Sewage treatment partition aeration control method based on LSTM feedforward prediction and feedback

Abstract

The invention relates to the technical field of sewage treatment, in particular to a sewage treatment partition aeration control method based on LSTM feedforward prediction and feedback, which comprises the following steps: S1, collecting real-time key data of a sewage treatment tank, the real-time key data comprising a dissolved oxygen measured value, an ammonia nitrogen measured value, a mixed liquid suspended solid concentration measured value and a flow measured value; s2, acquiring historical key data, inputting the historical key data into the long-short-term memory neural network model, and outputting an aeration prediction amount required by aeration; s3, calculating a deviation value between a dissolved oxygen measured value and a dissolved oxygen set value and a change rate of the dissolved oxygen measured value, and carrying out primary adjustment on the aeration prediction quantity based on the deviation value, the change rate and a fuzzy control algorithm; calculating a correction parameter K according to the sludge concentration, and adjusting the aeration prediction amount again based on the correction parameter K to obtain a target aeration amount; s4, determining the number and the opening degree of air blowers based on the target aeration rate; and S5, determining the opening degree of the valve based on the target aeration rate. The energy consumption can be reduced, and the sewage treatment efficiency can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sewage treatment, and in particular to a sewage treatment partitioned aeration control method based on LSTM feedforward prediction and feedback. BACKGROUND

[0002] Traditional sewage treatment aeration control methods mainly rely on manual adjustment or simple PID control systems, which have many limitations in practical applications. First, the aeration amount is difficult to control accurately, leading to problems of excessive aeration or insufficient aeration. Excessive aeration can cause energy waste, and according to statistics, the aeration system often accounts for 40%-60% of the total energy consumption of a sewage treatment plant, while insufficient aeration affects the treatment effect of the activated sludge process, causing unstable effluent quality and even exceeding the discharge standard. Second, existing equipment is prone to quality problems such as aeration head blockage, material aging, and disc-type aeration membrane damage, which not only increases maintenance costs but also may cause system failure and environmental pollution. For example, the generation of biological foam can interfere with the normal display of instruments, affect the operation of the automatic control system, and have a negative impact on the surrounding environment.

[0003] In addition, traditional methods lack real-time prediction and feedback mechanisms for water quality parameters such as dissolved oxygen and ammonia nitrogen, and cannot adapt to dynamic changes in influent water quality. Although there have been attempts at biological treatment, membrane technology, and intelligent monitoring in existing technologies, these methods often have low processing efficiency, high cost, and cannot achieve energy self-sufficiency and resource recovery. For example, the aeration control system has difficulty in optimizing oxygen supply when dealing with high or low load scenarios, leading to increased energy consumption and decreased processing efficiency. In recent years, although some sewage treatment plants have introduced automated aeration control, they still face the problem of insufficient accuracy, which cannot be adjusted in real time according to flow and biochemical oxygen demand (BOD), leading to rising operating costs and increased environmental impact. SUMMARY

[0004] The present application aims to provide a sewage treatment partitioned aeration control method based on LSTM feedforward prediction and feedback, which can reduce energy consumption and improve sewage treatment efficiency.

[0005] To solve the above technical problems, the present application provides the following technical solutions: The present application provides a sewage treatment partitioned aeration control method based on LSTM feedforward prediction and feedback, comprising the following steps: S1: Collecting real-time key data of the sewage treatment tank, the real-time key data including dissolved oxygen measured value, ammonia nitrogen measured value, mixed liquor suspended solids concentration measured value, and flow measured value; S2: Obtaining historical key data corresponding to the real-time key data, inputting the historical key data into a long short-term memory neural network model, and outputting aeration prediction required for aeration; S3: Calculate the deviation value of the measured value of dissolved oxygen and the set value of dissolved oxygen and the change rate of the measured value of dissolved oxygen, and adjust the aeration prediction value based on the deviation value, the change rate and the fuzzy control algorithm; calculate the correction parameter K based on the sludge concentration, and adjust the aeration prediction value again based on the correction parameter K to obtain the target aeration amount; S4: Determine the number and opening degree of the blower based on the target aeration amount; S5: Determine the valve opening degree based on the target aeration amount.

