Method for intelligent control of treatment process based on sewage chemical characteristic identification

Abstract

The present application relates to the technical field of sewage treatment, and discloses a treatment process intelligent control method based on sewage chemical characteristic identification, wherein spectral sensors deployed in a sewage pipe network are used to obtain water inlet chemical fingerprint data, and the time when the water inlet reaches a treatment plant is predicted; before the water inlet reaches, the fingerprint data is input into a biochemical reaction potential analysis model to obtain a biochemical reaction potential vector, and based on the biochemical reaction potential vector, a water force affinity score of each parallel biochemical treatment unit is calculated; combined with sludge function portrait data obtained from the plant, a targeted gain affinity score for allocating specific return sludge sources to specific biochemical treatment units is calculated. Based on the two scores, the optimal flow distribution and sludge allocation strategy are determined through optimization solution, so as to effectively cope with water quality impact, ensure stable operation of the system, and significantly improve the impact resistance and stability of the system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sewage treatment, in particular to a treatment process intelligent control method based on sewage chemical characteristic identification. BACKGROUND

[0002] Sewage treatment is a key link of environmental protection, and it is crucial to ensure the stable operation of its treatment process and the continuous compliance of effluent water quality. However, modern sewage treatment plants, especially large centralized treatment facilities, have a wide service range and a complex upstream pipe network system, resulting in significant dynamic fluctuations and unpredictability of their influent water quality, which is particularly susceptible to impact loads caused by sudden industrial discharges and other events.

[0003] The control method in the prior art usually relies on online monitoring instruments installed at the entrance of the sewage treatment plant. The fundamental limitation of this mode is the lag in response, i.e., only when sewage with specific water quality characteristics has reached the plant, can the control system respond passively. In the face of rapid impact loads, this lagging emergency control often fails to provide sufficient buffer time for the system, easily leading to overload of the treatment unit, disorder of the microbial system, and even causing temporary effluent water quality exceeding standard events.

[0004] In addition, when distributing loads among multiple biochemical treatment units operating in parallel, traditional control strategies often use a homogenized distribution principle, such as average flow distribution according to the capacity of each unit. This approach does not fully consider the actual state differences of microbial communities in different biochemical treatment units, nor does it assess the biochemical reaction adaptability between specific influent water quality and different treatment units. Therefore, when facing influent water with specific chemical characteristics (such as high ammonia nitrogen or refractory organic matter), this "one-size-fits-all" distribution method cannot accurately guide the load to the most capable unit, resulting in waste of treatment resources and decline in overall treatment efficiency.

[0005] Furthermore, existing control logic often treats the distribution of hydraulic load and the allocation of functional microorganisms (such as return sludge) as two independent decision-making steps. Operators may first adjust the flow, and then adjust the sludge return according to experience, lacking a coordinated optimization mechanism. This step-by-step, decoupled decision-making approach easily leads to suboptimal results, i.e., the optimal solution for flow distribution may not be optimal for sludge resource utilization, and vice versa, thus failing to maximize the overall operational efficiency of the system. SUMMARY

[0006] The existing sewage treatment control method is usually based on the water quality monitoring data in the treatment plant or at the inlet to respond, which has time lag and is difficult to cope with the rapid change of water quality impact in the sewage pipe network. In addition, when there are multiple parallel treatment units in the sewage treatment plant, there is a lack of a technical means capable of distributing hydraulic load and scheduling biological efficiency in a forward-looking and synergistic manner according to the adaptability of the influent chemical characteristics to each treatment unit and microbial population, which limits the stability and treatment efficiency of the whole treatment system.

[0007] To solve the above technical problems, the present application provides a sewage chemical characteristic recognition-based treatment process intelligent control method, which comprises the following steps:

[0008] S1, acquiring the influent chemical fingerprint data collected by the spectrum sensor deployed in the sewage pipe network system in real time; the data collected by the spectrum sensor can reflect the comprehensive chemical composition of various pollutants such as organic matter, nitrogen and phosphorus in the influent.

[0009] S2, based on the influent chemical fingerprint data and the preset pipe network hydraulic transport model, predicting the time when the influent carrying the corresponding chemical characteristics reaches the sewage treatment plant.

[0010] S3, before the influent reaches the sewage treatment plant, inputting the influent chemical fingerprint data into a preset biochemical reaction potential analysis model to obtain a biochemical reaction potential vector (BRP, Biochemical Reaction Potential Vector) representing the potential influence of the influent on the subsequent biochemical treatment process.

[0011] S4, according to the biochemical reaction potential vector and in combination with the historical performance data of each parallel biochemical treatment unit in the sewage treatment plant, calculating a hydraulic affinity score (HAS, Hydraulic Affinity Score) for each biochemical treatment unit.

[0012] S5, based on the hydraulic affinity score, determining a distribution ratio for the flow distribution of the influent, and according to the time predicted in step S2, controlling the flow distribution equipment of the sewage treatment plant according to the determined distribution ratio.

[0013] In one embodiment, in step S3, the biochemical reaction potential vector BRP includes at least one potential energy component:

[0014] aerobic oxygen uptake rate potential (AOUR_P), nitrification inhibition potential (NIP), sludge yield potential (SYP), or toxicity impact potential (TIP). The biochemical reaction potential analytical model outputs the quantified potential component values by matching and calculating the influent chemical fingerprint data with a pre-established database.

[0015] Further, the method further comprises:

[0016] After step S3 and before step S4, acquiring sludge function profile data collected by biological monitoring devices deployed at each source of return sludge in the wastewater treatment plant;

[0017] inputting the sludge function profile data into a preset sludge function pedigree analytical model to obtain a sludge function pedigree vector (SFP, Sludge Functional Pedigree Vector) representing the functional characteristics of the source of return sludge.

[0018] Preferably, the sludge function pedigree vector SFP comprises at least one of the following functional components: general biological activity (GBA), tolerance to specific toxicity (STR), or specific degradation effect on specific refractory organic matter (SDR).

[0019] In one specific embodiment, the method further comprises, after step S4 and before step S5:

[0020] According to the biochemical reaction potential vector BRP and the sludge function pedigree vector SFP obtained in step S3, a targeted gain affinity score (TGAS, Targeted Gain Affinity Score) is calculated for each combination of the biochemical treatment unit and the source of return sludge. The targeted gain affinity score is used to quantify the expected gain effect of deploying the sludge from a specific source of return sludge to a specific biochemical treatment unit to cope with a specific influent.

