Supervision system and method for sewage treatment process

Abstract

The invention relates to the technical field of sewage treatment, and discloses a supervision system and method for a sewage treatment process, and the system comprises a multi-parameter real-time monitoring module, a data fusion and analysis module, an automatic dosing control module and a feedback optimization module. The method is applied to the system. According to the supervision system and method for the sewage treatment process, dynamic and accurate supervision of the sewage treatment process is realized through the synergistic effect of the multi-parameter real-time monitoring module, the data fusion and analysis module, the automatic dosing control module and the closed-loop feedback optimization module; the problems of parameter simplification, control lag, high energy consumption and the like in the prior art are solved, the standard reaching rate of effluent is increased, and meanwhile, the agent cost and the equipment maintenance cost are reduced.

Description

Technical Field

[0001] This application relates to the technical field of sewage treatment, and specifically to a supervision system and method for the sewage treatment process. Background Art

[0002] Currently, the supervision systems for the sewage treatment process generally have the following technical problems:

[0003] Single real-time monitoring parameters: Existing systems usually only monitor a few water quality parameters such as COD and pH, and cannot comprehensively reflect the dynamic changes of sewage components (such as heavy metals and suspended solids) and process parameters such as flow rate and flow velocity, resulting in low dosing control accuracy.

[0004] Dosing control relies on fixed thresholds: Traditional systems use fixed thresholds (such as adding medicine at a fixed ratio when COD exceeds the threshold), and cannot adapt to sudden changes in water quality or flow fluctuations, resulting in waste of medicine or unqualified treatment effects.

[0005] Lack of dynamic optimization and closed-loop control: Existing algorithms (such as PID control) do not combine machine learning to dynamically adjust the dosing strategy, and do not form a closed-loop feedback mechanism, resulting in a lag in the system's response to abnormal conditions (such as an instantaneous increase in COD) and high energy consumption.

[0006] High energy consumption and equipment maintenance costs: The start-stop frequency of the dosing pump is fixed, and the start-stop strategy is not optimized according to the real-time working conditions, resulting in increased equipment wear and high maintenance costs. Moreover, the dosing amount of the medicine is not dynamically constrained, and there is a risk of secondary pollution.

[0007] These problems make it difficult for the existing technology to achieve precise control, energy conservation and consumption reduction, and long-term stable operation of sewage treatment. Summary of the Invention

[0008] The purpose of this application is to provide a supervision system and method for the sewage treatment process to solve the technical problems raised in the above background art.

[0009] To achieve the above purpose, this application discloses the following technical solutions:

[0010] In the first aspect, this application discloses a supervision system for the sewage treatment process, including:

[0011] A multi-parameter real-time monitoring module, including a water quality sensor group and a flow monitoring device. The water quality sensor group is used to collect data on chemical oxygen demand, biochemical oxygen demand, pH value, heavy metal concentration, and suspended solid concentration of sewage in real time; the flow monitoring device is used to collect data on the flow rate, flow velocity, and temperature of sewage in real time;

[0012] A data fusion and analysis module, which is used to receive the feedback data from the multi-parameter real-time monitoring module, and perform dynamic correlation analysis on water quality parameters through machine learning algorithms to generate chemical dosing control instructions;

[0013] An automated chemical dosing control module, which is used to add one or more of coagulants, flocculants or disinfectants to the sewage treatment system through a chemical dosing pump and a flow regulating valve according to the chemical dosing control instructions, and adjust the dosing amount in real time;

[0014] A feedback optimization module, which is used to perform iterative optimization on the chemical dosing control instructions through an adaptive algorithm based on the feedback data from the multi-parameter real-time monitoring module to form a closed-loop control.

[0015] Preferably, the machine learning algorithms include a spatio-temporal coupling regression model, a chemical dosing amount optimization model and a fuzzy PID control algorithm;

[0016] The spatio-temporal coupling regression model is used to extract features of the spatio-temporal distribution of chemical oxygen demand, biochemical oxygen demand and pH value in water quality parameters through a convolutional neural network, and combine a long short-term memory network to predict the water quality change trend in the future time, and its expression is:

[0017]

[0018] Among them, X t is the input data tensor at time step t, including the feature data extracted from the spatio-temporal distribution of chemical oxygen demand, biochemical oxygen demand and pH value in water quality parameters; f CNN is a convolutional neural network, which is used to extract local features in the spatial dimension; f LSTM is a long short-term memory network, which is used to model the dynamic changes in the time dimension; is the predicted water quality parameter value at the future moment t + Δt;

[0019] The chemical dosing amount optimization model is used to calculate the chemical dosing amount based on the multi-parameter non-linear relationship, and the calculation formula of the chemical dosing amount is:

[0020]

[0021] Among them, Q t is the chemical dosing amount at time t; α i , β i and γ i are time-varying coefficients obtained by training with historical data, and respectively correspond to the weights of chemical oxygen demand, biochemical oxygen demand and pH value; COD i , BOD i and pH iCorresponding to chemical oxygen demand, biochemical oxygen demand and pH value respectively, λ is the attenuation coefficient, where 0.05 ≤ λ ≤ 1, which is used to reflect the attenuation trend of water quality parameters over time; n is the number of water quality parameters participating in the calculation;

[0022] The proportional coefficient K of the fuzzy PID control algorithm p , the integral time T i and the differential time T d are adaptively adjusted according to the real-time flow rate Q 流量 and the flow velocity v; the specific adjustment formula is:

[0023]

[0024] T d = T d0 ·(1 + μ·ΔQ)

[0025] where, K p0 , T i0 and T d0 are the initial proportional coefficient, the initial integral time and the initial differential time respectively, Q 阈值 and v 阈值 are the preset flow rate threshold and flow velocity threshold, μ is the proportional coefficient gain factor, and ΔQ is the change amount of the chemical dosage.

