An intelligent control method for sewage treatment aeration based on environmental prior knowledge

Abstract

The present application relates to the technical field of sewage treatment, and particularly relates to a sewage treatment aeration intelligent control method based on environmental prior knowledge. According to water quality index monitoring data set obtained from water quality index monitoring in the process of influent monitoring, process monitoring and effluent monitoring, equipment working condition monitoring data set is obtained through equipment data collection of sewage treatment equipment; the water quality index monitoring data set and the equipment working condition monitoring data set are fused to expand into a fusion data set with environmental prior knowledge according to the environmental prior knowledge; the fusion data set with environmental prior knowledge is input into an adjustable parameter long short-term memory neural network machine learning model, and oxygen provided by a blower aeration is output according to the adjustable parameter adaptive long short-term memory neural network model, so that the aeration treatment process is controlled. The method realizes intelligent control of the sewage treatment aeration process, reduces energy consumption, improves the rationality of the aeration oxygen supply, and plays a positive role in sewage discharge up to standard.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sewage treatment, and in particular to a sewage treatment aeration intelligent control method based on environmental prior knowledge. BACKGROUND

[0002] Industrial wastewater and domestic wastewater discharged into natural water bodies without treatment will seriously pollute natural water bodies, leading to water quality deterioration, biodiversity reduction, and threats to human health; at the same time, harmful substances in these wastewater may penetrate into groundwater, further affecting drinking water safety.

[0003] Sewage biochemical treatment can effectively degrade organic pollutants, reduce harmful substances, improve water quality, protect the ecological environment, and ensure the safety of discharged water bodies, which is a key link in sewage treatment; the aeration process is crucial in the sewage biochemical treatment process, which promotes microbial activity by increasing oxygen supply, effectively decomposes organic pollutants, improves treatment efficiency, and ensures that the effluent water quality meets the standards.

[0004] At present, the control of the aeration stage in the sewage biochemical treatment process in China mainly focuses on manual fuzzy control, which is based on artificial experience and speculates whether the aeration amount is insufficient or excessive according to the changes in key water quality indicators, manually increases or reduces the aeration amount of the blower, and through manual fuzzy control, the water quality changes can be responded to a certain extent to maintain stable effluent water quality; but at the same time, there are slow response speed, low precision, strong artificial dependence, and other shortcomings, the operation is complex, the treatment efficiency is unstable, and it is easy to increase energy consumption, management and maintenance costs, and it is difficult to achieve the best sewage treatment effect.

[0005] In summary, the existing technology has the technical problem that the aeration stage relies on manual fuzzy control, which cannot ensure that the aeration amount can accurately meet the biochemical process, and there is a problem of environmental pollution caused by excessive discharge of sewage; therefore, the present application proposes an intelligent control method for sewage treatment aeration to improve the above-mentioned technical problems in the prior art. SUMMARY

[0006] The purpose of the present application is to provide an intelligent control method for sewage treatment aeration based on environmental prior knowledge to solve the problem of slow response speed, strong artificial dependence, and low control precision caused by the large reliance on manual fuzzy control in the existing aeration control process of sewage treatment plants.

[0007] To achieve the above-mentioned purpose, the present application adopts the following technical solution: an intelligent control method for sewage treatment aeration based on environmental prior knowledge, characterized in that it comprises the following steps:

[0008] S100, obtaining water quality index monitoring data set according to water quality index in the process of influent monitoring, process monitoring and effluent monitoring;

[0009] S200, collecting equipment data according to the sewage treatment equipment to obtain an equipment working condition monitoring data set;

[0010] S300, fusing the water quality monitoring data set and the equipment working condition monitoring data set, and expanding the fused data set with environmental priori knowledge according to environmental priori knowledge;

[0011] S400, building an adjustable parameter long short-term memory neural network adaptive model, wherein the adjustable parameter adaptive model is embedded in a control system terminal of the sewage treatment blower equipment;

[0012] S500, inputting the fused data into the adjustable parameter long short-term memory neural network adaptive model, and controlling the sewage treatment blower equipment according to the long short-term memory neural network and the adaptive parameters.