[0021] Further, the method further comprises:

[0022] determining a targeted return strategy based on the targeted gain affinity score; and controlling the return sludge deployment equipment of the wastewater treatment plant according to the predicted time obtained in step S2 and the targeted return strategy to deploy the sludge from a specified source of return sludge to a specified biochemical treatment unit.

[0023] In one preferred embodiment, the steps of determining the allocation ratio and the targeted return strategy are realized through a unified optimization process, and the process is specifically:

[0024] First, an optimization problem is constructed, in which a flow distribution proportion vector and a deployment strategy matrix of each return sludge source to each biochemical treatment unit are taken as decision variables, where is the number of biochemical treatment units, is the index of the return sludge source, is the index of the biochemical treatment unit.

[0025] Then, an objective function of the optimization problem is established . The objective function is the weighted sum of the hydraulic affinity score and the targeted gain affinity score, and its mathematical expression is:

[0026] Maximize ;

[0027] subject to the following conditions:

[0028] wherein is the hydraulic affinity score of the th biochemical treatment unit, is the targeted gain affinity score of the th return sludge source combined with the th biochemical treatment unit, and is a preset weight coefficient, and + = 1.

[0029] Finally, by solving the optimization problem to maximize the objective function , the optimal distribution proportion vector and the targeted return strategy matrix are obtained simultaneously under the conditions of satisfying the flow balance constraint .

[0030] In one embodiment, the method further comprises:

[0031] After step S5, for each biochemical treatment unit that has received allocated flow, the operating parameters of the biochemical treatment unit are dynamically adjusted according to the biochemical reaction potential vector obtained in step S3 and in combination with the targeted return strategy formulated for the biochemical treatment unit.

[0032] Preferably, the operating parameters include aeration amount, internal return ratio, or reagent dosage.

[0033] In one specific embodiment, in step S1, the optical spectrum sensor is an ultraviolet-visible light full spectrum analyzer or a three-dimensional fluorescence spectrum analyzer.

[0034] The present application provides a sewage chemical feature recognition-based intelligent control method for a treatment process.

[0035] 1. The present application predicts the arrival time of influent by deploying a spectrum sensor upstream of the sewage pipe network and combining a pipe network hydraulic transport model, and prepositions the time window of the control decision. This overcomes the inherent response lag of the traditional control method which relies on monitoring at the entrance of the treatment plant, so that the control system can calculate and deploy the coping strategy in advance before the actual arrival of the influent with specific chemical characteristics, thereby realizing the forward-looking regulation of the treatment process.

[0036] 2. The present application realizes non-equilibrium flow distribution based on water quality and treatment unit adaptability by analyzing the biochemical reaction potential vector of the influent and calculating the hydraulic affinity score for each parallel treatment unit. This method can accurately guide the influent with specific chemical characteristics to the unit with the highest treatment efficiency, thereby improving the utilization efficiency of treatment resources and the specificity of the system in responding to complex water quality changes.

[0037] 3. The present application integrates the decisions on hydraulic load scheduling and biological resource efficiency improvement by constructing an optimization problem that takes flow distribution ratio and targeted backflow strategy as unified decision variables, and taking hydraulic affinity and targeted gain affinity score as collaborative optimization objectives. This collaborative optimization mechanism can generate a globally coordinated control strategy, avoiding suboptimal or decision conflicts caused by single or step-by-step decisions, thereby improving the stability and effectiveness of the overall operation of the system. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 FIG. 1 is an execution environment schematic diagram of a sewage chemical feature recognition-based intelligent control method for a treatment process according to an embodiment of the present application;

[0039] Figure 2 FIG. 2 is a collaborative optimization problem solving flowchart according to an embodiment of the present application;

[0040] Figure 3 FIG. 3 is a comparison diagram of the effect of the method of the present application in response to impact load. DETAILED DESCRIPTION

[0041] With reference to the accompanying drawings, the technical solutions in the embodiments of the present application will be clearly and completely described in the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0042] Please refer to the accompanying drawings Figure 1 , Figure 1 is an execution environment schematic diagram of a sewage chemical feature recognition-based process intelligent control method according to an embodiment of the present application. The present application provides a process intelligent control method, which is executed in a system with an upstream monitoring unit 100, an in-plant monitoring and execution unit 200 and a central control and calculation unit 300 in a specific implementation, and specifically includes the following steps:

[0043] Step 1: Real-time acquisition of multi-source heterogeneous data. This step aims to obtain comprehensive and real-time working condition information for subsequent optimization decision. This step is specifically implemented by the following ways:

[0044] Acquisition of influent chemical fingerprint data: At least one optical spectrum sensor 110 (such as an ultraviolet-visible light full spectrum analyzer or a three-dimensional fluorescence spectrum analyzer) in the upstream monitoring unit 100 deployed at the upstream pipe network node of the sewage treatment plant is used to continuously or high-frequency online spectrum scanning of the sewage about to enter the plant, so as to obtain the influent chemical fingerprint data representing the water quality chemical component information.

[0045] Acquisition of sludge function portrait data: The biological monitoring equipment 210 deployed at each source of backflow sludge in the plant is used to monitor the biological activity state of each backflow sludge, so as to obtain the sludge function portrait data representing the strength of specific degradation function (such as nitrification, denitrification, toxicity resistance, etc.).

[0046] The above two types of data collected are transmitted in real time to the central control and calculation unit 300 for processing through the respective data transmission modules.

[0047] Step 2: Collaborative optimization decision based on data analysis. This step is the core of the method of the present application, and the processor in the central control and calculation unit 300 executes the preset program to complete it. This step receives the above-mentioned collected data and performs a series of calculations and decisions, specifically including:

[0048] Analysis of affinity score: calling biochemical reaction potential analysis model, the influent chemical fingerprint data is analyzed into quantitative hydraulic affinity score HAS; at the same time, calling sludge function spectrum analysis model, the sludge function portrait data is combined with the influent characteristics to analyze into quantitative targeted gain affinity score TGAS.

[0049] Construct and solve optimization problem: based on the affinity score parsed, a decision optimization module is executed. The core task of this module is to construct and solve a mixed integer linear programming (MILP) problem, i.e. to find a set of optimal decision variables (flow allocation ratio vector) and (sludge allocation strategy matrix) that can maximize the total objective function Maximize under a series of physical and process constraints.