[0026] [[ID=??]]Preferably, the spatio-temporal coupling regression model is also used to compensate for the time delay and noise of the sensor data through the Kalman filtering algorithm.

[0027] Preferably, the automatic chemical dosing control module further includes a multi-stage chemical dosing distributor and a chemical concentration correction unit;

[0028] The multi-stage chemical dosing distributor is used to achieve segmented dosing of the chemical through a pressure sensor and a solenoid valve, adjust the dosing position of the chemical according to the sewage depth, and optimize the flocculation effect through dynamic mixing ratio adjustment. The dynamic mixing ratio formula is: P r = 1:(1 + k·Conc 悬浮物 ·sin(ω·t)), where, P r is the calculated mixing ratio, k is the empirical coefficient, Conc 悬浮物 is the suspended solid concentration, ω is the adjustment frequency, and ω = 2π·0.5Hz;

[0029] The chemical concentration correction unit is used to correct the dosing ratio of the chemical in real time according to the heavy metal concentration data through an ion-selective electrode. The correction formula is: where, C 校正 is the corrected chemical concentration, C 初始 is the initial chemical concentration, Conc 重金属 Note: There seems to be an incomplete formula in ID=37. The translation is done as accurately as possible based on the provided text.is the heavy metal concentration data, Conc 重金属阈值 is the preset heavy metal concentration threshold.

[0030] Preferably, the feedback optimization module includes an abnormal condition identification unit and an energy consumption optimization sub-module. The abnormal condition identification unit detects mutations or abnormal fluctuations in water quality parameters through cluster analysis. When it detects that the instantaneous increase in COD exceeds 20%, it triggers an emergency dosing mode. The energy consumption optimization sub-module is used to optimize the start-stop frequency of the dosing pump.

[0031] Preferably, in the emergency dosing mode, the dosing amount is adjusted through a dosing amount adjustment formula. The specific dosing amount adjustment formula is:

[0032]

[0033] where Q 当前 is the dosing amount before triggering the emergency dosing mode, ΔParm is the change amount of any one of chemical oxygen demand, biochemical oxygen demand, and pH value in the water quality parameters, Parm 阈值 is the trigger threshold of any one of chemical oxygen demand, biochemical oxygen demand, and pH value in the preset water quality parameters, and σ is the non-linear gain coefficient.

[0034] Preferably, the objective function of the energy consumption optimization sub-module is: where E(t) is the energy consumption at time t, τ is the penalty coefficient for balancing energy consumption and dosing accuracy, W(t) is the actual dosing amount at time t, and W 目标 is the preset dosing amount.

[0035] Preferably, the energy consumption optimization sub-module is configured with a dynamic constraint algorithm. The constraint conditions include dynamic limits on the start-stop frequency of the dosing pump and dynamic constraints on the dosing amount;

[0036] The dynamic limit on the start-stop frequency of the dosing pump is specifically: where ΔS t is the start-stop times of the dosing pump at time t, N max (t) is the dynamic start-stop times threshold, Q 流量 (t) and v(t) are the real-time flow rate and flow velocity, Q 阈值 and v 阈值 are the preset flow rate threshold and flow velocity threshold, and N 基准 is the preset reference start-stop times;

[0037] The dynamic constraint on the dosing amount is specifically: W min (t) ≤ W(t) ≤ W max (t), Wmin (t) is the minimum chemical dosage, W max (t) is the maximum chemical dosage.

[0038] Preferably, the water quality sensor group includes an electrochemical sensor array, a spectral analysis unit, and a multi-sensor data fusion unit. The electrochemical sensor array is used to monitor pH, dissolved oxygen, and redox potential in real time; the spectral analysis unit is used to perform on-line qualitative and quantitative analysis of organic pollutants through Raman spectroscopy or fluorescence spectroscopy techniques; the multi-sensor data fusion unit is used to fuse sensor data through the Bayesian filtering algorithm to eliminate noise interference.

[0039] In a second aspect, the present application discloses a supervision method for the sewage treatment process, which is applied to the supervision system for the sewage treatment process as described above. The method includes the following steps:

[0040] Collect chemical oxygen demand, biochemical oxygen demand, pH value, heavy metal concentration, suspended solid concentration, flow rate, flow velocity, and temperature data of the sewage through a multi-parameter sensor group and a flow monitoring device;

[0041] According to the collected data, perform dynamic correlation analysis on water quality parameters through a machine learning algorithm to generate a chemical dosing control instruction;

[0042] According to the chemical dosing control instruction, add one or more of coagulants, flocculants, or disinfectants to the sewage treatment system through a chemical dosing pump and a flow regulating valve, and adjust the dosing amount in real time;

[0043] Based on real-time feedback data, perform iterative optimization on the chemical dosing control instruction through an adaptive algorithm to form a closed-loop control.