[0006] By adopting the above technical scheme, the LSTM model is adopted to perform feedforward prediction of the aeration gas amount required by the sewage treatment tank based on historical key data, the key parameters of the sewage treatment tank are collected in real time through a sensor, a feedback control model is combined, the aeration gas amount that needs to be adjusted is calculated, and finally the aeration gas amount required is calculated. The precise control of the blower and the valve is realized according to the required aeration gas amount. The method includes data acquisition, LSTM prediction, feedback adjustment, blower opening degree control and valve opening degree adjustment. Through the method, the aeration process can be effectively optimized, the energy consumption is reduced, the sewage treatment efficiency is improved, and the method is suitable for scenes such as municipal sewage treatment plants.

[0007] Optionally, the sewage treatment tank is one or more.

[0008] Optionally, the S2 comprises: The historical key data is acquired and normalized, and the missing values are filled with 0. A long short-term memory neural network model is constructed, the historical key data of the last 5 time steps is input, and the aeration prediction value of the next time step is output. The aeration prediction value is inversely normalized.

[0009] Optionally, the input sequence length of the historical key data in the S2 is 5.

[0010] Optionally, the S4 comprises: Read the remote state and actual opening degree of each blower; Determine and start the target blower according to the target aeration amount, and close the non-target blower; Calculate the difference diff between the target opening degree and the actual opening degree, if diff <-5, adjust to actual opening degree-5; if -5 ≤diff ≤ 5, use the target opening degree; if diff>5, adjust to actual opening degree+5; Write the adjusted opening degree, and wait for 5 seconds to verify the state.

[0011] Optionally, the S6 comprises: When the sewage treatment tank is one, the target aeration amount is distributed to the inlet section, the middle section and the outlet section respectively, and the valve opening degree is determined according to the distribution proportion of the inlet section, the middle section and the outlet section.

[0012] Optionally, the S6 further comprises: When there are multiple sewage treatment tanks, the proportion of the target aeration amount of tank A to the total target aeration amount, the proportion of the target aeration amount of the intermediate section of tank A to the total target aeration amount of tank A, and the proportion of the target aeration amount of the intermediate section of tank B to the total target aeration amount of tank B are calculated. According to the proportion of the target aeration amount of tank A to the total target aeration amount in different intervals, different valve control strategies are adopted to determine the valve opening degrees of the inlet section of tank A, the outlet section of tank A, the inlet section of tank B, and the outlet section of tank B. According to the proportion of the target aeration amount of the intermediate section of tank A to the total target aeration amount of tank A and the proportion of the target aeration amount of the intermediate section of tank B to the total target aeration amount of tank B, the valve opening degrees of the intermediate section of tank A and the intermediate section of tank B are determined.

[0013] Optionally, a sewage treatment partition aeration control method based on LSTM feedforward prediction and feedback further comprises: realizing data interaction through an OPC interface.

[0014] Optionally, the data interaction through the OPC interface comprises: writing the valve opening degree, the blower opening degree, the alarm state, and the time point into designated OPC point positions, and writing the time point in a string format as "YYYY-MM-DD HH:MM:SS".

[0015] In summary, the present application at least includes the following beneficial technical effects: 1. Improved control accuracy: through LSTM neural network model for time series prediction of dissolved oxygen (DO) and ammonia nitrogen (AD), combined with fuzzy control, dynamic adjustment of aeration amount, significantly improving control accuracy, ensuring stable effluent water quality.

[0016] 2. Reduce energy consumption: based on the predicted aeration demand to optimize the blower and valve opening degree, limit the blower opening degree change amplitude not more than 5, avoid excessive aeration, reduce the energy consumption of sewage treatment plant. At the same time, the blower start-stop time limit (30 minutes after starting to stop, 30 minutes after stopping to start) reduces frequent switching, prolongs the service life of the equipment.

[0017] 3. Enhance system stability: through OPC UA interface to read the blower and valve state in real time, verify the deviation of actual opening degree and set value, and trigger alarm when the deviation is out of limit, combined with time stamp recording operation, ensure the traceability and stability of system operation.