[0050] Third step: optimal decision-based collaborative control instruction execution. This step aims to convert the abstract mathematical solution calculated in the second step into precise control of physical equipment. The central control and calculation unit 300 converts the optimal solution and into specific equipment control instructions and sends them to the plant monitoring and execution unit 200. This unit performs the following collaborative actions according to the instructions:

[0051] Execute flow allocation: drive the flow allocation equipment 220 (such as PLC-controlled adjustable valves) installed on the inlet pipelines of each parallel biochemical tank to accurately adjust the opening degree, so that the actual inlet flow ratio of each unit strictly follows the requirements of the optimal solution .

[0052] Perform targeted sludge allocation: drive the return sludge allocation equipment 230 composed of sludge pumps and valve systems, according to the instructions corresponding to the elements with a value of 1 in the optimal solution , accurately add specific return sludge sources to designated biochemical treatment units.

[0053] Perform dynamic adjustment of operating parameters: drive the operating parameter adjustment equipment 240 (such as air blowers, internal return pumps, etc.) associated with each biochemical tank, according to the process load and treatment targets implied by the optimal solution and , feed-forward adjust key process parameters such as dissolved oxygen and internal return ratio to create the optimal local biochemical environment for the allocated tasks.

[0054] By repeatedly executing the above steps, the method of the present application can achieve intelligent and collaborative closed-loop control of the entire wastewater treatment process based on real-time sensing of water quality characteristics.

[0055] In one specific embodiment of the present application, the process of executing step S1 to obtain the inlet water chemical fingerprint data is completed by the spectrum sensor 110 deployed in the pipe network system.

[0056] In one embodiment, the optical sensor 110 is a three-dimensional fluorescence spectrograph. The spectrograph is connected to the sewage pipeline through a flow cell, so that the sewage in the pipeline can continuously flow through the measurement area of the spectrograph. The light source of the spectrograph emits a series of excitation light of preset wavelengths, which successively irradiate the sewage sample flowing through the flow cell. For each excitation wavelength, the photodetector of the spectrograph synchronously records the emission spectrum intensity in a complete wavelength range in the direction perpendicular to it, thereby forming a raw fluorescence data matrix.

[0057] The raw fluorescence data matrix is subjected to a series of correction and preprocessing steps to eliminate instrument background noise and physical interference, thereby forming influent chemical fingerprint data that accurately reflects the chemical composition of the water sample. The preprocessing process includes: first, by measuring and deducting the blank matrix generated by ultrapure water (Milli-Q water) under the same test conditions, to eliminate the Raman scattering of water and the background fluorescence of the device.

[0058] Secondly, the data matrix after deducting the blank is corrected for inner filter effect. The inner filter effect is a nonlinear decay of fluorescence intensity caused by the absorption of high-concentration substances in the sample to excitation light or emission light. The correction process can be carried out by the following formula:

[0059] wherein, is the corrected fluorescence intensity, is the fluorescence intensity measured by the instrument, is the absorbance of the sample at the excitation wavelength, is the absorbance of the sample at the emission wavelength. The absorbance value can be measured synchronously by the same device or an auxiliary ultraviolet-visible light absorption spectrometer.

[0060] After the above preprocessing steps, a two-dimensional excitation-emission matrix (EEM) is obtained, which is an influent chemical fingerprint data at a specific time . The X-axis of the matrix is the emission wavelength, the Y-axis is the excitation wavelength, and the value of each element in the matrix is the corrected fluorescence intensity measured at the excitation wavelength at the emission wavelength .

[0061] The influent chemical fingerprint data is digitized and sent to the central control and calculation unit 300 through the data transmission module of the upstream monitoring unit 100, for subsequent calculation in steps S2 and S3. This process is repeated at a preset time frequency (e.g. every 5 minutes), thereby forming a time series data set that continuously reflects the changes in the chemical characteristics of the upstream water quality.

[0062] After the acquisition of the influent chemical fingerprint data, and before the influent carrying the corresponding chemical characteristics reaches the wastewater treatment plant, the central control and computing unit 300 performs step S3, inputs the influent chemical fingerprint data into the preset biochemical reaction potential analysis model, to calculate the biochemical reaction potential vector (BRP).

[0063] In a specific embodiment, the biochemical reaction potential analysis model is a pre-trained multivariate linear regression model, such as a Partial Least Squares Regression (PLSR) model. The model construction process is completed before system deployment, which includes the following steps: first, a large number of historical wastewater samples are collected, for each sample, the influent chemical fingerprint data , and a set of biochemical reaction parameters such as oxygen consumption rate, nitrification inhibition rate, sludge yield, and toxicity inhibition level of specific microorganisms are measured through laboratory-scale batch reaction experiments.

[0064] Secondly, the collected data is processed. Each two-dimensional influent chemical fingerprint data matrix is unfolded into a one-dimensional row vector . The multiple laboratory-determined biochemical reaction parameter values corresponding to this sample form a response vector . Combining the data of all historical samples forms an independent variable matrix and a dependent variable matrix.

[0065] Finally, the Partial Least Squares Regression algorithm is applied to analyze the statistical relationship between the independent variable matrix and the dependent variable matrix, and a stable mathematical relationship that can map the input fingerprint vector to the output response vector is established, which is defined by a regression coefficient matrix . The regression coefficient matrix is stored in the biochemical reaction potential analysis model of the central control and computing unit 300.

[0066] In the real-time running phase of the system, when a new influent chemical fingerprint data is acquired and transmitted to the central control and computing unit 300, the biochemical reaction potential analysis model performs the following calculations:

[0067] First, the two-dimensional matrix is unfolded into a one-dimensional row vector in the same way as during training.

[0068] Then, the biochemical reaction potential vector is calculated by matrix multiplication operation:

[0069] wherein: is the output biochemical reaction potential vector, in the form of ; is the aerobic oxygen uptake rate potential component; is the nitrification inhibition potential component; is the sludge yield potential component; is the toxicity shock potential component; is a one-dimensional row vector unfolded from the current influent chemical fingerprint data; is the regression coefficient matrix stored in the pre-trained PLSR model.

[0070] The calculated biochemical reaction potential vector is a digital vector containing multiple quantitative components. Each component of the vector represents the potential impact of the influent on a specific biochemical process in the subsequent biochemical treatment process. The vector is then passed to the decision optimization module as the core data input for executing step S4 and subsequent steps.