[0044] Beneficial effects: The supervision system and method for the sewage treatment process of the present application achieve dynamic and precise supervision of the sewage treatment process through the coordinated action of a multi-parameter real-time monitoring module, a data fusion and analysis module, an automatic chemical dosing control module, and a closed-loop feedback optimization module, solve the problems of single parameters, control lag, high energy consumption, etc. in the prior art, improve the effluent compliance rate to ≥95%, and reduce the chemical agent cost and equipment maintenance cost at the same time. Description of the Drawings

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present application. For those skilled in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0046] Figure 1It is a structural block diagram of a supervision system for sewage treatment process provided by an embodiment of the present application. Detailed implementation manners

[0047] The technical solutions in the embodiments of the present application will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0048] In this article, the term "including" is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such a process, method, article or device. Without more limitations, the elements defined by the statement "including..." do not exclude the existence of other identical elements in the process, method, article or device including the said elements.

[0049] In the first aspect, this embodiment discloses a supervision system for sewage treatment process as Figure 1 shown, aiming to specifically solve the technical pain points existing in the prior art through technical solutions such as multi-parameter real-time monitoring, dynamic machine learning model, and closed-loop feedback optimization.

[0050] This system includes a multi-parameter real-time monitoring module, a data fusion and analysis module, an automatic chemical dosing control module, and a feedback optimization module. The multi-parameter real-time monitoring module, the data fusion and analysis module, and the automatic chemical dosing control module are sequentially communicatively connected, and the feedback optimization module is communicatively connected to both the multi-parameter real-time monitoring module and the data fusion and analysis module.

[0051] The multi-parameter real-time monitoring module includes a water quality sensor group and a flow monitoring device. The water quality sensor group is used to collect data on chemical oxygen demand (COD), biochemical oxygen demand (BOD), pH value, heavy metal concentration, and suspended solid concentration of sewage in real time; the flow monitoring device is used to collect data on the flow rate, flow velocity, and temperature of sewage in real time. These data are transmitted to the data fusion and analysis module through wireless transmission or wired connection.

[0052] The data fusion and analysis module is used to receive the feedback data from the multi-parameter real-time monitoring module and perform dynamic correlation analysis on water quality parameters through machine learning algorithms to generate chemical dosing control instructions.

[0053] The automatic chemical dosing control module is used to add one or more of coagulants, flocculants, or disinfectants to the sewage treatment system according to the chemical dosing control instructions through a chemical dosing pump and a flow regulating valve, and adjust the dosing amount in real time.

[0054] A feedback optimization module, which is used to iteratively optimize the chemical dosing control instruction through an adaptive algorithm based on the feedback data of the multi-parameter real-time monitoring module to form a closed-loop control.

[0055] Specifically, the machine learning algorithm includes a spatio-temporal coupling regression model, a chemical dosing amount optimization model, and a fuzzy PID control algorithm.

[0056] The spatio-temporal coupling regression model is used to extract features of the spatio-temporal distribution of chemical oxygen demand, biochemical oxygen demand, and pH value in water quality parameters through a convolutional neural network, and combine a long short-term memory network to predict the water quality change trend in the future time. For example, assuming we have a historical database containing water quality data for the past 24 hours, we can use this data to train the model to predict the water quality change in the next 10 minutes. Simply put, the model combines the spatial distribution characteristics of water quality parameters (such as the spatial correlation of COD, BOD, pH, and turbidity at different monitoring points) and time series characteristics (such as the change law of water quality parameters over time) through joint modeling of spatio-temporal features to predict the water quality state at a future moment, and its expression is:

[0057]

[0058] Among them, X t is the input data tensor at time step t, including the feature data extracted from the spatio-temporal distribution of chemical oxygen demand, biochemical oxygen demand, and pH value in water quality parameters; f CNN is the convolutional neural network, which is used to extract local features in the spatial dimension; f LSTM is the long short-term memory network, which is used to model the dynamic changes in the time dimension; is the predicted water quality parameter value at the future moment t + Δt. The spatio-temporal coupling regression model combines CNN-LSTM to predict the water quality change trend in the next 10 minutes, and the chemical dosing error is reduced to ≤3%, which is significantly better than the traditional PID control (error ≥10%).

[0059] In this embodiment, the core principle of spatio-temporal coupling modeling is as follows: (1) Spatial dimension modeling: CNN automatically captures the spatial correlation of water quality parameters through local receptive fields (such as whether the COD concentration changes at adjacent monitoring points are synchronous), solving the problem of manually defining spatial weights in traditional methods; (2) Temporal dimension modeling: LSTM dynamically retains or discards historical information through gating mechanisms (input gate, forget gate), effectively dealing with long-term dependencies in time series (such as the delayed increase in BOD concentration after rainfall); (3) Joint modeling: Combining spatial features (output of CNN) with temporal dynamics (output of LSTM) to form spatio-temporal joint features, avoiding the limitations of single-dimension modeling (such as ignoring spatial heterogeneity by only using time series). Compared with traditional technologies, traditional methods rely on manually defining spatial weights, while this model automatically learns spatial patterns through CNN; LSTM solves the problem that traditional regression models cannot capture multi-time scale dependencies; and, directly from raw spatio-temporal data to prediction results, without complex preprocessing. Using the spatio-temporal coupling regression model of this application, although the calculation time is slightly higher, the improvement in its prediction accuracy (such as a 49% reduction in COD prediction error) significantly improves the real-time performance and reliability of water quality management.

