, and this does not make an exhaustive list): Noise removal: Effect: Remove noise caused by equipment failure, signal interference or other reasons, which helps to ensure the stability and reliability of the data, and avoid misleading information caused by noise.

[0055] Outlier detection and correction: Effect: Detect and correct outliers that may be caused by equipment failure or other abnormal conditions. Correcting outliers can improve the consistency and authenticity of the data.

[0056] Data interpolation and smoothing: Effect: For data missing due to insufficient sampling or sensor failure, use interpolation method to fill in missing values. Smoothing operation helps to eliminate short-term fluctuations and better show the trend of the data.

[0057] Calibration and calibration: Effect: Periodic calibration and calibration are necessary for some sensors, which helps to ensure the accuracy and reliability of the measurement results, especially after a long time of operation.

[0058] Unit conversion: Effect: Convert the data generated by different sensors or instruments to the same unit, which helps to ensure the consistency and comparability of the data, and facilitates subsequent data analysis and reporting.

[0059] Filtering: Effect: Apply digital filtering techniques to eliminate high-frequency noise or fluctuations and retain meaningful signals, which helps to better show the trend and periodic changes of the data.

[0060] Store the processed data in the database of the system to provide data basis for the perception monitoring system calculation.

[0061] The perception monitoring system calculates the sludge load F / M, sludge age SRT, internal reflux ratio RN, external reflux ratio RW, gas-water ratio, specific oxygen consumption rate SOUR, and pollutant removal rate. According to the real-time state during detection, a dynamic baseline model is established. The following are some common water quality parameters and their corresponding calculation methods: Sludge load F / M: Effect: Directly determines the activity and metabolic state of microorganisms, affecting treatment efficiency, sludge settling properties, and final effluent quality.

[0062] Calculation formula: F / M = (Q * S0) / (V * X).

[0063] F / M: sludge loading [kg BOD5 / (kg MLSS·d)]; Q: influent flow rate [m 3 / d]; S0: influent BOD5 concentration [mg / L or kg / m 3 ]; V: effective volume of biological reaction tank (aeration tank) [m 3 ]; X: concentration of mixed liquor suspended solids (MLSS) in the aeration tank [mg / L or kg / m 3 ] Sludge age SRT: Effect: Key parameter for controlling sludge activity and nitrogen and phosphorus removal efficiency.

[0064] Calculation formula: SRT = (V * X) / (Q_w * X_r + Q_e * X_e).

[0065] SRT: sludge age [d]; V: effective volume of biological reaction tank [m 3 ]; X: MLSS concentration in the aeration tank [kg / m 3 ]; Q_w: daily discharge of residual sludge flow rate [m 3 / d]; X_r: residual sludge concentration [kg / m 3 ]; Q_e: daily effluent flow rate [m 3 / d]; X_e: effluent suspended solids (ESS) concentration [kg / m 3 ] (usually very small and can be ignored).

[0066] Internal reflux ratio RN: Effect: Internal reflux (nitrification liquid reflux) returns nitrate (NO3 - -N) produced in the aerobic tank to the front end of the anoxic tank, providing an electron acceptor for denitrifying bacteria, thereby completing the denitrification process.

[0067] Calculation formula: RN = R 内 / Q.

[0068] RN: internal reflux ratio [%]; R 内: Flow rate of internal reflux pump [m 3 / d]; Q: Influent flow rate [m 3 / d].

[0069] External reflux ratio RW: Function: The sludge precipitated in the secondary sedimentation tank is re-pumped to the front end of the aeration tank to supplement the active sludge lost in the aeration tank and maintain stable MLSS concentration in the tank.

[0070] Calculation formula: RW=R 外 / Q.

[0071] RW: External reflux ratio [%]; R 外 : Flow rate of external reflux pump [m 3 / d]; Q: Influent flow rate [m 3 / d].

[0072] Gas-water ratio: It refers to the ratio of the amount of gas introduced per hour to the amount of water per hour.

[0073] Function: By adjusting the gas-water ratio (volume ratio of gas to wastewater), the intermittent operation time of the aeration device can be optimized to achieve dynamic balance of energy consumption; a reasonable gas-water ratio can meet the oxygen demand of microorganisms and avoid excessive aeration causing energy waste.