[0013] Further, in the S100, the following steps are included:

[0014] SA100, monitoring the water quality of the water sample collected through the influent sampling pipe according to the data acquisition device to obtain an influent monitoring data set;

[0015] SB100, monitoring the water quality of the water sample collected through the biochemical process section sampling pipe according to the data acquisition device to obtain a biochemical process monitoring data set;

[0016] SC100, monitoring the water quality of the water sample collected through the effluent sampling pipe according to the data acquisition device to obtain a historical effluent monitoring data set.

[0017] Further, in the SA100-SC100, the sampling time of the influent monitoring data and the biochemical process monitoring data is the current time t, and the sampling time of the historical effluent monitoring data is the previous time t-1; the data acquisition device includes multiple types, each type of data acquisition device includes at least two water quality monitors, and the average value is used to avoid accidental errors; the data in the data set is preprocessed, missing values are filled, and abnormal values are deleted.

[0018] Further, the equipment working condition monitoring data set includes a digital control monitoring data set and a real-time control monitoring data set; the data types in the equipment working condition monitoring data set include sewage anaerobic hydraulic retention time data, anoxic hydraulic retention time data, aerobic hydraulic retention time data, sludge transportation time, and advanced treatment chemical addition time.

[0019] Further, in the S300, the following steps are included:

[0020] S310, align the data rows and data columns of the water quality index monitoring data set in S100 and the equipment working condition monitoring data set in S200, and set the data interval to the minimum sampling time;

[0021] S320, according to the removal rate and the removal percentage of the environmental prior knowledge, construct the difference and percentage difference between the influent water quality and the biochemical process section water quality, the influent water quality and the historical effluent water quality, and the biochemical process section water quality and the historical effluent water quality;

[0022] S330, according to the average value of the environmental prior knowledge, construct the multi-day average value of the influent water quality, the biochemical process section water quality and the historical effluent water quality;

[0023] S340, according to the environmental prior knowledge of the contrast value, construct the multiple variable contrast values of the influent water quality, the biochemical process section water quality and the historical effluent water quality;

[0024] S350, according to the environmental prior knowledge of the difference value, the difference percentage, the average value and the contrast value, data fusion is carried out on the constructed features to obtain the fusion data of the sewage treatment plant which can be used for model building.

[0025] Further, in the S400, the adjustable parameter long short-term memory neural network controls the information flow through the forget gate, the input gate and the output gate; the forget gate decides how much memory of the previous moment is discarded, the input gate decides the storage of the current input information, and the output gate decides the output state at the current moment; each gate has a corresponding weight matrix and a bias parameter;

[0026] The construction method of the adjustable parameter long short-term memory neural network is: the error at the time step is calculated and back propagated to adjust the weights and biases of the network; during the training process, the input sequence data is calculated by forward propagation to calculate the output, calculate the loss function, and then calculate the gradient of each time step by the back propagation algorithm, and continuously update the parameters of the long short-term memory neural network until the percentage error of the model is less than 5%, and output the temporary adjustable parameters;

[0027] The adjustable parameter adaptive model includes the number of memory cells, the number of neural network hidden layers, and the dimension of neural network input data.

[0028] Further, the adjustable parameter long short-term memory neural network mathematical model is as follows:

[0029]

[0030]

[0031]

[0032]

[0033]

[0034]

[0035]

[0036] wherein:

[0037] is an activation function;

[0038] is an input value, including water quality monitoring data set independent variables, biochemical process section water quality monitoring data set independent variables, historical effluent water quality monitoring data set independent variables, equipment working condition monitoring data set independent variables and environmental prior constructed water quality characteristic independent variables;

[0039] is a forget gate;

[0040] is a weight matrix of the forget gate, applied to the current input ;