[0018] 4. Adapt to dynamic changes: LSTM model uses historical data (sequence length of 5) to predict future aeration demand, adapts to dynamic changes of influent water quality and flow, is superior to traditional PID control static adjustment, especially suitable for high load or low load scenarios.

[0019] 5. High scalability: The method supports multi-pool systems (e.g., A-pool and B-pool) and can be extended to more pools or parameters through modular design (data acquisition, prediction, feedback control, execution), which is suitable for complex scenarios such as municipal wastewater treatment plants. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 Figure 1 is a flow chart of a wastewater treatment partitioned aeration control method based on LSTM feedforward prediction and feedback. Figure 2 Figure 2 is a schematic diagram of an aeration automatic control model. Figure 3 Figure 3 is an aeration process diagram. Figure 4 Figure 4 is a schematic diagram of an aeration equipment. DETAILED DESCRIPTION

[0021] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below in combination with the drawings and examples. It should be understood that the specific examples described herein are only used to explain the present application and are not intended to limit the present application.

[0022] The terms used in the following examples of the present application are only for the purpose of describing specific examples and are not intended to be limiting on the present application. As used in the specification and the appended claims of the present application, the singular forms "a," "an" and "the" are intended to include plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "and / or" as used herein refer to and include any or all possible combinations of one or more of the listed items. The terms "first," "second," are used only for the purpose of description and cannot be understood as implying or suggesting relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first," "second" can explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, the meaning of "a plurality of" is two or more, unless otherwise specified.

[0023] The embodiment of the present application provides a wastewater treatment partitioned aeration control method based on LSTM feedforward prediction and feedback.

[0024] Reference Figures 1-4 A wastewater treatment partitioned aeration control method based on LSTM feedforward prediction and feedback, comprising the following steps: S1: Collecting real-time key data of the wastewater treatment tank, the real-time key data including dissolved oxygen (DO) measured value, ammonia nitrogen (AD) measured value, mixed liquid suspended solids concentration (MLSS) measured value, and flow measured value.

[0025] Real-time key data is collected by corresponding sensors. Among them, the sewage treatment tank can be one or multiple, which is a biological tank (oxygen tank). In this application, two biological tanks A and B are taken as examples.

[0026] After collecting the real-time key data, the real-time key data is stored.

[0027] S2: Obtain the historical key data corresponding to the real-time key data, input the historical key data into the long short-term memory (LSTM) neural network model, and output the aeration prediction value required for aeration.

[0028] First, after obtaining the historical key data, it is normalized and the missing values are filled with 0. Then, the long short-term memory neural network model is constructed, the historical key data of the last 5 time steps (the input sequence length of the historical key data is 5) is input, and the aeration prediction value of the next time step is output. Finally, the aeration prediction value is inversely normalized for subsequent steps.

[0029] S3: Calculate the deviation value of the measured value of dissolved oxygen and the set value of dissolved oxygen and the change rate of the measured value of dissolved oxygen, and based on the deviation value, the change rate and the fuzzy control algorithm, the aeration prediction value is adjusted for the first time; calculate the correction parameter K according to the sludge concentration, and adjust the aeration prediction value again based on the correction parameter K to obtain the target aeration amount.

[0030] The deviation value of the measured value of dissolved oxygen and the set value of dissolved oxygen and the change rate of the measured value of dissolved oxygen are calculated, and based on the deviation value D1 of dissolved oxygen, the change rate D2 of dissolved oxygen before and after two adjustment beats, the aeration feedback compensation value is calculated by using the fuzzy control algorithm; calculate the correction parameter K according to the sludge concentration, and adjust the aeration feedback value again based on the correction parameter K, and obtain the target aeration amount by comprehensively considering the aeration feedforward prediction value. For details, please refer to Table 1.

[0031] Table 1 Dissolved oxygen control table For the correction parameter K, according to the change range of the sludge concentration, different coefficient values are preset to form a control table. During actual calculation, according to the sludge concentration value, the corresponding preset coefficient can be found by looking up the table.