[0074] SOUR: Function: It represents the oxygen consumption of unit mass of active sludge per unit time, and is a sensitive indicator for measuring the strength of sludge activity and judging whether the influent contains toxic substances.

[0075] Calculation formula:

[0076] SOUR: Specific oxygen uptake rate [mgO2 / (gMLSS·h)]; OUR: Oxygen uptake rate [mgO2 / (L·h)], i.e. oxygen consumption per unit volume of sludge liquid per unit time; MLSS: Mixed liquid sludge concentration for measurement [g / L]; f is the organic fraction of sludge; θ is the temperature coefficient, which is 1.07 when the water temperature is less than 20℃, and 1.05 when the water temperature is greater than or equal to 20℃; T is the water temperature.

[0077] Pollutant removal rate: Effect: By calculating the pollutant removal rate, it can be directly judged whether the wastewater treatment process is running normally. Operators can find the weak links in the process by monitoring the removal rate of different pollutants in different process sections, and then adjust the operating parameters.

[0078] Calculation formula:

[0079] : Pollutant removal rate, unit is %; : Concentration of a certain pollutant in the influent, usually in mg / L (such as COD concentration, ammonia nitrogen concentration); : Concentration of the same pollutant in the effluent, the unit is the same as the influent.

[0080] Through the calculation of sludge load F / M, sludge age SRT, internal reflux ratio RN, external reflux ratio RW, gas-water ratio, specific oxygen consumption rate SOUR and pollutant removal rate by the sensing monitoring system, the real-time operation state of the biological system is determined, SOUR value data and its associated parameter data are obtained, and a water quality monitoring feature data set of SOUR value data and associated parameter data is constructed.

[0081] A built-in dynamic baseline model is constructed using the deep learning library TensorFlow; the built-in dynamic baseline model takes SOUR value as the main parameter, obtains historical SOUR value data and classifies SOUR value, determines the threshold range of SOUR value data at different levels and the threshold range of SOUR's associated parameter data at different levels; the built-in dynamic baseline model includes an input layer for inputting historical water quality monitoring feature data set, the historical water quality monitoring feature data set includes training set, validation set and test set, the model is trained and the model performance is evaluated.

[0082] The built-in dynamic baseline model is defined with SOUR value as the main parameter. The SOUR in the bioactivity dynamic baseline model is classified, and the span between each level is preset to be 5 mgO2 / (gMLVSS·h), the starting value is 0 mgO2 / (gMLVSS·h), and the maximum value is 25 mgO2 / (gMLVSS·h). Therefore, the levels are L1: 0-5 mgO2 / (gMLVSS·h), L2: 5 mgO2 / (gMLVSS·h)-10 mgO2 / (gMLVSS·h), L3: 10 mgO2 / (gMLVSS·h)-15 mgO2 / (gMLVSS·h), L4: 15 mgO2 / (gMLVSS·h)-20 mgO2 / (gMLVSS·h), and L5: 20 mgO2 / (gMLVSS·h)-25 mgO2 / (gMLVSS·h). If the value exceeds 25 mgO2 / (gMLVSS·h), it is counted in the range of 20 mgO2 / (gMLVSS·h)-25 mgO2 / (gMLVSS·h).

[0083] Subsequently, the training set in the data obtained by the perception detection system is input into the built-in dynamic baseline model, 365 consecutive SOUR values before the current value are taken, classified according to the classification method, and the average value of each level is calculated i and skewness , so as to determine the level and range of the next dynamic baseline model.

[0084] Level:

[0085] Skewness: sk i =

[0086] Wherein: : the kth SOUR sample value in the ith level; : the SOUR average value in the ith level; : the SOUR standard deviation in the ith level; n: sample number; Secondary classification:

[0087] Wherein : threshold boundary of the ith level.

[0088] The critical value of the range of each level of the dynamic baseline model is defined as the average value i , and the level i is determined by the skewness value (X i - i ).

[0089] If the average value i If the corresponding deviation degree is greater than 0, the range of the current level is i-1 - i If the average value 0=0mgO2 / (gMLVSS・h), if the average value i If the corresponding deviation degree is less than 0, the range of the current level is i - i+1 , i+1 If the average value i If the corresponding deviation degree is greater than 0, the average value i+1 If the corresponding deviation degree is less than 0, an additional level is added between the i-th level and the i+1-th level, and the range is i - i+1 .