[0041] is a weight matrix of the forget gate, applied to the hidden state of the previous moment ;

[0042] is a bias term of the forget gate;

[0043] is an input gate;

[0044] is a weight matrix of the input gate, applied to the current input ;

[0045] is a weight matrix of the input gate, applied to the hidden state of the previous moment ;

[0046] is a bias term of the input gate;

[0047] is a hyperbolic tangent activation function;

[0048] is a memory cell state update;

[0049] is a weight matrix of the memory cell state, applied to the current input and the hidden state of the previous moment ;

[0050] is the bias term of the memory cell state;

[0051] is the output gate, for the air volume of the air blower;

[0052] is the weight matrix of the output gate, applied to the current input ;

[0053] is the weight matrix of the output gate, applied to the hidden state of the previous moment ;

[0054] is the bias term of the output gate;

[0055] is the hidden state.

[0056] Further, 20% of the fusion data in S300 is used for pre-training, the fusion data is used as an input variable, a long short-term memory neural network is trained, and temporary adjustable parameters are obtained.

[0057] Further, in S500, the following steps are included:

[0058] S510, set the parameters of the long short-term memory neural network to the temporary adjustable parameters, input 100% of the fusion data into the long short-term memory neural network, and the model outputs the recommended air volume of the air blower;

[0059] S520, the control module adjusts the direction and size of the guide vane according to the recommended air volume of the air blower, thereby affecting the air volume, and realizes precise control;

[0060] S530, according to the water quality monitoring of the water sample collected through the real-time effluent sampling pipe by the data acquisition device, the effluent water quality at time t is obtained, and the effluent monitoring data set is obtained;

[0061] S540, judge whether the water quality index meets the standard, if the water pollution index meets the standard, obtain the sewage discharge instruction.

[0062] The beneficial effects of the present application are:

[0063] The method has the beneficial effects of energy consumption reduction and process simplification; solves the technical problems in the prior art that the process parameter setting of the sewage biochemical treatment depends on artificial fuzzy control, which leads to insufficient or excessive oxygen supply of the blower in the sewage treatment aeration process, and the sewage treatment biochemical process cannot ensure effective biochemical treatment of the sewage due to insufficient or excessive aeration, realizes the intelligent control of the sewage biochemical treatment aeration process, improves the rationality of the oxygen supply in the aeration process, the scientificity of the sewage biochemical treatment process parameter setting, the effectiveness of the sewage biochemical treatment, and the technical effect of ensuring that the sewage discharge meets the specified index requirements. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 is a flowchart of the method of the present application;

[0065] Figure 2 is a schematic diagram of the sampling position and time involved in the method of the present application;

[0066] Figure 3 is a flowchart of the construction of the environmental prior knowledge feature of the method of the present application;

[0067] Figure 4 is a structural schematic diagram of the control system of the method of the present application. DETAILED DESCRIPTION

[0068] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0069] EMBODIMENT

[0070] As shown in Figure 1 , a sewage treatment aeration intelligent control method based on environmental prior knowledge comprises the following steps:

[0071] S100, obtaining water quality index monitoring data set according to water quality index in the process of influent monitoring, process monitoring and effluent monitoring;

[0072] SA100, water quality monitoring is performed on the water sample collected via the influent sampling pipe by the data acquisition device to obtain the influent monitoring data set, wherein the data acquisition device includes multiple types, and each type of data acquisition device includes at least two water quality monitors;

[0073] SA110, obtaining past influent sampling data set of the target sewage treatment plant;

[0074] SA120, performing data preprocessing, missing value filling and outlier deletion according to the past influent sampling data set in SA110;

[0075] SB130, average two or more data, and configure parameters of the water quality monitor according to fluctuation characteristics of water quality indexes.

[0076] SB100, acquire the biochemical process monitoring data set by monitoring water quality of water samples collected via the biochemical process section sampling pipe by the data acquisition device, wherein the data acquisition device includes multiple types, and each type of data acquisition device includes at least two water quality monitors.