[0032] In the calculation process of the deviation value, since the measured value of dissolved oxygen has been obtained in S1, only the set value of dissolved oxygen needs to be obtained in this step, and then the absolute value is obtained by subtracting the two. It should be understood that the measured value of dissolved oxygen can be one or more, and the average value can be calculated when multiple values are taken.

[0033] In the calculation process of the change rate, a change rate calculation formula is used, and multiple measured values of dissolved oxygen need to be obtained, and the time intervals of the multiple measured values of dissolved oxygen are the same. In this embodiment, the number of values of the measured values of dissolved oxygen is 4, which are the present value, the value 15 minutes ago, the value 30 minutes ago and the value 45 minutes ago, and the time interval is 15 minutes. In other embodiments, the number of values and the time interval can be set according to the time, which is not limited here.

[0034] When the sewage treatment tank is one, the target aeration quantity is corresponding to the sewage treatment tank.

[0035] When the sewage treatment tank is multiple, the target aeration quantity of a single sewage treatment tank is distributed according to the proportion of the target aeration quantity in the total target aeration quantity.

[0036] S4: Determine the number and opening of the blower based on the target aeration quantity.

[0037] Firstly, the remote state and actual opening of each blower are read through the OPC interface, and then the target blower is determined and started according to the target aeration quantity, and the non-target blower is closed. Then, the difference diff between the target opening and the actual opening is calculated, if diff<-5, it is adjusted to actual opening-5; if-5 ≤ diff ≤ 5, the target opening is used; if diff>5, it is adjusted to actual opening+5. Finally, the adjusted opening is written, and the state is verified for 5 seconds.

[0038] S5: Determine the valve opening based on the target aeration quantity.

[0039] In the sewage treatment tank, the sewage treatment tank can be divided into an inlet section, an intermediate section and an outlet section, and the inlet section, the intermediate section and the outlet section all need to be provided with valves.

[0040] When the sewage treatment tank is one, the target aeration quantity is distributed to the inlet section, the intermediate section and the outlet section respectively, and the valve opening is determined according to the distribution proportion of the inlet section, the intermediate section and the outlet section.

[0041] When the sewage treatment tank is multiple, the valve opening is determined by the same principle. For the convenience of understanding, A tank and B tank are taken as examples.

[0042] According to the proportion of the target aeration quantity of A tank in the total target aeration quantity, the proportion of the target aeration quantity of the intermediate section of A tank in the total target aeration quantity of A tank and the proportion of the target aeration quantity of the intermediate section of B tank in the total target aeration quantity of B tank, the opening of the six valves is calculated, and the set value is written after waiting for 20 seconds, the deviation between the actual value and the set value is verified, and the alarm state is updated.

[0043] More specifically, first, the proportion of the target aeration amount of the A pool to the total target aeration amount, the proportion of the target aeration amount of the intermediate section of the A pool to the total target aeration amount of the A pool, and the proportion of the target aeration amount of the intermediate section of the B pool to the total target aeration amount of the B pool are calculated. Then, according to the proportion of the target aeration amount of the A pool to the total target aeration amount, different valve control strategies are adopted in different intervals to determine the valve opening of the inlet section of the A pool, the outlet section of the A pool, the inlet section of the B pool, and the outlet section of the B pool. Then, according to the proportion of the target aeration amount of the intermediate section of the A pool to the total target aeration amount of the A pool and the proportion of the target aeration amount of the intermediate section of the B pool to the total target aeration amount of the B pool, the valve opening of the intermediate section of the A pool and the intermediate section of the B pool is determined, and the corresponding opening value is written into the OPC interface point. Finally, 20 seconds are waited, the actual opening is read, the deviation is calculated, if the deviation exceeds the tolerance, the alarm is triggered and the OPC point is written.

[0044] A wastewater treatment partitioned aeration control method based on LSTM feedforward prediction and feedback, further comprising data interaction through an OPC interface after S5, including: writing the valve opening, the blower opening, the alarm state and the time point into the specified OPC point, the time point is written in string format, and the format is "YYYY-MM-DD HH:MM:SS".

[0045] The above-described embodiments are only used to specifically describe the technical solutions of the present application, and the above-described embodiments are only used to help understand the method of the present application and its core idea, and should not be understood as limiting the present application. Within the technical scope disclosed by the present application, changes or replacements that can be easily thought of by those skilled in the art should be covered within the protection scope of the present application.