[0071] In a specific embodiment, the method of the present application further comprises, after step S3 and before step S4, obtaining sludge functional profile data from the biological monitoring device 210 deployed at each source of return sludge, and inputting the data into a pre-set sludge functional spectrum analysis model to obtain a sludge functional spectrum vector (SFP) representing the functional characteristics of the sludge source.

[0072] The biological monitoring device 210 can be an online respirometer. The online respirometer is connected to the return sludge pipeline and can automatically and periodically extract sludge samples into the reaction chamber inside it. In the reaction chamber, the online respirometer can accurately measure and record the oxygen uptake rate of the sludge under different conditions, and these oxygen uptake rate data constitute the sludge functional profile data. The sludge functional spectrum analysis model is a set of pre-set calculation rules and formulas integrated in the central control and calculation unit 300, used to convert the sludge functional profile data into a standardized sludge functional spectrum vector .

[0073] One component of the sludge functional spectrum vector is the overall biological activity (GBA). The calculation process is as follows: the online respirometer first measures the endogenous oxygen uptake rate of the extracted sludge sample under the condition of no external carbon source supply ( ). The sludge functional spectrum analysis model then calculates the GBA component by the following formula :

[0074] wherein, is a pre-set reference oxygen uptake rate value, which represents the average biological activity level of the activated sludge under normal operation of the treatment plant.

[0075] Sludge functional profile vector Another component of the sludge functional profile vector is the specific toxicity resistance (STR). The calculation process is as follows: after the determination of the endogenous oxygen uptake rate, the online respirometer automatically adds a standard concentration of a pre-selected representative toxic substance (e.g. 3,5-dichlorophenol) to the same sludge sample in the reaction chamber. After a fixed contact time, the oxygen uptake rate of the sludge at this time is measured again, denoted as the post-inhibition oxygen uptake rate (R ). The STR component is calculated by the following formula:

[0076] This ratio directly quantifies the tolerance of the sludge microorganisms to the specific toxic substance.

[0077] Sludge functional profile vector Another component of the sludge functional profile vector is the specific degradation resistance (SDR) to a specific refractory organic substance. The calculation process is similar to that of STR, with the difference being that the substance added to the reaction chamber is one or more pre-selected representative refractory organic substances. By measuring the increment of the oxygen uptake rate after adding the substance, the specific oxygen uptake rate (R ) for the specific substance can be calculated. The SDR component can be represented as: wherein, is the total oxygen uptake rate measured after adding the specific refractory organic substance.

[0078] By performing the above series of measurements and calculations, the central control and calculation unit 300 generates a sludge functional profile vector containing multiple functional components for each source of return sludge. This vector is digitally stored and transmitted to the decision optimization module for calculating the targeted gain affinity score and formulating the targeted return strategy.

[0079] After the central control and calculation unit 300 generates the biochemical reaction potential vector (BRP), step S4 is performed to calculate a hydraulic affinity score (HAS) for each biochemical treatment unit j (j = 1, 2,..., n) of the n parallel biochemical treatment units in the wastewater treatment plant.

[0080] In a storage module of the central control and calculation unit 300, a historical performance portrait vector is pre-stored for each biochemical treatment unit j. The historical performance portrait vector is composed of the statistical analysis results of long-term operation monitoring data of the unit, which contains the biochemical reaction potential vector performance indicators corresponding to each component. In one specific embodiment, the historical performance portrait vector is in the form of: where each component is a value extracted from the historical database and processed by standardization: is the historical average aerobic loading treatment capacity of unit j, representing its performance in responding to high organic matter concentration shock; is the historical nitrification stability index of unit j, representing its performance in maintaining nitrification function in the face of nitrification inhibitors; is the inverse of the historical sludge yield coefficient of unit j, representing its performance in low sludge production; is the historical toxicity shock tolerance of unit j, representing its performance in biological activity recovery after encountering toxic substances.

[0081] The calculation process of the hydraulic affinity score is to perform mathematical operations on the biochemical reaction potential vector and the historical performance portrait vector of each biochemical treatment unit j, thereby obtaining a scalar value quantitatively evaluating the degree of adaptation of unit j to the current influent chemical characteristics.

[0082] In one specific embodiment, the hydraulic affinity score is calculated by a linear weighted summation model. The mathematical expression of the model is:

[0083] wherein, is the calculated hydraulic affinity score of biochemical treatment unit j; is the index of the components in the biochemical reaction potential vector and the historical performance portrait vector, representing a specific biochemical reaction dimension (such as AOUR, NIP, SYP, TIP); is the preset weight coefficient of the th component, which is set to reflect the relative importance of different biochemical reaction dimensions in the overall goal of wastewater treatment; is the value of the th component of the biochemical reaction potential vector ; is the value of the th component of the historical performance portrait vector of biochemical treatment unit j; correspond to the weight coefficients of the four dimensions of "aerobic oxygen consumption rate", "nitrification inhibition", "sludge yield", and "toxic shock", respectively; ​​​​The biochemical reaction potential components of the influent in the four dimensions of "aerobic oxygen consumption rate", "nitrification inhibition", "sludge production rate" and "toxic shock" are respectively compared. , , , These correspond to the historical performance profile components of the j-th biochemical treatment unit in the four dimensions of "aerobic oxygen consumption rate", "nitrification inhibition", "sludge yield" and "toxic shock".

[0084] The structure of this calculation formula allows for a higher specific historical performance ( Biochemical processing units with high potential (high value) are facing biochemical processing units with correspondingly high reaction potential (high value) When the inflow value is large, a higher hydraulic affinity score can be obtained.

[0085] The central control and computing unit 300 is for all parallel-operating biochemical processing units (j=1 to...). ) calculate their respective Then, the obtained score vector The output is sent to the decision optimization module as the core basis for determining the traffic allocation ratio.

[0086] In one specific implementation, the method, after step S4 and before step S5, further includes configuring each biochemical treatment unit j with each return sludge source. A combination of these factors is used to calculate a Target Gain Affinity Score (TGAS). The score is... Used to quantify the source of returned sludge. The sludge is mixed into the biological treatment unit j to meet the expected gain effect when the specific influent is used.