[0077] SB110, acquire the past process biochemical section sampling data set of the target sewage treatment plant.

[0078] SB120, perform data preprocessing, missing value filling, and outlier deletion according to the past process biochemical section sampling data set in SB110.

[0079] SB130, average two or more data, and configure parameters of the water quality monitor according to fluctuation characteristics of water quality indexes.

[0080] SC100, acquire the historical effluent monitoring data set by monitoring water quality of water samples collected via the effluent sampling pipe by the data acquisition device, wherein the data acquisition device includes multiple types, and each type of data acquisition device includes at least two water quality monitors.

[0081] SC110, acquire the past historical effluent sampling data set of the target sewage treatment plant.

[0082] SC120, perform data preprocessing, missing value filling, and outlier deletion according to the past historical effluent sampling data set in SC110.

[0083] SC130, average two or more data, and configure parameters of the water quality monitor according to fluctuation characteristics of water quality indexes.

[0084] As shown in Figure 2 the sampling time of the influent monitoring data and the biochemical process monitoring data is the current time t, and the sampling time of the historical effluent water quality monitoring data is the historical last time t-1.

[0085] Specifically, in this embodiment, the target sewage treatment plant is a professional sewage treatment institution that adopts a biochemical treatment method for sewage treatment. The sewage undergoes sewage biochemical treatment processes such as sewage influent adjustment, effluent adjustment, anaerobic reaction, anoxic reaction, aerobic reaction, sludge treatment and disposal, advanced treatment, and flocculation and sedimentation in the sewage treatment equipment of the target sewage treatment plant.

[0086] In this embodiment, various data acquisition devices are involved, each performing its own function to monitor different pollution indicators. For example, a flow meter measures water volume, a temperature meter measures temperature, a COD meter measures COD, a total nitrogen meter measures total nitrogen content in wastewater, an ammonia nitrogen meter measures ammonia nitrogen content in wastewater, and a total phosphorus meter measures total phosphorus in wastewater. Each data acquisition device includes at least two water quality analyzers, using an average value to avoid random errors.

[0087] S200. By collecting equipment data from the sewage treatment equipment, a data set of equipment operating condition monitoring data is obtained.

[0088] Specifically, in this embodiment, wastewater undergoes various biological treatment processes within the wastewater treatment equipment of the target wastewater treatment plant, including wastewater influent regulation, effluent regulation, anaerobic reaction, anoxic reaction, aerobic reaction, sludge treatment and disposal, advanced treatment, and flocculation sedimentation. Based on the control system, the anaerobic hydraulic retention time, anoxic hydraulic retention time, aerobic hydraulic retention time, sludge removal time, and advanced treatment agent addition time are adjusted and controlled to establish the anaerobic, anoxic, aerobic, sludge discharge, and normal advanced treatment conditions of the wastewater treatment equipment, so as to effectively carry out the biological treatment process of wastewater.

[0089] Data is collected from the control system used to control the sewage treatment equipment, and the actual operating data of the sewage treatment equipment is collected to obtain the equipment condition monitoring dataset. The data types in the equipment condition monitoring dataset include anaerobic hydraulic retention time data, anoxic hydraulic retention time data, aerobic hydraulic retention time data, sludge transportation time, and deep treatment agent addition time.

[0090] The equipment operating condition monitoring dataset includes a digital control monitoring dataset and a real-time control monitoring dataset. The digital control monitoring dataset contains control parameters for anaerobic hydraulic retention time (HHD), anoxic hydraulic retention time (HHD), aerobic hydraulic retention time (HHD), sludge removal time, and advanced treatment reagent addition time, all obtained from data acquisition of the control system. The real-time control monitoring dataset contains real-time parameters for anaerobic hydraulic retention time (HHD), anoxic hydraulic retention time (HHD), aerobic hydraulic retention time (HHD), sedimentation time, drainage time, and blower aeration rate, all obtained from data acquisition of the wastewater treatment equipment.

