Intelligent control method and system for sewage treatment based on Gaussian function feature extraction

Abstract

The present invention belongs to the field of sewage treatment technology and provides a method and system for intelligent sewage treatment control based on Gaussian function feature extraction, aiming to optimize the sewage treatment process through a data-driven approach. The method includes the following steps: first, collecting time series data from an external reactor through a simulation simulator; then preprocessing the time series data; then establishing a substrate kinetic function for the sewage treatment plant's reaction tank; based on this kinetic function, combined with differential equations, establishing a data-driven model and calibrating and verifying it; further, dynamically controlling the sewage treatment plant's key control parameters to achieve minimum energy consumption and optimal effluent quality; finally, by updating the external reactor data and Gaussian parameters in real time, generating real-time control signals to optimize the operation of the sewage treatment plant. The method has adaptive capabilities and can optimize the sewage treatment process in real time, reducing energy consumption and improving treatment efficiency. It is particularly suitable for the intelligent control of AO denitrification processes.

Description

Technical Field

[0001] The present invention belongs to the technical field of sewage treatment, and in particular relates to an intelligent sewage treatment control method and system based on Gaussian function feature extraction. Background Art

[0002] In the current wastewater treatment process, achieving real-time dynamic online dynamic control is crucial to ensuring stable treatment and cost-effectiveness. Machine learning-based models can provide more precise data-based dynamic control. For traditional machine learning-based models, a large amount of real-time online monitoring data of water quality, mud quality and environmental variables needs to be trained to obtain more accurate predictions. However, collecting these monitoring data is expensive, requires more additional online monitoring equipment, and introduces high-cost construction and operation processes, which is not cost-effective for sewage treatment plants. Therefore, in real-time online dynamic control, how to use the least amount of monitoring equipment to achieve reliable dynamic predictions while realizing dynamic regulation of operating control parameters has become an urgent problem that needs to be solved urgently. Summary of the Invention

[0003] To solve the above technical problems, the present invention provides a sewage treatment intelligent control method and system based on Gaussian function feature extraction to solve the problems in the prior art. The technical solution adopted by the present invention is:

[0004] The intelligent control method for sewage treatment based on Gaussian function feature extraction includes:

[0005] Step 1. Use two external reactors as simulation simulators for the anoxic tank and aerobic tank of the sewage treatment plant to collect time series data. Simultaneously, collect time series data on the influent water quality and effluent quality of the sewage treatment plant to establish a database of the external reactor data set and the sewage treatment plant data set.

[0006] Step 2. Preprocessing the time series data of each indicator in each set of data in the external reactor data set, then calculating the second-order derivative of the time series data of each indicator in each set of data with respect to time to obtain second-order derivative time series data, fitting the second-order derivative time series data with a Gaussian function to obtain Gaussian parameters under different initial concentrations and aeration rates of each monitoring indicator of the external reactor, and establishing a characteristic parameter database including the initial values ​​of each monitoring indicator of the external reactor, the oxygen transfer coefficient, and the Gaussian parameters;

[0007] Step 3. Perform surface fitting on the initial values ​​and aeration rates of the two external reactors in the characteristic parameter database and the corresponding Gaussian parameters to obtain a Gaussian parameter polynomial. Substitute the Gaussian parameter polynomial into the Gaussian function and integrate the Gaussian function with time as the variable to obtain the substrate dynamics, which is the substrate dynamics of the sewage treatment plant reaction tank.

[0008] Step 4. Based on the obtained substrate kinetics of the sewage treatment plant reactor and the modeling method of continuous flow sewage treatment plant differential equations, a data-driven model is established. This model uses the sewage treatment plant influent water quality as the output to obtain a predicted value of the effluent water quality. The data-driven model is calibrated and validated based on the actual measured influent and effluent water quality.

[0009] Step 5. Combining a multi-objective optimization algorithm and a data-driven model, with the goal of minimizing energy consumption and effluent water quality, the sewage treatment plant's operating control parameters, oxygen transfer coefficient and nitrification solution return rate, are regulated to generate a dynamic control strategy.

[0010] Step 6. Update the polynomial of Gaussian parameters in real time using the newly collected time series data of the external reactor, update the data-driven model, and update the operation and control strategy in real time based on the updated data-driven model and multi-objective optimization algorithm. Further, convert the updated strategy into a control signal and transmit it to the aerator and nitrification liquid return pump of the sewage treatment plant.