[0090] The dynamic baseline model dynamically classifies the SOUR according to the distribution characteristics of the SOUR, and dynamically divides the number of levels in the SOUR standard dynamic baseline model into i levels.

[0091] For example, according to the above method, the SOUR is divided into L1 (≤2.98mgO2 / (gMLVSS・h)), L2 (2.98~6.95mgO2 / (gMLVSS・h)), L3 (6.95~12.75mgO2 / (gMLVSS・h)), L4 (12.75~16.79mgO2 / (gMLVSS・h)), and L5 (16.79~22.87mgO2 / (gMLVSS・h)), which are five levels; and the numerical values at both ends of each level are the threshold boundaries of the level.

[0092] The water quality monitoring feature data obtained by the perception detection system, such as SOUR, sludge age, internal reflux ratio, external reflux ratio, gas-water ratio, ammonia nitrogen removal rate, total nitrogen removal rate, and total phosphorus removal rate, is divided into a training set, a validation set, and a test set in sequence with time as a node, a model is trained, and the performance of the model is evaluated, for example, the training set includes related data continuously running for 6 months, the validation set includes data continuously running for 3 months, and the test set includes data continuously running for 3 months.

[0093] The mean square error loss function is used to measure the difference between the change trend of SOUR and the change trend of related data.

[0094] The built-in dynamic baseline model is trained using the training set, and the model parameters are updated through the back propagation algorithm to minimize the loss function; the performance of the built-in dynamic baseline model is determined and optimized by calculating the mean square error index using the validation set.

[0095] The model is fine-tuned according to the performance feedback of the validation set, and the model parameters are adjusted until the performance no longer improves significantly or the preset number of iterations is reached.

[0096] When the real-time SOUR is in the corresponding level, the relationship between the mode and the mean in the current level is determined by the same logic, and the standard threshold of each parameter is determined.

[0097] The sludge activity intelligent early warning system calculates the real-time pollutant removal rate through the calculation module in the sensing monitoring system according to the detected water quality, sludge parameters and pollutant effluent index requirements of the sewage treatment plant; the current level Li and its threshold range are determined through the pollutant removal rate and the built-in dynamic baseline model.

[0098] As shown in Figure 2 , the current SOUR value is processed by moving average and compared with the dynamic baseline model, and the event exceeding the current model threshold is found; moving average means that the newly collected data of the day is averaged with the existing data as the value for judgment, which is used to avoid the large fluctuations caused by the influent, and the influence of instantaneous change can be eliminated after averaging, such as using the average of all values within 3 days, that is, the average of 72 continuous data.

[0099] After SOUR abnormal alarm, trigger abnormality tracing engine, in built-in dynamic baseline model, according to SOUR value belongs to level Li, get relevant parameter Pi standard threshold range Pmin~Pmax.

[0100] At the same time, calculate the amount of parameter Pi deviating from the threshold, that is, the deviation (Pi-Pmean) / Pstd, where Pi is the real-time parameter value, Pmean is the mean value of the parameter in this level, and Pstd is the standard deviation.

[0101] List the parameter Pi with the largest deviation as the key inducement and sort it by deviation size, output Top3 key inducement information to expert library early warning system.

[0102] As shown in Figure 3 , the expert library early warning system includes a water quality-activity composite rule library, and the expert library early warning system outputs adjustment suggestions according to the rule library content corresponding to the Top3 key inducements.