[0091] S300, such as Figure 3 As shown, by fusing influent water quality indicators, process biochemical water quality indicators, historical effluent water quality indicators, and equipment operating condition monitoring datasets, environmental prior knowledge features are constructed; data fusion is completed by constructing removal amount, removal rate, average value, and comparison value, respectively.

[0092] S310, align the influent water quality index, process biochemical section water quality index, historical effluent water quality index and equipment working condition monitoring data set in data row and data column, and set the data interval to the minimum sampling time;

[0093] S320, according to the environmental prior knowledge of removal rate and removal percentage, the difference and percentage difference between influent water quality and biochemical process section water quality, influent water quality and historical effluent water quality, biochemical process section water quality and historical effluent water quality are constructed;

[0094] Specifically, the difference and percentage difference between influent and process biochemical section include: water quantity change difference between influent and process biochemical section, temperature change difference between influent and process biochemical section, COD change difference between influent and process biochemical section, total nitrogen change difference between influent and process biochemical section, ammonia nitrogen change difference between influent and process biochemical section, total phosphorus change difference between influent and process biochemical section. Water quantity change difference percentage between influent and process biochemical section, temperature change difference percentage between influent and process biochemical section, COD change difference percentage between influent and process biochemical section, total nitrogen change difference percentage between influent and process biochemical section, ammonia nitrogen change difference percentage between influent and process biochemical section, total phosphorus change difference percentage between influent and process biochemical section.

[0095] The difference and percentage difference between process biochemical section and historical effluent include: water quantity change difference between process biochemical section and historical effluent, temperature change difference between process biochemical section and historical effluent, COD change difference between process biochemical section and historical effluent, total nitrogen change difference between process biochemical section and historical effluent, ammonia nitrogen change difference between process biochemical section and historical effluent, total phosphorus change difference between process biochemical section and historical effluent. Water quantity change difference percentage between process biochemical section and historical effluent, temperature change difference percentage between process biochemical section and historical effluent, COD change difference percentage between process biochemical section and historical effluent, total nitrogen change difference percentage between process biochemical section and historical effluent, ammonia nitrogen change difference percentage between process biochemical section and historical effluent, total phosphorus change difference percentage between process biochemical section and historical effluent.

[0096] The difference and percentage difference between influent and historical effluent include: water quantity change difference between influent and historical effluent, temperature change difference between influent and historical effluent, COD change difference between influent and historical effluent, total nitrogen change difference between influent and historical effluent, ammonia nitrogen change difference between influent and historical effluent, total phosphorus change difference between influent and historical effluent. Water quantity change difference percentage between influent and historical effluent, temperature change difference percentage between influent and historical effluent, COD change difference percentage between influent and historical effluent, total nitrogen change difference percentage between influent and historical effluent, ammonia nitrogen change difference percentage between influent and historical effluent, total phosphorus change difference percentage between influent and historical effluent.

[0097] S330, according to the average value of the environmental prior knowledge, the multi-day average value of the influent water quality, the biochemical process section water quality and the historical effluent water quality is constructed, including 1 day, 2 days, 3 days, 5 days, 10 days, 30 days multi-day average value;

[0098] Specifically, the multi-day average value of the influent, the biochemical process section, and the effluent single position includes the following characteristics: the multi-day average value of the water quantity of the influent, the biochemical process section, and the effluent; the multi-day average value of the temperature of the influent, the biochemical process section, and the effluent; the multi-day average value of the COD of the influent, the biochemical process section, and the effluent; the multi-day average value of the total nitrogen of the influent, the biochemical process section, and the effluent; the multi-day average value of the ammonia nitrogen of the influent, the biochemical process section, and the effluent; the multi-day average value of the total phosphorus of the influent, the biochemical process section, and the effluent.

[0099] The multi-day average value of the difference and the percentage difference between the influent water quality and the biochemical process section water quality, the influent water quality and the historical effluent water quality, and the biochemical process section water quality and the historical effluent water quality includes the following characteristics: the multi-day average value of all the constructed characteristics of S320.