[0087] This calculation process is executed by the central control and computing unit 300. The process uses the current biochemical reaction potential vector. m sludge functional spectrum vectors from m return sludge sources (i=1,2,…,m), and the historical performance profile vector of each biochemical treatment unit. As input.

[0088] Target gain affinity score The calculation aims to evaluate the functional characteristics of sludge source i (by... Characterization) and the performance shortcomings of biochemical treatment unit j (by Characterization) in addressing the biochemical challenges posed by influent (by The degree of collaborative matching during characterization.

[0089] In one specific implementation, the score is calculated using the following mathematical expression:

[0090] wherein, is the targeted gain affinity score calculated for the combination of the return sludge source i and the biochemical treatment unit j; and are preset weight coefficients for the toxicity tolerance dimension and the specific degradation dimension, respectively; is the biochemical reaction potential vector with the toxicity impact potential component; is the sludge functional spectrum vector of the ith return sludge source with the tolerance component to the specific toxicity; is the historical performance portrait vector of the jth biochemical treatment unit with the normalized (between 0 and 1) historical toxicity impact tolerance component. The factor represents the “performance short board” or “gain demand” of the unit in terms of toxicity tolerance; is the biochemical reaction potential vector with the degradation demand potential component representing the degradation demand of the influent containing the specific refractory organics; is the sludge functional spectrum vector of the ith return sludge source with the degradation specificity component to the specific refractory organics; is the historical performance portrait vector of the jth biochemical treatment unit with the normalized historical specific substance degradation performance component corresponding to . The factor represents the “performance short board” of the unit in terms of the specific degradation function.

[0091] Through this modified calculation model, the score is determined by three parties: the influent has a clear challenge ( value is high), the sludge has the ability to respond (c value is high), and the target unit is exactly deficient in this respect ((1−h) value is high). This makes the score truly and accurately reflect the logic of “targeted gain” and provides a solid data foundation for subsequent collaborative optimization decisions.

[0092] The central control and calculation unit 300 calculates for each possible combination of the return sludge source and the biochemical treatment unit j according to the above model, thereby forming a targeted gain affinity score matrix of mxn dimensions. This matrix is then output to the decision optimization module as the core basis for determining the targeted return strategy.

[0093] After calculating the hydraulic affinity score and the target gain affinity score, the decision optimization module inside the central control and computing unit 300 executes step S5 to generate control decisions by constructing and solving a co-optimization problem. The first step of this process is to precisely define the decision variables of the co-optimization problem.

[0094] In one specific embodiment of the present invention, the collaborative optimization problem includes two sets of decision variables: one set for determining the allocation of influent flow rate and the other set for determining the distribution of return sludge.

[0095] The first set of decision variables is the flow allocation ratio vector. This vector is used to determine the distribution ratio of the total influent flow rate among the n parallel-operating biological treatment units. Flow rate distribution ratio vector The mathematical form is a group containing A one-dimensional vector with n elements: , where each element As a continuous variable, its physical meaning is assigned to the first... The proportion of the influent flow rate of each biochemical treatment unit to the total influent flow rate. Therefore, each element... As a continuous variable, its physical meaning is assigned to the first... The proportion of the influent flow rate of each biological treatment unit to the total influent flow rate. Therefore, The range of values ​​for is limited to: Furthermore, the sum of all elements must be exactly equal to 1 to satisfy the fundamental physical constraint of flow conservation: .

[0096] The second set of decision variables is the sludge allocation strategy matrix. This matrix represents a specific scheme for allocating m different sources of return sludge to n biological treatment units. Sludge Allocation Strategy Matrix The mathematical form of is a OK A two-dimensional matrix of columns:

[0097] , where each element Let be a binary integer variable, whose value is defined as: ;

[0098] when When, it indicates that the execution will be the first The sludge from the first return sludge source was allocated to the first... Operation of a biochemical processing unit; when When the time is specified, it indicates that the operation will not be performed.

[0099] To ensure the specificity and effectiveness of targeted gain, a sludge preparation strategy matrix is ​​used. The elements in the above equation also need to satisfy the following constraints:

[0100] The physical meaning of this constraint is that, in a single control cycle, each specific return sludge source i (especially those with special functional characteristics) can be allocated to at most one target biochemical treatment unit Or not used.

[0101] The flow distribution ratio vector And the sludge allocation strategy matrix Together constitute the complete decision variable set of the collaborative optimization problem. The goal of the decision optimization module is to calculate the optimal values of these two groups of variables.

[0102] In one specific embodiment of the present application, the objective function of the collaborative optimization problem is defined as And is set as a maximization problem. This objective function performs a weighted summation of the hydraulic affinity score and the targeted gain affinity score to form a comprehensive index that can reflect the overall operational benefit of the system.

[0103] The mathematical expression of the objective function is: Maximize Wherein, is the objective function to be maximized, and its calculation result is a scalar value representing the "overall operational affinity" or "comprehensive benefit score" that the entire wastewater treatment system can obtain under the specific control strategy defined by the decision variables (flow distribution ratio vector) and (sludge allocation strategy matrix).

[0104] The first term Represents the total hydraulic affinity score. This term aims to reward decisions that allocate a larger proportion of the incoming water flow (larger ) to treatment units that are more suitable for the current water quality (i.e., have a higher hydraulic affinity score ). It reflects the optimization level of the system at the macro flow distribution level.

[0105] The second term Represents the total targeted gain score. This term aims to quantify the cumulative benefits brought by all executed biological gain operations (i.e., all allocations with = 1). Since is a binary variable, it acts as a "switch": only when the decision model decides to implement a certain allocation from the sludge source to the treatment unit , the potential gain of this allocation will be counted into the total score, otherwise its contribution is zero.

[0106] + are preset weight coefficients. These two dimensionless parameters are hyperparameters that can be preset by system operators or high-level control logic to balance and trade-off between the two optimization objectives. In one specific embodiment, these two weight coefficients satisfy the constraint conditions + = 1 and , ≥ 0. By adjusting these two coefficients, the optimization objectives can be made to focus more on improving the efficiency and stability of the treatment of conventional pollutants (e.g., set = 0.8, = 0.2) or more on actively and quickly responding to sudden toxic or refractory substance shocks (e.g., set = 0.3, = 0.7).

[0107] The final goal of the decision optimization module is to find a set of optimal decision variables and that maximize the objective function under the premise of satisfying all the constraints defined above (e.g., and ) through standard mixed-integer linear programming (MILP) solvers or other applicable optimization algorithms.