[0060] The chemical dosage optimization model is used to calculate the chemical dosage based on the multi-parameter non-linear relationship, and the calculation formula for the chemical dosage is:

[0061]

[0062] where Q t is the chemical dosage at time t; α i , β i and γ i are time-varying coefficients obtained by training with historical data (such as ridge regression), and respectively correspond to the weights of chemical oxygen demand, biochemical oxygen demand, and pH value; COD i , BOD i and pH i respectively correspond to chemical oxygen demand, biochemical oxygen demand, and pH value, λ is the attenuation coefficient 0.05 ≤ λ ≤ 1, which is used to reflect the attenuation trend of water quality parameters over time and is determined by cross-validation; n is the number of water quality parameters participating in the calculation. Compared with the traditional fixed-threshold method, this non-linear chemical dosage optimization model can more accurately reflect the water quality change trend, with the error rate reduced by more than 30%, thus improving the chemical dosing efficiency and effect. Capturing the non-linear relationship of water quality parameters through non-linear combinations (logarithm, square root, etc.), such as the sensitivity when the pH value deviates from neutral (pH i -7). e -λ·t considers the time-variability of water quality parameters (such as the attenuation of pollutant concentration over time), avoiding the lag effect.

[0063] The proportional coefficient K of the fuzzy PID control algorithm p, integral time T i and differential time T d are adaptively adjusted according to the real-time flow rate Q 流量 and flow velocity v; the specific adjustment formula is:

[0064]

[0065] T d = T d0 ·(1 + μ·ΔQ)

[0066] where, K p0 , T i0 and T d0 are the initial proportionality coefficient, initial integral time and initial differential time respectively, Q 阈值 and v 阈值 are the preset flow threshold and flow velocity threshold, μ is the proportionality coefficient gain factor, and ΔQ is the change in the chemical dosage. For example, when the flow rate increases, the proportionality coefficient K p will also increase accordingly to respond faster to water quality changes and respond quickly; when the flow rate is stable, the integral time T i is extended to reduce integral accumulation. This adaptive adjustment mechanism can better adapt to complex and changing working conditions than traditional PID control. When the flow rate fluctuates by ±20%, the control error is reduced to within 5%.

[0067] Furthermore, in order to ensure the accuracy of spatio-temporal feature extraction, the spatio-temporal coupling regression model is also used to compensate for the time delay and noise of sensor data through the Kalman filtering algorithm. This step is crucial for ensuring the reliability of subsequent analysis results.

[0068] Secondly, in order to ensure the reliability of the processing process, in this embodiment, the spatio-temporal coupling regression model also includes a dynamic weight allocation mechanism, which dynamically adjusts the weights of each parameter in the regression model according to the real-time change rate of water quality parameters. The weight calculation formula is: where, θ i is the weight of the i-th parameter, θ i ∈(α i , β i , γ i ); ΔX i is the current change rate of the i-th parameter; X i is the current value of the i-th parameter, X 目标 is the preset target value; η is the sensitivity coefficient, which is used to adjust the sensitivity of the weight to the deviation of the parameter from the target value, and η = 0.3 is set. The dynamic weight allocation mechanism adjusts the weight according to the parameter change rate, so that the response time of the system to a sudden change in COD (such as an instantaneous increase of 20%) ≤ 2 seconds, avoiding processing lag.

[0069] As a preferred implementation manner of this embodiment, the automatic chemical dosing control module further includes a multi-stage chemical dosing distributor and a chemical concentration correction unit.

[0070] The multi-stage chemical dosing distributor is used to achieve segmented dosing of chemicals through a pressure sensor and a solenoid valve, adjust the dosing position of the chemicals according to the sewage depth, and optimize the flocculation effect through dynamic mixing ratio adjustment, so as to achieve precise segmented dosing of chemicals and multi-chemical synergistic control, improve the flocculation effect and reduce chemical waste. The technical principle is: stratify according to the depth of the aeration tank (such as the surface layer, middle layer, bottom layer), monitor the sewage flow rate and pressure in different areas through a pressure sensor, and the solenoid valve controls the chemical dosing position to ensure uniform dispersion of the chemicals. The flow control accuracy of the segmented dosing is ±0.5%, and this accuracy parameter is achieved through a high-precision solenoid valve (such as a stepping motor-driven valve) and closed-loop PID control. The segmented dosing logic is: divide the area according to the depth of the aeration tank (such as 0-2m, 2-4m, 4-6m); the pressure sensor real-time feedbacks the pressure of each layer, and dynamically adjusts the opening of the solenoid valve to ensure uniform distribution of the chemicals in the target area.