[0011] Furthermore, in step 1, data from the reactor of the aerobic tank is collected once every hour, and data from the reactor of the anoxic tank is collected once every three hours. The collected data include BOD, ammonia nitrogen, total nitrogen and dissolved oxygen; BOD probes, ammonia nitrogen probes and total nitrogen probes are respectively arranged at the water inlet and outlet of the sewage treatment plant for continuous online monitoring, and the monitoring frequency is one data point every 10 seconds.

[0012] Furthermore, in step 2, the time series data of each indicator in each group of data are respectively calculated for the second-order derivative with respect to time. The second-order derivative is achieved by the second-order central difference method. The formula is as follows:

[0013]

[0014] Where x is the time point, f″(x) is the second-order derivative of the desired time point; Δx is the time interval between data points in each time series; f(x+Δx), f(x), and f(x-Δx) are the values ​​of the data points at time point x and at the positions of the Δx time intervals before and after it.

[0015] The second-order derivative data in step 2 will be fitted with a Gaussian function, and the formula is as follows:

[0016]

[0017] Where A is the amplitude of the Gaussian function, μ is the mean, and σ is the standard deviation.

[0018] Further, the approximate substrate kinetics of the sewage treatment plant reaction tank in step 3 is as follows:

[0019]

[0020] Where f′(x) i,j is the kinetic function of substrate j in reaction process i, γ is the correction parameter, HRT is the hydraulic retention time of reaction process i, A i,j , μ i,j , σ i,j are the parameters of the Gaussian function, where i represents the anoxic or aerobic process, and j represents the BOD, ammonia nitrogen or total nitrogen index.

[0021] Furthermore, in step 4, the predicted value of the sewage treatment plant influent quality is corrected by comparing the data-driven model with the experimental data.

[0022] Furthermore, the multi-objective optimization algorithm in step 5 adopts a multi-objective genetic algorithm.

[0023] The intelligent control system for sewage treatment based on Gaussian function feature extraction includes: an external reactor; the external reactor includes two, namely an anoxic reactor and an aerobic reactor;

[0024] The anoxic reactor and the aerobic reactor are both connected to a computer;

[0025] The water inlet of the anoxic reactor is connected to the water inlet of the anoxic tank to simulate the anoxic tank;

[0026] The water inlet end of the aerobic reactor is connected to the water inlet end of the aerobic tank to simulate the aerobic tank.

[0027] The present invention has the following beneficial effects:

[0028] 1. The reaction kinetics of anoxic and aerobic tanks in continuous flow sewage treatment plants can be quickly obtained through external reactors;

[0029] 2. By calculating the second-order derivative of the external reactor time series data and using the Gaussian function to extract characteristic parameters, the lengthy time series is converted into Gaussian parameters, significantly reducing the amount of data;

[0030] 3. The sewage treatment data-driven model based on Gaussian function feature extraction has high and fast computing efficiency;

[0031] 4. Update the data-driven model by updating the Gaussian parameter database. The parameter update rate is fast and the model is stable.

[0032] 5. Based on an efficient data-driven model computing platform, the update rate of the optimization control strategy is significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a flow chart of the method of the present invention;

[0034] Figure 2 Schematic diagram of the system structure of the present invention. DETAILED DESCRIPTION

[0035] The following is a combination of the embodiments of the present invention Figure 1-Figure 2 , the technical solutions in the embodiments of the present invention are described clearly and completely. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Unless otherwise specified, the technical means used in the embodiments are conventional means well known to those skilled in the art.

[0036] This invention aims to achieve reliable dynamic prediction of wastewater treatment processes and regulation of operational control parameters using only a small amount of monitoring equipment. To this end, the invention proposes a novel feature extraction technology that, based on online monitoring data, accurately extracts the kinetic characteristics of key biochemical processes in wastewater treatment. This technology enables dynamic intelligent control of both anoxic and aerobic denitrification processes, resulting in a more efficient and stable wastewater treatment process and reduced operating energy consumption and costs for wastewater treatment plants.