[0100] S340, according to the environmental prior knowledge of the comparison value, the multi-variable comparison value of the influent water quality, the biochemical process section water quality and the historical effluent water quality is constructed;

[0101] Specifically, the comparison value of the influent and the biochemical process section includes the following: the COD comparison value of the influent and the biochemical process section, the total nitrogen comparison value of the influent and the biochemical process section, the ammonia nitrogen comparison value of the influent and the biochemical process section, and the total phosphorus comparison value of the influent and the biochemical process section.

[0102] The comparison value of the biochemical process section and the historical effluent includes the following: the COD comparison value of the biochemical process section and the historical effluent, the total nitrogen comparison value of the biochemical process section and the historical effluent, the ammonia nitrogen comparison value of the biochemical process section and the historical effluent, and the total phosphorus comparison value of the biochemical process section and the historical effluent.

[0103] The comparison value of the influent and the historical effluent includes the following: the COD comparison value of the influent and the historical effluent, the total nitrogen comparison value of the influent and the historical effluent, the ammonia nitrogen comparison value of the influent and the historical effluent, and the total phosphorus comparison value of the influent and the historical effluent.

[0104] S350, according to the environmental prior knowledge of the difference value, the difference percentage, the average value and the comparison value, the constructed characteristics are data fused to obtain the fusion data of the sewage treatment plant which can be used for model building.

[0105] S400, a long short-term memory neural network adaptive model with adjustable parameters is built, wherein the long short-term memory neural network adaptive model with adjustable parameters is embedded in the control system terminal of the sewage treatment blower equipment;

[0106] S410. Construct a long short-term memory neural network with adjustable parameters. The adjustable parameter adaptive model includes the number of memory cells, the number of hidden layers in the neural network, and the dimension of the input data of the neural network.

[0107] S420. Use 20% of the fused data for pre-training, and use the fused data as input variables to train the long short-term memory neural network and obtain temporary adjustable parameters.

[0108] The mathematical model formula for the constructed Long Short-Term Memory neural network is as follows:

[0109]

[0110]

[0111]

[0112]

[0113]

[0114]

[0115]

[0116] in:

[0117] yes Activation function;

[0118] The input values ​​include independent variables from the influent water quality monitoring dataset, independent variables from the biochemical process water quality monitoring dataset, independent variables from the historical effluent water quality monitoring dataset, independent variables from the equipment operating condition monitoring dataset, and independent variables from water quality characteristics constructed prior to the environment.

[0119] It is the Gate of Oblivion;

[0120] It is the weight matrix of the forget gate, applied to the current input. ;

[0121] It is the weight matrix of the forget gate, applied to the hidden state of the previous time step. ;

[0122] It is the bias term of the forget gate;

[0123] It is an input gate;

[0124] is the weight matrix of the input gate, applied to the current input ;

[0125] is the weight matrix of the input gate, applied to the previous time step of the hidden state ;

[0126] is the bias term of the input gate

[0127] is the hyperbolic tangent activation function

[0128] is the memory cell state update

[0129] is the weight matrix of the memory cell state, applied to the current input and the previous time step of the hidden state ;

[0130] is the bias term of the memory cell state

[0131] is the output gate, for the blower aeration amount

[0132] is the weight matrix of the output gate, applied to the current input ;

[0133] is the weight matrix of the output gate, applied to the previous time step of the hidden state ;

[0134] is the bias term of the output gate

[0135] is the hidden state

[0136] The adjustable parameter long short-term memory neural network controls the flow of information through the forget gate, the input gate, and the output gate. The forget gate determines how much memory from the previous time step to discard, the input gate determines the storage of current input information, and the output gate determines the output state at the current time step. Each gate has a corresponding weight matrix and bias parameter.