[0071] When treating wastewater containing multiple pollutants, the best treatment effect can be achieved by adjusting the ratio of different chemicals. The dynamic mixing ratio formula is: P r = 1:(1 + k·Conc 悬浮物 ·sin(ω·t)), where P r is the calculated mixing ratio, k is an empirical coefficient, Conc 悬浮物 is the suspended solid concentration, ω is the adjustment frequency, and ω = 2π·0.5Hz. By periodically adjusting the mixing ratio through the sine function, high / low concentration zones are formed alternately during the flocculation process, enhancing the particle collision efficiency. The implementation steps are: the sensor real-time monitors the turbidity and pH value; according to the current water quality data, dynamically select the ω value (such as increasing the frequency when the water quality fluctuates greatly); control the opening of the solenoid valve through the PLC and distribute the chemicals in proportion.

[0072] Based on the design of the above multi-stage chemical dosing distributor, the ineffective loss of chemicals on the surface layer is reduced, the flocculation efficiency of the bottom layer is improved, and the chemical utilization rate is increased by 15%-20%. And through periodic ratio adjustment, the flocculation particles form denser flocs, and the suspended solid removal rate is increased to more than 95%.

[0073] The chemical agent concentration correction unit is used to correct the dosing ratio of the chemical agent in real time according to the heavy metal concentration data through an ion-selective electrode, avoiding over-dosing or insufficient treatment. For example, if an increase in heavy metal concentration is detected, the dosing amount of the chemical agent is appropriately increased to ensure the treatment effect. The ion-selective electrode (ISE) monitors the concentration of heavy metal ions (such as lead and cadmium) in real time and adjusts the chemical agent concentration through a non-linear correction formula to inhibit the antagonistic effect between heavy metals and flocculants. The correction formula adopted by the chemical agent concentration correction unit is: where 0.5 is the linear gain coefficient to control the correction amplitude; C 校正 is the corrected chemical agent concentration; C 初始 is the initial chemical agent concentration; Conc 重金属 is the heavy metal concentration data; Conc 重金属阈值 is the preset heavy metal concentration threshold, set to 10 mg / L, which is a preset safety concentration threshold. When it is exceeded, the dosing of the chemical agent needs to be strengthened. The implementation steps are as follows: The ISE detects the heavy metal concentration every 5 seconds; the data is input into the correction formula to calculate C 校正 ; the frequency of the chemical agent dosing pump is adjusted to make the actual concentration approach C 校正 .

[0074] Through the design of the chemical agent concentration correction unit, the heavy metal removal rate is increased to 99%, while reducing the over-dosing of the chemical agent (saving about 10%-15% in cost). Moreover, the competitive adsorption between heavy metals and flocculants is inhibited, ensuring stable flocculation effect.

[0075] As a preferred implementation manner of this embodiment, the feedback optimization module includes an abnormal condition identification unit and an energy consumption optimization sub-module. The abnormal condition identification unit detects mutations or abnormal fluctuations in water quality parameters through cluster analysis. When it detects that the instantaneous increase in COD exceeds 20%, it triggers an emergency chemical agent dosing mode. The energy consumption optimization sub-module is used to optimize the start-stop frequency of the chemical agent dosing pump.

[0076] In order to quickly respond to water quality mutations and avoid treatment failure, in the emergency chemical agent dosing mode, the dosing amount of the chemical agent is adjusted through the dosing amount adjustment formula. During daily monitoring, the water quality data (such as COD, etc.) is analyzed in real time through the DBSCAN clustering algorithm to detect abnormal fluctuations. When the instantaneous increase in COD exceeds the threshold, emergency chemical agent dosing is started. The specific dosing amount adjustment formula is:

[0077]

[0078] where Q 当前 is the dosing amount before triggering the emergency chemical agent dosing mode, ΔParm is the change amount of any one of chemical oxygen demand, biochemical oxygen demand, and pH value in the water quality parameters, Parm 阈值is the trigger threshold for any one of the chemical oxygen demand, biochemical oxygen demand, and pH value in the preset water quality parameters, such as COD 阈值 = 100 mg / L, σ is the non-linear gain coefficient, which controls the response slope (the larger σ is, the more sensitive the response is), and σ = 0.75 is taken. The implementation steps are as follows: The DBSCAN algorithm clusters the real-time COD data to identify mutation points; when ΔCOD exceeds the 20% threshold, calculate Q 加急 ; Trigger the emergency mode, and the dosing pump adds the reagent (such as flocculant + coagulant aid) at the maximum flow rate.

[0079] Through the design of the emergency dosing mode, rapid response can be achieved. For example, when the COD suddenly increases by 30%, the concentration can be restored below the threshold within 5 minutes; and, through non-linear gain, over-dosing is avoided, and the energy consumption in the emergency mode only increases by 15%.