[0108] In one specific embodiment of the present application, the solution of the collaborative optimization problem needs to be carried out under the following set of constraint conditions:

[0109] Decision variable inherent constraints: This set of constraints is a basic definition of the value range and type of the decision variables themselves and is a formal component of building the optimization model:

[0110] Flow allocation proportion range constraints: This constraint ensures that the proportion of flow allocated to any treatment unit is a valid percentage.

[0111] Sludge allocation decision type constraints: This constraint explicitly defines that the sludge allocation decision is a binary switch of "execute" or "not execute".

[0112] Material conservation and logic constraints: This set of constraints is the key to ensuring the correctness of the system's material balance and core control logic:

[0113] Total flow conservation constraints: , which ensures that the sum of the proportions of the total influent flow received by all parallelly operated biochemical treatment units is exactly 100%, i.e. all influent flow is allocated without omission or creation.

[0114] Sludge source uniqueness constraint: This inequality constraint is the core logic of achieving "targeted" gain. It stipulates that each reflux sludge source (particularly those identified as having specific functions, such as high-efficiency nitrogen removal or toxicity resistance) can only be directed to one target biochemical treatment unit at most in a single control cycle. This avoids the inefficient dispersion of valuable special microorganism resources.

[0115] Operating process and physical limit constraints: This set of constraints is a manifestation of the invention's further deepening of technical details and ensuring the feasibility of the solution, which takes into account the upper limits of physical and biological treatment capacity of the treatment units:

[0116] Treatment unit hydraulic load constraint: wherein is the total influent flow (m 3 / h) at the current time, and and are the minimum and maximum operating influent amounts respectively that the th biochemical treatment unit is designed to allow. This constraint prevents reactor hydraulic shock, short flow, or excessive aeration, microbial activity reduction, etc. due to uneven distribution of flow (excessive ) or too small flow (excessive ).

[0117] Treatment unit sludge receiving constraint: This constraint limits from the receiving end, stipulating that any biochemical treatment unit can only receive targeted addition from one sludge source in a single control cycle. This initiative has important process significance: it avoids the simultaneous addition of multiple sludge with different functional characteristics (which may be antagonistic) to a reactor, thereby simplifying process control and making it possible to evaluate and verify specific biological enhancement behaviors.

[0118] All the above constraints collectively define a "feasible region" of the optimization problem. The solution process of the decision optimization module is essentially a process of finding the unique solution that maximizes the objective function within this feasible region.

[0119] Refer to the attached Figure 2, Figure 2 is a schematic diagram of the solution process of a co-optimization problem according to an embodiment of the present application. Once the objective function and all the constraints are completely defined, they collectively constitute a formalized, structured mathematical optimization problem. The solution process of this problem is automatically executed by the decision optimization module inside the central control and computation unit 300, with the goal of efficiently and accurately computing the optimal control decisions.

[0120] In one specific embodiment of the present application, through the analysis of the structure of this optimization problem, it can be classified as a Mixed-Integer Linear Programming (MILP) problem. The basis of this classification is:

[0121] The objective function is linear: the objective function is a linear weighted sum of the decision variables and .

[0122] The constraints are linear: all the constraints, including the equality constraints and the inequality constraints, are linear expressions with respect to the decision variables.

[0123] The variable types are mixed: the set of decision variables contains both continuous variables (flow allocation ratios) that take values in the interval [0, 1] and binary integer variables (solid waste allocation switches) that take values in the set {0, 1}.

[0124] The decision optimization module integrates one or more specialized solvers that are optimized for solving MILP problems of this type. At the beginning of each control decision cycle, this module first substitutes the real-time computed hydraulic affinity scores and targeted gain affinity scores as constant parameters into the expressions of the objective function and the constraints, thus generating a specific mathematical instance to be solved at this moment.

[0125] Subsequently, the module calls the MILP solver to solve this instance. The solver internally employs robust algorithms such as the combination of Branch and Bound and Simplex Method. This algorithm is capable of systematically searching within the complex, multi-dimensional feasible region defined by all the constraints, to guarantee finding an optimal solution that globally maximizes the objective function within a limited time.

[0126] The final output of this solving process is not only the maximum value of the objective function, but also the values of the optimal decision variables corresponding to this maximum value, denoted as and .

[0127] Optimal flow distribution ratio vector where each element is a specific numerical value, such as [0.55, 0.45], which means that 55% of the total influent flow should be allocated to the biochemical treatment unit 1 and 45% to unit 2.

[0128] Optimal sludge allocation strategy matrix which is an m x n matrix composed of 0s and 1s, where the element with a value of 1 (e.g. ) precisely instructs that the sludge from which reflux sludge source (source 2) should be targeted to which biochemical treatment unit (unit 1).

[0129] The central control and computing unit 300 finally converts this set of optimal solutions into specific physical device control instructions, such as precisely allocating flow by adjusting the valve opening on the relevant pipeline, and performing specific sludge allocation tasks by controlling the start and stop of pumps and flow direction switches. This entire solving and execution process is repeated at a preset time period (e.g. every 30 minutes) to achieve continuous, dynamic and intelligent collaborative optimization control of the wastewater treatment process.

[0130] After the decision optimization module solves the optimal flow distribution ratio vector and the optimal sludge allocation strategy matrix , these decision instructions are not immediately and synchronously issued. Considering that a wastewater treatment plant is a large physical system with significant time lag effects, the invention further discloses a time prediction-based collaborative control execution method to ensure the accuracy and timeliness of control actions.

[0131] The core idea of this method is to convert the static optimal decision solution into a dynamic, accurately time-scheduled execution sequence. After receiving the optimal solution and , the control execution module inside the central control and computing unit 300 will perform the following steps:

[0132] Retrieve and calculate time lag parameters The control execution module first retrieves or calculates a set of key time lag parameters from the system configuration database or according to real-time operating conditions. These parameters are quantifications of the time required for control instructions to produce actual process effects from issuance.

[0133] Hydraulic transport time lag ): This refers to the point from which the total intake water is distributed (e.g., the flow distribution well) to the point where the water flow reaches the [number missing]. The time required for the inlet of a biochemical treatment unit. In one specific implementation, this time delay is not a fixed constant, but rather depends on the pipe geometry parameters (length). Pipe diameter ) and the current flow allocated to this unit ( The velocity is dynamically calculated to reflect the changes in flow rate under different flow rates.