[0137] The construction method of the adjustable parameter long short-term memory neural network is to calculate the error at the time step and back-propagate to adjust the weights and biases of the network. During training, the input sequence data is calculated through forward propagation to calculate the output, calculate the loss function, and then calculate the gradient of each time step step by step through the back propagation algorithm, and continuously update the parameters of the long short-term memory neural network until the model percentage error is less than 5%, and output the temporary adjustable parameters.

[0138] The wastewater treatment aeration intelligent control system module based on environmental prior knowledge is embedded in the air-blowing aerator as shown in Figure 4

[0139] S500, the fusion data is input into the adjustable parameter long short term memory network model, and the wastewater treatment air-blowing machine equipment is controlled according to the long short term memory neural network and the adaptive parameters;

[0140] S510, the parameters of the long short term memory neural network are set as temporary adjustable parameters, 100% of the fusion data is input into the long short term memory neural network, and the model outputs the air-blower recommended aeration amount;

[0141] S520, the control module adjusts the direction and size of the guide vane according to the air-blower recommended aeration amount, thereby affecting the aeration amount, and precise control is realized;

[0142] S530, the water quality monitoring is performed on the water sample collected through the real-time effluent sampling pipe according to the data acquisition device, the effluent water quality at the moment t is acquired, and the effluent monitoring data set is acquired;

[0143] S540, it is judged whether the water quality index meets the national standard water quality index, if the water pollution index meets the national standard water quality index, the wastewater discharge instruction is acquired.

[0144] Comparative example

[0145] The wastewater treatment aeration control intelligent control based on the environmental prior knowledge of the present application is compared with the traditional artificial fuzzy control, and the results are as follows:

[0146]

[0147] It can be seen that, by using the method of the present application, the COD and total nitrogen in the effluent are both significantly smaller than those of the artificial fuzzy control method, and the present application has obvious advantages.

[0148] The present application is not limited to the above-mentioned best embodiment, and anyone can derive other various forms of products under the inspiration of the present application, but regardless of any changes in shape or structure, any technical solution with the same or similar technical solutions as the present application falls within the protection scope of the present application.​