[0037] The intelligent control method for sewage treatment based on Gaussian function feature extraction includes:

[0038] Step 1: Use the two external reactors as the anoxic tank and aerobic tank of the sewage treatment plant respectively to simulate the simulator to collect time series data. At the same time, collect the time series data of the sewage treatment plant's influent water quality and quantity and effluent water quality, and then establish a database containing the external reactor data set and the sewage treatment plant data set;

[0039] The data set of the external reactor is established by using two external reactors as the anoxic tank and aerobic tank of the sewage treatment plant to simulate and collect time series data. Figure 2As shown, the sewage treatment intelligent control system based on Gaussian function feature extraction simulates the operation of the anoxic tank 1 of the sewage treatment plant through the external anoxic reactor 8, and uses the mixed liquid 15 at the water inlet end of the anoxic tank 1 as the inlet water of the external reactor. Every hour, the anoxic reactor 8 is completely emptied to the middle part 17 of the anoxic tank of the sewage treatment plant through the drainage pump 16 of the external anoxic reactor 8, and then the mixed liquid at the water inlet end of the anoxic tank is re-injected through the water inlet pump 14 of the external anoxic reactor 8. In this process, the agitator 11 of the external anoxic reactor 8 is kept in continuous operation, and the BOD probe 9 is used to measure the mixed liquid. , ammonia nitrogen probe 10, total nitrogen probe 12 and dissolved oxygen probe 13 are used for online monitoring. The monitoring frequency is one data point every 10 seconds, and the data signal is transmitted to the computer 26 through the ModBus serial port tool 25. Then, the data signal is converted into a data value through the data acquisition software in the computer 26 and recorded in the external reactor data set of the computer. The time series data within each residence time is regarded as a group of data. Therefore, a group of time data of the anoxic reactor 8 can be collected every 1 hour. The group of time series data includes the time data of BOD, ammonia nitrogen, total nitrogen and dissolved oxygen.

[0040] Similarly, the aerobic tank 4 of the sewage treatment plant is simulated by the external aerobic reactor 18, with the mixed liquid 21 at the water inlet end of the aerobic tank 4 being used as the inlet water. Every three hours, the reactor is completely emptied to the middle part 23 of the aerobic tank 4 of the sewage treatment plant through the drainage pump 22 of the external aerobic reactor 18, and then the mixed liquid at the water inlet end of the aerobic tank 4 is re-injected through the water inlet pump 20 of the external aerobic reactor 18; the aerobic reactor 18 is continuously aerated through the aeration disk 19 at the bottom of the reactor by the aeration pump 24 of the external aerobic reactor, and the air-water ratio of the aeration is kept consistent with that of the sewage treatment plant; then Online monitoring is performed through another set of BOD probes, ammonia nitrogen probes, total nitrogen probes and dissolved oxygen probes. The monitoring frequency is one data point every 10 seconds, and the data signal is transmitted to the computer 26 through the ModBus serial port tool 25. The data signal is then converted into a data value through the data acquisition software in the computer and recorded in the external reactor data set of the computer. The time series data within each residence time is regarded as a group of data. Therefore, a group of aerobic reactor data can be collected every 3 hours. The group of time series data includes the time data of BOD, ammonia nitrogen, total nitrogen and dissolved oxygen.

[0041] The data set of the sewage treatment plant is collected by arranging BOD probes, ammonia nitrogen probes, and total nitrogen probes at the water inlet and outlet of the sewage treatment plant for continuous online monitoring. The monitoring frequency is one data point every 10 seconds, and the data signal is transmitted to the computer 26 through the ModBus serial port tool 25. The data signal is then converted into data values ​​through the data acquisition software in the computer and recorded in the sewage treatment plant data set of the computer.

[0042] Step 2: preprocessing the time series data of each indicator of each group of data in the external reactor data set, and then calculating the second-order derivative of the time series data of each indicator in each group of data with respect to time to obtain second-order derivative time series data. The second-order derivative data is further fitted with a Gaussian function to obtain Gaussian parameters under different initial concentrations and aeration rates of each monitoring indicator of a large number of external reactors, and establishing a characteristic parameter database including the initial values ​​of each monitoring indicator of the external reactor, the oxygen transfer coefficient, and the Gaussian parameters;

[0043] The time series data for BOD, ammonia nitrogen, total nitrogen, and dissolved oxygen indicators in each data set of the external reactor data set collected by the computer were preprocessed. Since this time series data was obtained from online monitoring data, it may contain missing data and outliers, so data preprocessing was required. The quartile method was used to handle outliers and missing values ​​in the data. First, the data were sorted by numerical value and the first quartile (Q1), median (Q2), and third quartile (Q3) were calculated. Outliers were then identified based on the interquartile range (IQR = Q3 - Q1). Specifically, data points below Q1 - 1.5 × IQR or above Q3 + 1.5 × IQR were considered outliers. These outliers were handled by using median replacement. This quartile preprocessing method effectively improved the quality and reliability of the data, ensuring the accuracy of subsequent analysis and modeling.