[0080] It is known that each time the dosing pump is started, the motor needs to overcome static friction and acceleration inertia, resulting in a significant increase in instantaneous energy consumption (especially in high-power pumps). Frequent start-stop will accumulate additional energy consumption. Reducing the start-stop times can reduce the start-up energy consumption, but may cause fluctuations in the dosing amount (due to continuous operation or long-term shutdown of the pump). Correspondingly, the pump can adjust the dosing amount more frequently (such as quickly responding to water quality changes), making the actual dosing amount at time t closer to the preset dosing amount, but may cause an increase in energy consumption due to frequent start-stop. And the pump runs more smoothly, but may cause an increase in the deviation between the actual dosing amount and the preset dosing amount at time t due to the inability to adjust the dosing amount in time (such as insufficient dosing when the influent turbidity suddenly increases). Therefore, in this embodiment, the energy consumption optimization sub-module optimizes the start-stop frequency of the dosing pump through the dynamic programming algorithm to achieve a balance between energy consumption and dosing accuracy. Its technical principle is: Establish a weighted objective function of energy consumption and dosing accuracy, combine dynamic constraint conditions (start-stop frequency limit, dosing range), and use the dynamic programming algorithm to solve the optimal start-stop strategy. The objective function of the energy consumption optimization sub-module is: where E(t) is the energy consumption (kW·h) at time t, including start-up energy consumption (such as 0.5 kW·h / time) and running energy consumption (such as 0.1 kW·h / min), τ is the penalty coefficient used to balance energy consumption and dosing accuracy, balancing energy consumption (low priority) and accuracy (high priority), and τ = 15 is taken, W(t) is the actual dosing amount at time t, and W 目标 is the preset dosing amount.

[0081] Corresponding to the objective function, the energy consumption optimization sub-module is configured with a dynamic constraint algorithm, and the constraint conditions include dynamic limits on the start-stop frequency of the dosing pump and dynamic constraints on the dosing amount of the reagent.

[0082] The dynamic limit on the start-stop frequency of the dosing pump is specifically: where, ΔS t is the start / stop times of the chemical dosing pump at time t, N max (t) is the dynamic start / stop times threshold. When the flow rate or velocity increases, N max (t) increases linearly, allowing more frequent start / stop to cope with load changes. Q 流量 (t) and v(t) are the real-time flow rate and velocity, Q 阈值 and v 阈值 are the preset flow rate threshold and velocity threshold, N 基准 is the preset reference start / stop times, taking N 基准 = 5 times / hour.

[0083] Secondly, the dosing amount constraint is:

[0084] W min (t) ≤ W(t) ≤ W max (t)

[0085] Reduce chemical dosing at low COD

[0086] Increase the upper limit at high COD to avoid insufficient treatment

[0087] where, Conc COD is the concentration value of COD, W min (t) is the minimum dosing amount, W max (t) is the maximum dosing amount, W 基准 is the reference dosing amount, COD 阈值 is the preset COD concentration threshold.

[0088] The implementation steps are as follows: The dynamic programming algorithm takes 1 minute as the time step to predict the energy consumption and error in the next T = 60 minutes; select the start / stop decision at each time point to meet the constraint conditions and minimize the objective function; output the optimal start / stop sequence to control the pump to execute.

[0089] Through the design of the above dynamic constraint algorithm, by adjusting the upper limit of the start / stop times of the chemical dosing pump (N max (t)) according to the real-time flow rate and velocity, compared with the fixed threshold (such as 5 times / hour), the energy consumption is reduced by 15% - 20%. The dynamic constraint of the chemical dosing amount (W min (t) and W max (t)) avoids over-dosing, and the chemical cost is reduced by 20% - 25%.

[0090] In this embodiment, the water quality sensor group includes an electrochemical sensor array, a spectral analysis unit, and a multi-sensor data fusion unit. The electrochemical sensor array is used to monitor pH, dissolved oxygen, and redox potential in real time. The sampling frequency of the electrochemical sensor is 10 Hz, and the resolution is 0.01 pH / 0.1 mg / L / mV. The spectral analysis unit is used to perform on-line qualitative and quantitative analysis of organic pollutants through Raman spectroscopy or fluorescence spectroscopy technology, identify the types of organic pollutants, assist in the selection of chemical agent types (such as selective flocculants), and the detection limit is ≤ 3 mg / L. It can simultaneously analyze TOC, benzene series, polycyclic aromatic hydrocarbons, etc. The multi-sensor data fusion unit is used to fuse sensor data through the Bayesian filtering algorithm to eliminate noise interference. Any one of the existing technologies can be adopted for the Bayesian filtering algorithm. By eliminating sensor drift and environmental noise, the data confidence level is increased by 30%. The implementation steps are as follows: The sensor data is uploaded to the central controller every second; the Bayesian filter calculates weights and fuses the data; the fusion result drives the dosing control and abnormal alarm.

[0091] Through the configuration of the water quality sensor group, the following are achieved: the pH error ≤ ±0.1, and the heavy metal detection limit reaches the ppb level; multi-component identification: simultaneously quantitatively analyze 5 - 10 kinds of organic pollutants; anti-interference ability: the data availability is increased by 50% in a noisy environment.

[0092] In summary, this application solves the core problems of single parameters, control lag, high energy consumption, and high maintenance costs in the existing sewage treatment supervision system by integrating innovative technologies such as multi-parameter monitoring, dynamic machine learning models, and closed-loop feedback optimization, and realizes precise, energy-saving, and stable sewage treatment process control, with significant technological progressiveness and practicality.