Claims

1. An intelligent control method for sewage treatment aeration based on environmental prior knowledge, characterized in that, It comprises the following steps: S100, obtaining water quality index monitoring data set according to water inlet monitoring, process monitoring and water outlet monitoring; The S100 comprises the following steps: SA100, obtaining water inlet monitoring data set by monitoring the water quality of the water sample collected through the water inlet sampling pipe by the data acquisition device; SB100, obtaining biochemical process monitoring data set by monitoring the water quality of the water sample collected through the biochemical process section sampling pipe by the data acquisition device; SC100, obtaining historical water outlet monitoring data set by monitoring the water quality of the water sample collected through the water outlet sampling pipe by the data acquisition device; In the SA100-SC100, the sampling time of the water inlet monitoring data and the biochemical process monitoring data is the current time t, and the sampling time of the historical water outlet monitoring data is the previous time t-1; the data acquisition device includes multiple types, each type of data acquisition device includes at least two water quality monitors, and the average value is used to avoid accidental errors; the data in the data set is preprocessed, missing values are filled, and abnormal values are deleted; S200, obtaining equipment working condition monitoring data set by collecting equipment data of the sewage treatment equipment; The equipment working condition monitoring data set includes digital control monitoring data set and real-time control monitoring data set; the data types in the equipment working condition monitoring data set include sewage anaerobic hydraulic retention time data, anoxic hydraulic retention time data, aerobic hydraulic retention time data, sludge transportation time and advanced treatment chemical addition time; S300, fusing the water quality index monitoring data set and the equipment working condition monitoring data set, and expanding into a fusion data set with environmental priori knowledge according to environmental priori knowledge; S310, aligning the data rows and data columns of the water quality index monitoring data set in S100 and the equipment working condition monitoring data set in S200, and setting the data interval to the minimum sampling time; S320, constructing the difference and percentage difference between the water quality of the water inlet, the water quality of the biochemical process section, the water quality of the water inlet and the historical water outlet, and the water quality of the biochemical process section and the historical water outlet according to the environmental priori knowledge of removal rate and removal percentage; S330, constructing the multi-day average value of the water quality of the water inlet, the water quality of the biochemical process section and the historical water outlet according to the environmental priori knowledge of average value; S340, constructing multiple variable comparison values of the water quality of the water inlet, the water quality of the biochemical process section and the historical water outlet according to the environmental priori knowledge of comparison value; S350, according to the environmental priori knowledge of difference value, difference percentage, average value and comparison value, data fusion is performed on the constructed features to obtain sewage treatment plant fusion data that can be used for model building; S400, building an adjustable parameter long short-term memory neural network adaptive model, wherein the adjustable parameter adaptive model is embedded in the control system terminal of the sewage treatment blower equipment; In the S400, the adjustable parameter long short-term memory neural network controls information flow through the forget gate, the input gate and the output gate; the forget gate determines how much memory of the previous time to discard, the input gate determines the storage of the current input information, and the output gate determines the output state at the current time; each gate has a corresponding weight matrix and bias parameter. The adjustable parameter long short-term memory neural network is constructed by calculating the error at the time step and back-propagating to adjust the weights and biases of the network; during the training process, the input sequence data is calculated by forward propagation to calculate the output, calculate the loss function, and then calculate the gradient of each time step by the back propagation algorithm, and continuously update the parameters of the long short-term memory neural network until the percentage error of the model is less than 5%, and output the temporary adjustable parameters; The adjustable parameter adaptive model includes the number of memory cells, the number of hidden layers of the neural network, and the dimension of the input data of the neural network; The adjustable parameter long short-term memory neural network mathematical model is as follows: Wherein: is activation function; are input values, including water quality monitoring data set independent variables, biochemical process section water quality monitoring data set independent variables, historical effluent water quality monitoring data set independent variables, equipment working condition monitoring data set independent variables, and water quality characteristic independent variables constructed from environmental priors; is a forget gate; is a weight matrix of the forget gate applied to the current input ; is a weight matrix of the forget gate applied to the hidden state at the previous time step ; is a bias term for the forget gate; is an input gate; is a weight matrix of the input gate applied to the current input ; is the weight matrix of the input gate applied to the hidden state at the previous time step ; is a bias term for the input gate; is the hyperbolic tangent activation function; is a memory cell state update; is a weight matrix of the memory cell states applied to the current input and the hidden state of the previous time instant ; is a bias term for the memory cell state; is the output gate, for the blower aeration amount; is a weight matrix of the output gate applied to the current input ; is the weight matrix of the output gate applied to the hidden state at the previous time step ; is a bias term for the output gate; is a hidden state; S500, input the fusion data into the adjustable parameter long short-term memory neural network adaptive model, control the wastewater treatment blower equipment according to the long short-term memory neural network and the adaptive parameters.

2. The intelligent control method of wastewater treatment aeration based on environmental prior knowledge according to claim 1, characterized in that: 20% of the fusion data in S300 is used for pre-training, the fusion data is used as an input variable, the long short-term memory neural network is trained, and temporary adjustable parameters are obtained.

3. The method of claim 2, wherein the method is characterized by, The S500 includes the following steps: S510, set the parameters of the long short-term memory neural network to the temporary adjustable parameters, input 100% of the fusion data into the long short-term memory neural network, and the model outputs the recommended aeration amount of the blower; S520, the control module adjusts the direction and size of the guide vane according to the recommended aeration amount of the blower, thereby affecting the aeration amount, and realizing precise control; S530, the water quality monitoring is performed on the water sample collected through the real-time effluent sampling pipe by the data acquisition device, the effluent water quality at time t is obtained, and the effluent monitoring data set is obtained; S540, judge whether the water quality index meets the standard, if the water pollution index meets the standard, obtain the wastewater discharge instruction.

Images