Claims

1. A wastewater treatment zone aeration control method based on LSTM feedforward prediction and feedback, characterized in that, Includes the following steps: S1: Collect real-time key data from the wastewater treatment pond, including measured values ​​of dissolved oxygen, ammonia nitrogen, mixed liquor suspended solids concentration, and flow rate. S2: Obtain historical key data corresponding to real-time key data, input the historical key data into the long short-term memory neural network model, and output the predicted amount of aeration required for aeration. S3: Calculate the deviation between the measured dissolved oxygen value and the set dissolved oxygen value, as well as the rate of change of the measured dissolved oxygen value. Based on the deviation, the rate of change, and the fuzzy control algorithm, make an initial adjustment to the aeration prediction amount. Calculate the correction parameter K based on the sludge concentration. Based on the correction parameter K, adjust the aeration prediction amount again to obtain the target aeration amount. S4: Determine the number and opening degree of blowers based on the target aeration rate; S5: Determine the valve opening based on the target aeration rate.

2. The wastewater treatment zone aeration control method based on LSTM feedforward prediction and feedback as described in claim 1, characterized in that, The wastewater treatment pond may be one or more.

3. The wastewater treatment zone aeration control method based on LSTM feedforward prediction and feedback as described in claim 2, characterized in that, S2 includes: Acquire historical key data and normalize it, filling in missing values ​​with 0; construct a long short-term memory neural network model, input the historical key data of the last 5 time steps, and output the aeration prediction for the next time step; perform inverse normalization on the aeration prediction.

4. The wastewater treatment zone aeration control method based on LSTM feedforward prediction and feedback as described in claim 3, characterized in that, The length of the input sequence of historical key data in S2 is 5.

5. A wastewater treatment zone aeration control method based on LSTM feedforward prediction and feedback as described in any one of claims 1-4, characterized in that, S4 includes: Read the remote status and actual opening degree of each blower; Determine and start the target blower based on the target aeration volume, and shut down the non-target blowers; Calculate the difference between the target opening and the actual opening, diff. If diff < -5, adjust to the actual opening -5; if -5 ≤ diff ≤ 5, use the target opening; if diff > 5, adjust to the actual opening +5. Write the adjusted opening value and wait 5 seconds to verify the status.

6. The wastewater treatment zone aeration control method based on LSTM feedforward prediction and feedback as described in claim 5, characterized in that, S6 includes: When there is only one sewage treatment tank, the target aeration volume is allocated to the inlet section, the middle section and the outlet section respectively, and the valve opening is determined according to the allocation ratio of the inlet section, the middle section and the outlet section.

7. The wastewater treatment zone aeration control method based on LSTM feedforward prediction and feedback as described in claim 6, characterized in that, S6 further includes: When there are multiple sewage treatment tanks, calculate the proportion of the target aeration volume of tank A to the total target aeration volume, the proportion of the target aeration volume of the middle section of tank A to the total target aeration volume of tank A, and the proportion of the target aeration volume of the middle section of tank B to the total target aeration volume of tank B. Based on the proportion of the target aeration volume of pool A to the total target aeration volume, different valve control strategies are adopted in different intervals to determine the valve opening of the inlet section of pool A, the outlet section of pool A, the inlet section of pool B, and the outlet section of pool B. The valve opening degree of the middle section of pool A and the middle section of pool B is determined based on the proportion of the target aeration rate of the middle section of pool A to the total target aeration rate of pool A, and the proportion of the target aeration rate of the middle section of pool B to the total target aeration rate of pool B.

8. The wastewater treatment zone aeration control method based on LSTM feedforward prediction and feedback as described in claim 1, characterized in that, Also includes: Data interaction is achieved through the OPC interface.

9. The wastewater treatment zone aeration control method based on LSTM feedforward prediction and feedback as described in claim 8, characterized in that, The data interaction via the OPC interface includes writing valve opening degree, blower opening degree, alarm status and time point to a specified OPC point. The time point is written in string format, which is "YYYY-MM-DD HH:MM:SS".

Images