[0093] In the second aspect, this embodiment provides a supervision method for the sewage treatment process, which is applied to the supervision system for the sewage treatment process as described above. The method includes the following steps:

[0094] Collect chemical oxygen demand, biochemical oxygen demand, pH value, heavy metal concentration, suspended solid concentration, flow rate, flow velocity, and temperature data of sewage through a multi-parameter sensor group and a flow monitoring device;

[0095] According to the collected data, perform dynamic correlation analysis on water quality parameters through a machine learning algorithm to generate a dosing control instruction;

[0096] According to the dosing control instruction, add one or more of coagulants, flocculants, or disinfectants to the sewage treatment system through a dosing pump and a flow regulating valve, and adjust the dosing amount in real time;

[0097] Based on the real-time feedback data, perform iterative optimization on the dosing control instruction through an adaptive algorithm to form a closed-loop control.

[0098] It should be noted that the supervision method for the sewage treatment process in this embodiment corresponds to the aforementioned supervision system for the sewage treatment process. Therefore, for the parts not disclosed in the supervision method for the sewage treatment process in this embodiment (specific technical means and corresponding technical effects), reference can be made to the description in the aforementioned supervision system for the sewage treatment process, and details will not be repeated herein.

[0099] In the embodiments provided in this application, it should be understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, code, or any appropriate combination thereof. For hardware implementation, the processor can be implemented in one or more of the following units: application specific integrated circuit (ASIC), digital signal processor (DSP), digital signal processing device (DSPD), programmable logic device (PLD), field programmable gate array (FPGA), processor, controller, microcontroller, microprocessor, or other electronic units designed to implement the functions described herein, or a combination thereof. For software implementation, part or all of the processes of the embodiments can be completed by instructing relevant hardware through a computer program. When implemented, the above program can be stored in a computer-readable storage medium or transmitted as one or more instructions or codes on a computer-readable storage medium. The computer-readable storage medium includes computer storage media and communication media, where the communication media includes any medium that facilitates the transmission of a computer program from one place to another. The storage media can be any available medium that can be accessed by a computer. The computer-readable storage medium can include, but is not limited to, RAM, ROM, EEPROM, CD-ROM, or other optical disc storage, magnetic disk storage media, or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer.

[0100] Finally, it should be noted that the above are only the preferred embodiments of this application and are not used to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of this application shall be included in the protection scope of this application.

Claims

1. A supervision system for the sewage treatment process, characterized in that, Comprising: A multi-parameter real-time monitoring module, including a water quality sensor group and a flow monitoring device. The water quality sensor group is used to collect data on chemical oxygen demand, biochemical oxygen demand, pH value, heavy metal concentration, and suspended solid concentration of sewage in real time. The flow monitoring device is used to collect data on the flow rate, flow velocity, and temperature of sewage in real time. A data fusion and analysis module, which is used to receive the feedback data of the multi-parameter real-time monitoring module and perform dynamic correlation analysis on water quality parameters through machine learning algorithms to generate chemical dosing control instructions. An automated chemical dosing control module, which is used to add one or more of coagulants, flocculants, or disinfectants to the sewage treatment system through a chemical dosing pump and a flow regulating valve according to the chemical dosing control instructions, and adjust the dosing amount in real time. A feedback optimization module, which is used to iteratively optimize the chemical dosing control instructions through an adaptive algorithm based on the feedback data of the multi-parameter real-time monitoring module to form a closed-loop control.

2. The supervision system for sewage treatment process according to claim 1, characterized in that, The machine learning algorithms include a spatio-temporal coupling regression model, a chemical dosing amount optimization model, and a fuzzy PID control algorithm. The spatio-temporal coupling regression model is used to extract features of the spatio-temporal distribution of chemical oxygen demand, biochemical oxygen demand, and pH value in water quality parameters through a convolutional neural network, and predict the water quality change trend in the future time in combination with a long short-term memory network. Its expression is: Among them, X t is the input data tensor at time step t, including the feature data extracted from the spatio-temporal distributions of chemical oxygen demand, biochemical oxygen demand, and pH value among the water quality parameters; f CNN is a convolutional neural network for extracting local features in the spatial dimension; f LSTM is a long short-term memory network for modeling the dynamic changes in the time dimension; is the predicted water quality parameter value at the future time t + Δt; The chemical dosing amount optimization model is used to calculate the chemical dosing amount based on the multi-parameter non-linear relationship. The calculation formula for the chemical dosing amount is: Among them, Q t is the chemical dosing amount at time t; α i , β i and γ i are time-varying coefficients obtained by training with historical data, and respectively correspond to the weights of chemical oxygen demand, biochemical oxygen demand and pH value; COD i , BOD i and pH i respectively correspond to chemical oxygen demand, biochemical oxygen demand and pH value, λ is the decay coefficient 0.05 ≤ λ ≤ 1, which is used to reflect the decay trend of water quality parameters over time; n is the number of water quality parameters participating in the calculation; The proportional coefficient K of the fuzzy PID control algorithm p , the integral time T i and the differential time T d are adaptively adjusted according to the real-time flow rate Q 流量 and the flow velocity v; the specific adjustment formula is as follows: T d = T d0 ·(1 + μ·ΔQ) Among them, K p0 , T i0 and T d0 are the initial proportionality coefficient, the initial integral time, and the initial derivative time respectively. Q 阈值 and v 阈值 are the preset flow threshold and flow velocity threshold. μ is the proportionality coefficient gain factor, and ΔQ is the change in the chemical addition amount.