[0103] The water quality-activity composite rule library includes 8 levels, which are: Rule R1: ammonia nitrogen removal rate decreases, effluent ammonia nitrogen increases and SOUR is normal, output "increase aeration quantity, increase sludge concentration, control SOUR to Li level, adjust parameter Pi value to Pmin~Pmax"; Rule R2: ammonia nitrogen removal rate decreases, effluent ammonia nitrogen increases and SOUR decreases, output "increase aeration quantity, increase sludge concentration, add carbon source, adjust parameter Pi value to Pmin~Pmax"; Rule R3: total phosphorus removal rate decreases, effluent total phosphorus increases, SOUR is normal and biological pool effluent orthophosphorus is normal, output "check secondary sedimentation tank sludge discharge, increase phosphorus removal agent (such as PAC) dosage, increase dosage point position"; Rule R4: total phosphorus removal rate decreases, effluent total phosphorus increases, SOUR is normal and biological pool effluent orthophosphorus increases, output "check anaerobic tank phosphorus release, ensure anaerobic tank DO <0.2 mg / L, check secondary sedimentation tank sludge discharge, control SOUR to Li level, increase phosphorus removal agent (such as PAC) dosage, increase dosage point position to 2-3, adjust parameter Pi value to Pmin~Pmax"; Rule R5: total phosphorus removal rate decreases, effluent total phosphorus increases, SOUR decreases and biological pool effluent orthophosphorus increases, output "check anaerobic tank phosphorus release, ensure anaerobic tank DO <0.2 mg / L, check secondary sedimentation tank sludge discharge, increase aeration quantity, add carbon source (dosage is calculated according to C / P=20:1), increase phosphorus removal agent (such as PAC) dosage, increase dosage point position to 2-3, adjust parameter Pi value to Pmin~Pmax"; Rule R6: total nitrogen removal rate decreases, effluent total nitrogen increases, SOUR is normal and biological pool nitrate increases, output "check internal reflux DO, control internal reflux DO <0.5 mg / L, control SOUR to Li level, add carbon source to anoxic tank (dosage is calculated according to C / N=3:1), adjust parameter Pi value to Pmin~Pmax"; Rule R7: total nitrogen removal rate decreases, effluent total nitrogen increases, SOUR decreases and biological pool nitrate increases, output "check internal reflux DO, control internal reflux DO <0.5 mg / L, add carbon source to anoxic tank (dosage is calculated according to C / N=5:1), add nitrifying bacteria, adjust parameter Pi value to Pmin~Pmax"; Rule R8: two or more of ammonia nitrogen, total phosphorus and total nitrogen indicators simultaneously appear removal rate decrease, effluent increase, SOUR sudden decrease, output "check whether there is toxic substance in the influent, adjust influent quantity to 80% of the design load, add carbon source (dosage is calculated according to actual demand), control SOUR, adjust parameter Pi value to Pmin~Pmax".

[0104] It should be noted that the water quality analyzer usually has the ability to communicate with external systems, such as through Modbus, Profibus, or other communication protocols, and real-time data can be obtained by reading the communication interface of the instrument. This method usually requires knowledge of the instrument's communication protocol and related commands; Dynamic water quality characteristics: Water quality characteristics may exhibit dynamic changes in time and space, and large data packet compression rate fluctuations may make it difficult to accurately reflect these changes in a timely manner. In water quality testing, real-time updated data is crucial for accurate monitoring of dynamic water bodies.

[0105] The present application realizes intelligent abnormal monitoring of biological sludge activity by monitoring the data analysis system established by the biological system. The system combines multiple water quality, sludge, and process parameter data and evaluates through comprehensive analysis index. Once the abnormal SOUR is detected, the system can quickly and accurately generate early warning and adjustment suggestions, enabling operators to promptly know the problem of biological activity operation state.

[0106] The present application provides differentiated treatment strategies for biological activity systems by establishing a perception monitoring system, a decision-making system, and an expert library early warning system. This differentiated treatment strategy enables operators to take appropriate measures in the case of abnormal alarm of different factors, thereby effectively dealing with various situations and improving the stability and operability of the system.

[0107] The above formulas are dimensionless values calculated, and the formulas are obtained by software simulation of a large amount of data to obtain a formula of the latest real situation. The preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0108] The above-described embodiments can be implemented in part or in whole in a computer program product, which can be executed on a computer. The computer program product comprises one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded and executed on a computer, the whole or part of the processes or functions described in the embodiments of the present application are produced. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable apparatus. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center through a wired or wireless (for example, infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (for example, a U disk, a floppy disk, a hard disk), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0109] It should be understood that the size of the sequence number of each process described above in various embodiments of the present application does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0110] Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0111] In the embodiments provided in the present application, it should be understood that the disclosed system and method can be implemented in other ways. For example, the above-described embodiments are merely illustrative.

[0112] The present application is not limited to the above best mode, and anyone can derive other various forms of products under the inspiration of the present application, but regardless of any changes in shape or structure, any technical solution with the same or similar technical solutions as the present application falls within the scope of the present application.