[0044] The second-order derivative of the time series data of each indicator after preprocessing is calculated with respect to time. Since the time series data points of each indicator in each set of data are in the form of one data point every 10 seconds, the second-order derivative of the time series data of each indicator with respect to time is realized by the second-order central difference method. The formula is as follows:

[0045]

[0046] Where x is the time point, f″(x) is the second-order derivative at the desired time point, Δx is the time interval between data points in each time series, 10 seconds, and f(x+Δx), f(x), and f(x-Δx) are the values ​​of the data points at time point x and 10 seconds before and after it.

[0047] Through the above formula, the data of each indicator of each set of time series data in the external reactor data set is converted into a time series of the second-order derivative of each indicator with respect to time. Then, the time series of the second-order derivative of each indicator is fitted with a Gaussian function. The time series of the second-order derivative is approximately equal to the Gaussian function. The formula is as follows:

[0048]

[0049] Where A is the amplitude of the Gaussian function, μ is the mean, and σ is the standard deviation.

[0050] According to the above method, the time series of each indicator in each set of data corresponds to a Gaussian function, and therefore corresponds to a set of Gaussian function parameters A, μ and σ. And each set of data has a different initial concentration and aeration volume of each monitoring indicator of an external reactor. Therefore, all the data collected in the external reactor data set are extracted using the above Gaussian function method to extract the Gaussian function parameters, and a large number of different initial concentrations and aeration volume parameters of each monitoring indicator of the external reactor can be obtained - oxygen transfer parameter (K L a) Gaussian parameters, including the initial values ​​of BOD, ammonia nitrogen, and total nitrogen, K L The data of a, A, μ and σ constitute a characteristic parameter database.

[0051] Step 3: Perform surface fitting on a large number of initial values ​​and aeration rates of the two external reactors in the characteristic parameter database and the corresponding Gaussian parameters to obtain a polynomial of the Gaussian parameters. The Gaussian parameter polynomial is further substituted into the Gaussian function, and the Gaussian function with time as the variable is integrated once to obtain the substrate dynamics, which is approximately equal to the substrate dynamics of the sewage treatment plant reaction tank.

[0052] A large number of initial values ​​and aeration rates of the two external reactors are individually fitted with each corresponding Gaussian parameter to obtain a polynomial of the Gaussian parameter. The specific fitting formula is as follows:

[0053] A i,j =a1+a2·x+a3·y+a4·x·y+a5·x 2 +a6·y 2 (3)

[0054] μ i,j =b1+b2·x+b3·y+b4·x·y+b5·x 2 +b6·y 2 (4)

[0055] σ i,j =c1+c2·x+c3·y+c4·x·y+c5·x 2 +c6·y 2 (5)

[0056] Where A i,j , μ i,j , σ i,jare the parameters of the Gaussian function, where i represents the anoxic or aerobic process, j represents the BOD, ammonia nitrogen or total nitrogen index; x and y represent any two indicators of the initial values ​​of the BOD, ammonia nitrogen, total nitrogen or the aeration volume of the external reactor; a1~a6, b1~b6, c1~c6 represent the parameters of the polynomial of the surface fitting.

[0057] Substituting the polynomial of the above Gaussian parameters into the Gaussian function, we get the following formula:

[0058]

[0059] Integrating the above results once we can get

[0060]

[0061] Where f′(x) i,j is the kinetic function of substrate j in reaction process i, γ is the correction parameter, HRT is the hydraulic retention time of reaction process i, A i,j , μ i,j , σ i,j are the parameters of the Gaussian function, where i represents the anoxic or aerobic process, and j represents the BOD, ammonia nitrogen or total nitrogen index.

[0062] Step 4: Based on the obtained substrate kinetics of the sewage treatment plant reaction tank and the modeling method of the continuous flow sewage treatment plant differential equation, a data-driven model is established. The model uses the sewage treatment plant influent water quality as the output to obtain the predicted value of the effluent water quality. The data-driven model is further calibrated and verified based on the actual measured influent and effluent water quality.

[0063] Specifically, based on the obtained substrate kinetics of the external reactor and combined with the modeling method of the differential equation of the continuous flow sewage treatment plant, the substrate concentration accumulation rate of the reaction tank of the sewage treatment plant is obtained as follows:

[0064]

[0065] Where S j is the concentration of substrate j in the reaction pool, S j,inf is the influent water quality of substrate j, and HRT is the hydraulic retention time of reaction process i.