3. The supervision system for the sewage treatment process according to claim 2, wherein, The spatio-temporal coupling regression model is also used to compensate for the time delay and noise of sensor data through a Kalman filter algorithm.

4. The supervision system for the sewage treatment process according to claim 1, characterized in that, The automated chemical dosing control module further includes a multi-stage chemical dosing distributor and a chemical agent concentration correction unit. The multi-stage chemical dosing dispenser is used to achieve segmented dosing of chemicals through a pressure sensor and a solenoid valve, adjust the dosing position of the chemicals according to the sewage depth, and optimize the flocculation effect through dynamic mixing ratio adjustment. The dynamic mixing ratio formula is: P r = 1:(1 + k·Conc 悬浮物 ·sin(ω·t)), where P r is the calculated mixing ratio, k is an empirical coefficient, Conc 悬浮物 is the suspended solid concentration, ω is the adjustment frequency, and ω = 2π·0.5 Hz; The chemical agent concentration correction unit is used to correct the dosing ratio of the chemical agent in real time through an ion-selective electrode according to the heavy metal concentration data. The correction formula is: Among them, C 校正 is the corrected drug concentration, and C 初始 is the initial drug concentration. Conc 重金属 is the heavy metal concentration data, and Conc 重金属阈值 is the preset heavy metal concentration threshold.

5. The supervision system for the sewage treatment process according to claim 1, characterized in that, The feedback optimization module includes an abnormal condition identification unit and an energy consumption optimization sub-module. The abnormal condition identification unit detects mutations or abnormal fluctuations in water quality parameters through cluster analysis. When it detects that the instantaneous increase in COD exceeds 20%, it triggers an emergency chemical dosing mode. The energy consumption optimization sub-module is used to optimize the start-stop frequency of the chemical dosing pump.

6. The supervision system for the sewage treatment process according to claim 5, wherein, In the emergency chemical dosing mode, the chemical dosing amount is adjusted through a chemical dosing amount adjustment formula. The specific chemical dosing amount adjustment formula is: Among them, Q 当前 is the chemical dosage before triggering the emergency chemical addition mode, ΔParm is the change amount of any one of the chemical oxygen demand, biochemical oxygen demand, and pH value in the water quality parameters, and Parm 阈值 is the trigger threshold of any one of the chemical oxygen demand, biochemical oxygen demand, and pH value in the preset water quality parameters, and σ is the nonlinear gain coefficient.

7. The supervision system for the sewage treatment process according to claim 5, characterized in that, The objective function of the energy consumption optimization sub-module is: Among them, E(t) is the energy consumption at time t, τ is the penalty coefficient used to balance the energy consumption and the chemical dosing accuracy, W(t) is the actual chemical dosing amount at time t, and W 目标 is the preset chemical dosing amount.

8. The supervision system for sewage treatment process according to claim 7, characterized in that, The energy consumption optimization sub-module is configured with a dynamic constraint algorithm. The constraint conditions include dynamic limits on the start-stop frequency of the chemical dosing pump and dynamic constraints on the chemical agent dosing amount. The dynamic limit on the start-stop frequency of the chemical dosing pump is specifically: Among them, ΔS t is the start-stop times of the chemical dosing pump at time t, N max (t) is the dynamic start-stop times threshold, Q 流量 (t) and v(t) are the real-time flow rate and flow velocity, Q 阈值 and v 阈值 are the preset flow rate threshold and flow velocity threshold, N 基准 is the preset reference start-stop times; The dynamic constraint of the chemical dosage is specifically: W min (t) ≤ W(t) ≤ W max (t), where W min (t) is the minimum dosage, and W max (t) is the maximum dosage.

9. The supervision system for sewage treatment process according to claim 1, characterized in that, The water quality sensor group includes an electrochemical sensor array, a spectral analysis unit, and a multi-sensor data fusion unit. The electrochemical sensor array is used to monitor pH, dissolved oxygen, and oxidation-reduction potential in real time. The spectral analysis unit is used to perform on-line qualitative and quantitative analysis of organic pollutants through Raman spectroscopy or fluorescence spectroscopy technology. The multi-sensor data fusion unit is used to fuse sensor data through a Bayesian filtering algorithm to eliminate noise interference.

10. A supervision method for the sewage treatment process, applied to the supervision system for the sewage treatment process according to any one of claims 1-9, characterized in that, The method includes the following steps: Collect the data of chemical oxygen demand, biochemical oxygen demand, pH value, heavy metal concentration, suspended solid concentration, flow rate, flow velocity and temperature of sewage through a multi-parameter sensor group and a flow monitoring device; According to the collected data, dynamically analyze the correlation of water quality parameters through a machine learning algorithm to generate a chemical dosing control instruction; According to the chemical dosing control instruction, add one or more of coagulants, flocculants or disinfectants to the sewage treatment system through a chemical dosing pump and a flow regulating valve, and adjust the dosing amount in real time; Based on the real-time feedback data, iteratively optimize the chemical dosing control instruction through an adaptive algorithm to form a closed-loop control.

Images