Claims

1. A sludge activity intelligent detection and early warning system, characterized in that: This includes a perception and monitoring system, a judgment and decision-making system, and an expert database early warning system; The aforementioned sensing and monitoring system is responsible for acquiring and calculating water quality, sediment quality, and process parameters, acquiring SOUR value data and associated parameter data related to the SOUR value data, and constructing a water quality monitoring feature dataset of SOUR value data and associated parameter data. The aforementioned decision-making system incorporates a dynamic baseline model and an anomaly tracing engine. A built-in dynamic baseline model is constructed using the TensorFlow deep learning library. This model uses the SOUR value as the primary parameter, acquires historical SOUR value data, and classifies the data into different gradations. It then determines the threshold ranges for SOUR values ​​at different gradations, as well as the threshold ranges for associated parameter data at the corresponding SOUR gradations. The built-in dynamic baseline model includes an input layer for inputting historical water quality monitoring feature datasets, which consist of training, validation, and test sets. The model is trained and its performance is evaluated. Based on changes in influent parameters and the required pollutant removal rate, the current SOUR value grading is determined. The current SOUR value data acquired by the sensing and monitoring system is compared with the threshold range of the current grading to determine if the current SOUR value data falls within the threshold range of the current grading. When the current SOUR value deviates from the threshold range in its current execution level, the anomaly tracing engine is activated to reverse-check and detect related parameter data in order to pinpoint the key cause. The aforementioned expert database early warning system is used to trigger alarms for abnormal sludge activity, simultaneously output key causes, and generate control suggestions based on the matching expert database.

2. The intelligent sludge activity detection and early warning system according to claim 1, characterized in that: The associated parameter data includes real-time monitoring data of water temperature, influent COD, influent ammonia nitrogen, sludge concentration MLSS, aerobic rate OUR, effluent ammonia nitrogen, phosphorus, nitrate nitrogen, influent flow rate, external return flow rate, internal return flow rate, aeration volume, and excess sludge discharge flow rate, as well as detection and calculation data of sludge organic matter f, sludge load F / M, sludge age SRT, internal return ratio RN, external return ratio RW, air-to-water ratio, specific aerobic rate SOUR, and pollutant removal rate.

3. The intelligent sludge activity detection and early warning system according to claim 2, characterized in that: Before the sensing and monitoring system can perform calculations, the acquired data needs to be preprocessed and then compressed and stored in the system's database to provide a data foundation for the system's calculations. Preprocessing includes noise reduction, outlier detection and correction, data interpolation and smoothing, calibration and standardization, unit conversion, and filtering.

4. The intelligent sludge activity detection and early warning system according to claim 1, characterized in that: The built-in dynamic baseline model first performs an initial classification of historical SOUR value data according to a certain span. Then, it calculates the skewness of SOUR values ​​within each initial classification to determine their distribution characteristics within that classification. Based on the distribution characteristics, it determines the classification threshold and performs a second classification to achieve the classification of SOUR value data. It collects and records the current SOUR value and automatically optimizes the threshold of its classification according to the classification method to form a built-in dynamic baseline model that adapts to different processing requirements. When the real-time SOUR is in the corresponding classification, the relevant parameter data uses the same logic to determine the relationship between the mode and the mean in that classification, thereby determining the standard threshold of each relevant parameter.

5. The intelligent sludge activity detection and early warning system according to claim 4, characterized in that: The historical SOUR data includes the average of the daily SOUR data over the past 365 days.

6. The intelligent sludge activity detection and early warning system according to claim 1, characterized in that, The aforementioned anomaly tracing engine performs the following steps based on the built-in dynamic baseline model: S1. Based on the changes in influent parameters and the required pollutant removal rate, obtain the classification Li to which the current SOUR value belongs in the built-in dynamic baseline model; S2. After processing the current SOUR value data, compare it with the built-in dynamic baseline model; S3. Based on Li, provide feedback on the threshold range Pmin~Pmax of each associated parameter Pi at the grade Li and the mean Pm at the grade. S4. Automatically calculate the amount by which each associated parameter data Pi deviates from its respective threshold range, i.e., the deviation |(Pi-Pm) / Pm|×100%, where Pi is the real-time parameter value and Pm is the average value of the graded parameter; S5. List the correlation parameter data Pi with the largest deviation as the key cause and sort them according to the degree of deviation. Output the Top 3 key cause information to the expert database early warning system.