[0134] Sludge transport and mixing time delay ( ): This refers to the total time required from the start of the pump or valve transporting sludge from the i-th sludge source to the j-th biological treatment unit until that portion of the sludge is substantially uniformly mixed into the reaction zone of the biological treatment unit, sufficient to begin exerting its biosynergistic effect. This parameter integrates pumping time, pipeline transport time, and mixing time within the reactor.

[0135] Determining the target time for control synchronization is based on the calculated optimal flow allocation decision. The module will determine a control synchronization target time for each biochemical treatment unit j that receives the incoming water. The physical meaning of this time is the estimated time when the "batch" of influent currently being measured and analyzed at the flow distribution point is expected to arrive at biological treatment unit j. The calculation formula is: ,in, The start time for performing optimization calculations for the current control cycle.

[0136] The key task of the asynchronous execution instruction sequence control module is to ensure that the effects of control actions that require coordination (i.e., flow allocation and targeted sludge dosing) can be achieved within the target time. To achieve synchronization, it calculates the specific startup time of each physical device "backwards," generating an asynchronous instruction sequence:

[0137] Flow distribution command: Commands to adjust flow distribution valves can typically be issued immediately or with a very short system response delay. Execution time... Approximately .

[0138] Sludge blending instruction: For any sludge blending operation that is decided to be performed (i.e. The corresponding pump and valve start commands are not in Instead of being issued, it is scheduled for a calculated future time. Execution. The formula for calculating the execution time is: This calculation ensures that the sludge allocation action, although "pre-emptively" initiated, its bio-enhancing effect is precisely timed to coincide with the arrival of the target water body that it is intended to treat.

[0139] By the above mechanism, the present invention avoids the problem of control mismatching caused by physical time lag. For example, it prevents functional sludge from being prematurely dosed into a reactor when the characteristic pollutants it is intended to degrade are still far away in the incoming water pipeline; similarly, it also avoids the ineffective enhancement of functional sludge that arrives too late after the characteristic pollutants have already entered the reactor. This time-prediction-based collaborative execution strategy greatly improves the actual conversion efficiency of the targeted gain, accurately translating the mathematical optimal solution into the optimal process control effect in the physical world.

[0140] The collaborative optimal control of the present invention is not limited to macroscopic flow distribution and sludge allocation, but further deepens to the fine-tuning and self-adaptive adjustment of the internal operating environment of each independent biochemical treatment unit. At the same time when the collaborative control instruction is issued, a parameter dynamic adjustment mechanism closely coupled with the decision is activated, whose core goal is to create optimal local biochemical reaction conditions for the allocated task (defined by and ).

[0141] In a specific embodiment, the central control and calculation unit 300 contains an operating parameter adjustment submodule. This module receives the optimal solution and output by the decision optimization module as the main input, and combines the affinity score and with the operating condition interpretation information behind the decision to perform a feed-forward dynamic adjustment of the key operating parameters of each biochemical treatment unit .

[0142] This dynamic adjustment mainly manifests in the following aspects:

[0143] The feed-forward aeration amount adjustment operating parameter adjustment submodule will calculate the organic loading rate (OLR) that the jth treatment unit will bear in real time according to the optimal flow proportion and the total incoming water flow allocated to it. Compared to the traditional feedback control based on effluent dissolved oxygen (DO), the present invention adds a feed-forward control loop:

[0144] When a treatment unit j is allocated a higher flow proportion (i.e. When the incoming wastewater has a relatively high (or large) organic and ammonia load, the system predicts that the organic and ammonia load will increase significantly. To this end, the module proactively increases the dissolved oxygen (DO) setpoint of the aerobic zone of the unit (e.g., from the regular 2.0 mg / L to 2.5 mg / L) and correspondingly increases the blower output. This ensures that the microorganisms have sufficient oxygen for respiration and nitrification when the load shock arrives, avoiding a decrease in pollutant removal efficiency due to oxygen deficiency.

[0145] One of the core innovations of the process parameter optimization operation parameter adjustment submodule that targets gain functions is that it can "understand" sludge allocation decisions The strategic intention behind it, and the coordinated adjustment of process parameters to maximize the gain effect.

[0146] Example 1: Strengthening nitrification If the decision is to allocate a sludge source i rich in efficient nitrifying bacteria to the treatment unit j (i.e. , and The high score is because denitrification needs to be strengthened), the module not only ensures sufficient DO as above, but also may moderately reduce the internal recirculation (Mixed Liquor Recirculation, MLR) ratio of the unit. The purpose of this is to prolong the sludge retention time (SRT) in the aerobic zone, providing a more favorable breeding environment for slow-growing nitrifying bacteria, thereby maximizing the targeted gain effect.

[0147] Example 2: Strengthening denitrification If the decision is to allocate a sludge source i with high-efficiency denitrification capability to the treatment unit j, the module will take the opposite strategy. It will significantly increase the internal recirculation ratio of the unit, sending more nitrate (NO3 − −N) produced in the aerobic zone back to the anoxic zone quickly, providing sufficient electron acceptors for the newly added dominant denitrifying bacteria, thereby achieving efficient denitrification.

[0148] Differential energy-saving control of system energy efficiency This dynamic adjustment mechanism not only pursues treatment effect, but also strives to optimize overall energy consumption.

[0149] For those treatment units that are allocated a relatively low hydraulic load (i.e. smaller), and do not perform targeted gain operations (i.e. all ), the module will not adjust the process parameters of the unit. , the operating parameter adjustment sub-module will identify it as a "base load" operation mode. In this mode, the module will lower its DO set value to a lower level (e.g. 1.5 mg / L) that can meet the basic treatment needs, and reduce the frequency of the internal reflux pump. This differentiated energy-saving control supplies limited energy to the unit that undertakes the main treatment task, avoiding energy waste on low-load units.

[0150] In this way, the present application organically combines macroscopic allocation decisions with microscopic internal parameter adjustment of the unit, forming a multi-level, synergistic control system. It ensures that each control decision is not only optimal in terms of resource allocation, but also provides the most suitable biochemical environment for its execution level.