[0066] A data-driven model was established based on the substrate concentration accumulation rate described above, and then solved using an ODE solver. The differential equations of the data-driven model were solved using odeint from the Python scipy library. A loss function was then constructed by minimizing the mean squared error (MSE) between the actual sewage treatment plant data and the predicted values ​​of the data-driven model. The loss function is:

[0067]

[0068] Where s i is the predicted value of the data-driven model, is the corresponding true value.

[0069] Then solve the minimized MSE to train the model correction parameter γ, the formula is as follows:

[0070]

[0071] A genetic algorithm is used to solve the above minimization process, specifically the GeneticAlgorithm function in the Playtypus library in Python is used for the solution.

[0072] Step 5: Combining a multi-objective optimization algorithm and a data-driven model, with energy consumption and effluent water quality concentration as the control objectives, the sewage treatment plant's operating control parameters, oxygen transfer coefficient and nitrification solution return rate, are regulated to generate a dynamic control strategy;

[0073] Specifically, the data-driven model is used as the computing platform, and a multi-objective genetic algorithm is used to calculate the oxygen transfer coefficient (K L a) and nitrification liquid return rate (NNR) are optimized to achieve the best combination of energy consumption and water quality, and to achieve real-time adjustment of sewage treatment plant operation. The multi-objective optimization is shown in the following formula:

[0074]

[0075] Where W r,o is the energy consumption of aeration and nitrification liquid return, S BOD is the BOD concentration, is the ammonia nitrogen concentration, S TN is the total nitrogen concentration.

[0076] W r,o =0.012·K L a+0.003992592·Q IN ·NNR (12)

[0077] Where Q ID is the water inlet flow rate.

[0078] The multi-objective optimization algorithm is the NSGAII algorithm in the platypus library in Python.

[0079] In step 6, the polynomial of the Gaussian parameters is updated in real time by using the newly collected external reactor time series data, thereby updating the data-driven model. The operation control strategy is updated in real time based on the updated data-driven model and the multi-objective optimization algorithm, and the updated strategy is further converted into a control signal and transmitted to the aerator and nitrification liquid return pump of the sewage treatment plant.

[0080] The specific operation is to extract the features of a set of newly collected time series data of the reactor to obtain a set of feature parameter data including the initial value of the external reactor, the aeration volume and the corresponding Gaussian function parameters, and add this set of data to the previously extracted feature parameter database; then re-surface fit the feature parameter data in this feature parameter database, update the coefficients a1~a6, b1~b6 and c1~c6 of the Gaussian parameter polynomial function, and then bring the updated Gaussian parameter polynomial into formula (8), and re-correct the data-driven model based on the newly collected sewage treatment plant inlet and outlet water data; further, update the operation control strategy in real time based on the updated data-driven model and multi-objective optimization algorithm, and finally convert the real-time value of the updated strategy into a control signal and transmit it to the aerator and return pump of the sewage treatment plant, as well as the aerator and each pump of the external reactor. Figure 2 As shown, the electrical signal generated by the computer 26 is transmitted to the aerator 5 and nitrification liquid return pump 3 of the sewage treatment plant through the ModBus serial port tool 25. All aerators and return pumps are equipped with controllable drive hardware, and the signals transmitted through ModBus are used to control the start, stop, speed adjustment, etc. of the equipment.

[0081] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various deformations, modifications, and substitutions made to the technical solutions of the present invention by ordinary technicians in this field should fall within the scope of protection determined by the claims of the present invention.