7. The intelligent sludge activity detection and early warning system according to claim 6, characterized in that: In S2, the sensing and monitoring system acquires SOUR value data once per hour, and compares the current SOUR value data with the average of the previous 71 consecutive data as the new current SOUR value data with the built-in dynamic baseline model.

8. The intelligent sludge activity detection and early warning system according to claim 1, characterized in that, The expert database early warning system includes an expert knowledge graph containing water quality-activity composite rules; the water quality-activity composite rules include: Rule R1: When the ammonia nitrogen removal rate decreases, the effluent ammonia nitrogen increases, and the SOUR is normal, output "Increase the aeration rate, increase the sludge concentration, control the SOUR to the Li level, and adjust the Pi value to Pmin~Pmax". Rule R2: When the ammonia nitrogen removal rate decreases, the effluent ammonia nitrogen increases and the SOUR decreases, the output should be "Increase the aeration rate, increase the sludge concentration, add a carbon source, and adjust the Pi value to Pmin~Pmax". Rule R3: Total phosphorus removal rate decreases, total phosphorus in effluent increases, SOUR is normal and orthophosphorus in biological tank effluent is normal, output "Check sludge discharge from secondary sedimentation tank, increase phosphorus removal agent dosage, and increase dosage points"; Rule R4: Total phosphorus removal rate decreases, effluent total phosphorus increases, SOUR is normal and biological tank effluent orthophosphorus increases, output "Check the phosphorus release in the anaerobic tank, ensure DO in the anaerobic tank < 0.2 mg / L, check the sludge discharge in the secondary sedimentation tank, control SOUR to Li level, increase the dosage of phosphorus removal agent, increase the number of dosing points to 2-3, and adjust the Pi value to Pmin~Pmax". Rule R5: Total phosphorus removal rate decreases, effluent total phosphorus increases, SOUR decreases and biological tank effluent orthophosphorus increases. Output: "Check the phosphorus release in the anaerobic tank, ensure DO in the anaerobic tank < 0.2 mg / L, check the sludge discharge in the secondary sedimentation tank is normal, increase the aeration rate, add carbon source, increase the dosage of phosphorus removal agent, increase the number of addition points to 2-3, and adjust the Pi value to Pmin~Pmax". Rule R6: Total nitrogen removal rate decreases, effluent total nitrogen increases, SOUR is normal and nitrate nitrogen in biological tank increases, output "Check internal return DO, control internal return DO < 0.5 mg / L, control SOUR to Li level, add carbon source to anoxic tank, adjust Pi value to Pmin~Pmax"; Rule R7: Total nitrogen removal rate decreases, effluent total nitrogen increases, SOUR decreases and nitrate nitrogen in biological tank increases, output "Check internal return DO, control internal return DO < 0.5 mg / L, add carbon source to anoxic tank, add nitrifying bacteria, adjust Pi value to Pmin~Pmax"; Rule R8: If two or more of the three indicators of ammonia nitrogen, total phosphorus, and total nitrogen occur simultaneously, the removal rate decreases, the effluent increases, and the SOUR drops suddenly. The output should be: "Check if there are toxic substances in the influent, adjust the influent flow to 80% of the design load, add a carbon source, upgrade the SOUR control, and adjust the Pi value to Pmin~Pmax." 9. A warning method based on the intelligent sludge activity detection and early warning system according to any one of claims 1-8, characterized in that, Includes the following steps: a. Real-time acquisition of SOUR value data and related parameter data, with a sampling frequency of once per 60 minutes; among which, the organic content f of sludge is manually detected and updated once per month. b. Determine the current SOUR value classification based on changes in influent parameters and the required pollutant removal rate; c. After processing the current SOUR value, compare it with the baseline model. Perform a baseline comparison every hour to determine whether the current SOUR value data deviates from the threshold range of its current execution level. d. When the current SOUR value data offset exceeds the threshold range of its current execution level, initiate multi-parameter reverse tracing to investigate key causes; e. Match the expert database early warning system, issue an alarm, and output expert suggestions to the operation and maintenance personnel's terminal.

Images