[0151] In one embodiment, Figure 3 is a schematic diagram of the recorded effluent water quality change results of the system described in the present application in response to a specific scenario of a sudden high-concentration pollution impact load on the upstream pipe network. The graph is a two-dimensional coordinate graph, with the horizontal coordinate representing time and the vertical coordinate representing the concentration of a key pollutant (ammonia nitrogen) in the effluent of the wastewater treatment plant.

[0152] Referring to Figure 3 , the graph contains the following elements:

[0153] A horizontal line segment, labeled as discharge standard. This line segment represents a pre-set compliance threshold of the effluent pollutant concentration.

[0154] A blue dashed line, labeled as traditional control method. This curve represents the change process of the effluent pollutant concentration in a system that does not use the method described in the present application. Before , the value of the blue dashed line is below the threshold of the discharge standard. After , its value rises rapidly and is above the threshold defined by the discharge standard for a period of time, and then slowly decreases to below the threshold.

[0155] A black line, labeled as using the method of the present application. This curve represents the change process of the effluent pollutant concentration in a system that uses the method described in the present application. Throughout the time axis, the value of the black line is always below the threshold defined by the discharge standard. Around , the value of this curve appears a small range of fluctuations, and then returns to stable.

[0156] Figure 3 Two specific time points are also marked in the graph:

[0157] Time point : represents the moment when the upstream monitoring unit 100 detects that the chemical fingerprint of the sewage in the pipe network has changed significantly (corresponding to high-concentration pollution) by the spectrum sensor 110. The method of the present application starts the analysis of data and the calculation of collaborative optimization decisions after .

[0158] Time point : represents the moment when the high-concentration pollution sewage detected at is actually transported by the pipeline and arrives at the sewage treatment plant.

[0159] The figure shows that, when dealing with the technical problem of the same upstream high-concentration pollution impact, the traditional control method shown by curve A has the technical effect of exceeding the effluent quality standard after the impact load arrives (after ). In contrast, the method of the present application shown by the black line detects and performs predictive collaborative control at the upstream , so that the effluent quality is always maintained below the threshold value defined by the emission standard when the impact load arrives (at ), thereby solving the technical problem that the prior art cannot effectively deal with sudden impact loads.

Claims

1. A smart control method for wastewater treatment processes based on the identification of wastewater chemical characteristics, characterized in that, Includes the following steps: S1. Obtain in-situ chemical fingerprint data in real time from spectral sensors deployed in the sewage pipe network system; S2. Based on the influent chemical fingerprint data and the pipeline hydraulic transport model, predict the time when the influent carrying the corresponding chemical characteristics will arrive at the wastewater treatment plant. S3. Before the influent reaches the wastewater treatment plant, the chemical fingerprint data of the influent is input into a preset biochemical reaction potential analysis model to obtain a biochemical reaction potential vector characterizing the potential impact of the influent on the subsequent biochemical treatment process. S4. Based on the biochemical reaction potential vector and combined with the historical performance data of each parallel biochemical treatment unit in the wastewater treatment plant, calculate a hydraulic affinity score for each biochemical treatment unit. S5. Based on the hydraulic affinity score, determine the allocation ratio for the flow distribution of the influent, and control the flow distribution equipment of the wastewater treatment plant according to the determined allocation ratio based on the predicted arrival time of the influent carrying the corresponding chemical characteristics at the wastewater treatment plant.

2. The intelligent control method for wastewater treatment process based on chemical feature identification according to claim 1, characterized in that, In step S3, the biochemical reaction potential vector includes at least one of the following potential energy components: Aerobic oxygen consumption rate potential, nitrification inhibition potential, sludge yield potential, or toxicity shock potential.

3. The intelligent control method for wastewater treatment process based on chemical feature identification according to claim 1, characterized in that, Also includes: After step S3 and before step S4, obtain sludge functional profile data collected by biomonitoring devices deployed in each return sludge source within the wastewater treatment plant. The sludge functional profile data is input into a preset sludge functional spectrum analysis model to obtain a sludge functional spectrum vector characterizing the functional characteristics of the returned sludge source.

4. The intelligent control method for wastewater treatment process based on chemical feature identification according to claim 3, characterized in that, The sludge functional spectrum vector includes at least one of the following functional components: Overall biological activity, tolerance to toxicity, or specificity in the degradation of recalcitrant organic matter.

5. The intelligent control method for wastewater treatment process based on chemical feature identification according to claim 3, characterized in that, After step S4 and before step S5, the following is also included: Based on the biochemical reaction potential vector and the sludge functional spectrum vector, a targeted gain affinity score is calculated for each combination of the biochemical treatment unit and each return sludge source. The targeted gain affinity score is used to characterize the expected gain effect of distributing the sludge from the return sludge source to the biochemical treatment unit in response to the influent.

6. The intelligent control method for wastewater treatment process based on chemical feature identification according to claim 5, characterized in that, Also includes: Based on the target gain affinity score, a target reflux strategy is determined; Based on the predicted arrival time of the influent carrying the corresponding chemical characteristics obtained in step S2, the wastewater treatment plant's return sludge distribution equipment is controlled according to the targeted return strategy to distribute the sludge from the designated return sludge source to the designated biochemical treatment unit.

7. The intelligent control method for wastewater treatment process based on chemical feature recognition according to claim 6, characterized in that, The steps for determining the allocation ratio for the influent flow and the targeted reflux strategy are as follows: Construct an optimization problem, where the allocation ratio and the targeted reflux strategy are the decision variables; Establish an objective function for the optimization problem, wherein the objective function is a weighted sum of the hydraulic affinity score and the target gain affinity score; By solving the optimization problem to maximize the objective function, the optimal allocation ratio and the targeted backflow strategy can be obtained simultaneously.

8. The intelligent control method for wastewater treatment process based on chemical feature identification according to claim 6, characterized in that, After step S5, for each of the biochemical processing units that has received allocated flow, the operating parameters of the biochemical processing unit are dynamically adjusted based on the biochemical reaction potential vector obtained in step S3 and in conjunction with the targeted backflow strategy formulated for the biochemical processing unit.

9. The intelligent control method for wastewater treatment process based on chemical feature identification according to claim 8, characterized in that, The operating parameters include aeration rate, internal reflux ratio, or dosage of the agent.

10. The intelligent control method for wastewater treatment process based on chemical feature identification according to claim 1, characterized in that, In step S1, the spectral sensor is an ultraviolet-visible full-spectrum analyzer or a three-dimensional fluorescence spectrometer.

Images