Claims

1. An intelligent control method for sewage treatment based on Gaussian function feature extraction, characterized in that: include: Step 1. Use two external reactors as simulation simulators for the anoxic tank and aerobic tank of the sewage treatment plant to collect time series data. Simultaneously, collect time series data on the influent water quality and effluent quality of the sewage treatment plant to establish a database of the external reactor data set and the sewage treatment plant data set. Step 2. Preprocessing the time series data of each indicator in each set of data in the external reactor data set, then calculating the second-order derivative of the time series data of each indicator in each set of data with respect to time to obtain second-order derivative time series data, fitting the second-order derivative time series data with a Gaussian function to obtain Gaussian parameters under different initial concentrations and aeration rates of each monitoring indicator of the external reactor, and establishing a characteristic parameter database including the initial values ​​of each monitoring indicator of the external reactor, the oxygen transfer coefficient, and the Gaussian parameters; Step 3. Perform surface fitting on the initial values ​​and aeration rates of the two external reactors in the characteristic parameter database and the corresponding Gaussian parameters to obtain a Gaussian parameter polynomial. Substitute the Gaussian parameter polynomial into the Gaussian function and integrate the Gaussian function with time as the variable to obtain the substrate dynamics, which is the substrate dynamics of the sewage treatment plant reaction tank. Step 4. Based on the obtained substrate kinetics of the sewage treatment plant reactor and the modeling method of continuous flow sewage treatment plant differential equations, a data-driven model is established. This model uses the sewage treatment plant influent water quality as the output to obtain a predicted value of the effluent water quality. The data-driven model is calibrated and validated based on the actual measured influent and effluent water quality. Specifically, based on the obtained substrate kinetics of the external reactor and combined with the modeling method of the differential equation of the continuous flow sewage treatment plant, the substrate concentration accumulation rate of the reaction tank of the sewage treatment plant is obtained as follows: Where S j is the concentration of substrate j in the reaction pool, S j,inf is the influent water quality of substrate j, HRT is the hydraulic retention time of reaction process i; Step 5. Combining a multi-objective optimization algorithm and a data-driven model, with the goal of minimizing energy consumption and effluent water quality, the sewage treatment plant's operating control parameters, oxygen transfer coefficient and nitrification solution return rate, are regulated to generate a dynamic control strategy. Step 6. Using the newly collected time series data from the external reactor, the polynomial of the Gaussian parameters is updated in real time, and the data-driven model is updated. Based on the updated data-driven model and the multi-objective optimization algorithm, the operation control strategy is updated in real time. The updated strategy is further converted into control signals and transmitted to the aerator and nitrification liquid return pump of the sewage treatment plant; The approximate substrate kinetics of the sewage treatment plant reaction tank in step 3 is as follows: Where f'(x) i,j is the kinetic function of substrate j in reaction process i, γ is the correction parameter, HRT is the hydraulic retention time of reaction process i, A i,j , μ i,j , σ i,j are the parameters of the Gaussian function, where i represents the anoxic or aerobic process, and j represents the BOD, ammonia nitrogen or total nitrogen index.

2. The intelligent control method for sewage treatment based on Gaussian function feature extraction according to claim 1 is characterized in that: In step 1, data from the reactor of the aerobic tank is collected once every hour, and data from the reactor of the anoxic tank is collected once every three hours. The collected data include BOD, ammonia nitrogen, total nitrogen and dissolved oxygen; BOD probes, ammonia nitrogen probes and total nitrogen probes are respectively arranged at the water inlet and outlet of the sewage treatment plant for continuous online monitoring, and the monitoring frequency is one data point every 10 seconds.

3. The intelligent control method for sewage treatment based on Gaussian function feature extraction according to claim 1 is characterized in that: In step 2, the time series data of each indicator in each group of data are respectively calculated with respect to time. The second-order derivative is achieved by the second-order central difference method. The formula is as follows: Where x is the time point, f”(x) is the second-order derivative at the time point being sought; Δx is the time interval between data points in each time series; f(x+Δx), f(x), and f(x-Δx) are the values ​​of the data points at time point x and at the positions of the Δx time intervals before and after it. The second-order derivative data in step 2 will be fitted with a Gaussian function, and the formula is as follows: Where A is the amplitude of the Gaussian function, μ is the mean, and σ is the standard deviation.

4. The intelligent control method for sewage treatment based on Gaussian function feature extraction according to claim 1 is characterized in that: In step 4, the predicted value of the sewage treatment plant influent quality is corrected by comparing the data-driven model with the experimental data.

5. The intelligent control method for sewage treatment based on Gaussian function feature extraction according to claim 1 is characterized in that: The multi-objective optimization algorithm in step 5 adopts a multi-objective genetic algorithm.

6. An intelligent sewage treatment control system based on Gaussian function feature extraction, which adopts the intelligent sewage treatment control method based on Gaussian function feature extraction according to claim 1, characterized in that: include: External reactor; the external reactor includes two, namely an anoxic reactor (8) and an aerobic reactor (18); The anoxic reactor (8) and the aerobic reactor (18) are both connected to a computer (26); The water inlet of the anoxic reactor (8) is connected to the water inlet of the anoxic tank (1) to simulate the anoxic tank (1); The water inlet end of the aerobic reactor (18) is connected to the water inlet end of the aerobic tank (4) to simulate the aerobic tank (4).

Images