Intelligent control method for fluidization of filler in aerobic MBBR (Moving Bed Biofilm Reactor) reaction tank
By setting up multi-point monitoring in the MBBR reactor and using the YOLO algorithm to identify the fluidization state of the packing material, and combining this with a weighted allocation strategy to dynamically adjust aeration, the problem of precise and intelligent control of the packing material fluidization in the MBBR reactor was solved, achieving efficient and stable water treatment results.
Patent Information
- Application Number
- CN202512044640.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-31
- Publication Date
- 2026-02-24
AI Technical Summary
Existing fluidization control methods for MBBR reactor packing have qualitative judgment defects, strong subjectivity, and difficulty in achieving precision and intelligence. Furthermore, existing sensors and fluid dynamics models suffer from low monitoring accuracy, slow response speed, and poor applicability, failing to meet the requirements for efficient and stable operation of the MBBR process.
The YOLO algorithm, combined with multi-point monitoring and weight allocation strategy, is used to set up four monitoring points in the MBBR reactor. A pre-trained target detection model is used to identify the fluidization state of the packing material and calculate the global fluidization anomaly index. The aeration intensity is then dynamically adjusted to achieve uniform distribution of the packing material.
It realizes intelligent monitoring and automated control of the fluidization of the packing material in the MBBR reactor, improves the uniformity of packing material distribution, reduces aeration energy consumption, improves treatment efficiency and effluent quality, and reduces manual intervention.
Smart Images
Figure CN121554089A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent control technology for fluidized bed packing in water treatment, specifically to an intelligent control method for fluidized bed packing in an aerobic MBBR reactor. Background Technology
[0002] Fluidization state of the packing material is a core element for the efficient operation of moving bed biofilm processes. In MBBR systems, suspended packing material serves as a carrier for microbial attachment, and the uniformity of its free flow directly affects biofilm activity, pollutant mass transfer efficiency, and system stability. An ideal fluidization state requires achieving dynamic equilibrium of the packing material within the reaction tank, avoiding both localized accumulation that reduces biofilm-water contact and excessive fluidization that could lead to biofilm detachment. However, existing packing fluidization control methods have significant limitations and struggle to meet the precise and intelligent operation requirements of MBBR processes.
[0003] Currently, the assessment of the fluidization state of packing materials in existing technologies mainly relies on visual observation by operators or simple tools. For example, the movement trajectory and distribution of the packing materials are visually assessed through a transparent observation window, or a fixed aeration intensity is set based on historical experience. These methods have significant qualitative limitations, failing to quantify specific parameters of the packing material's fluidization state (such as the ambiguity of the critical conditions between "uniform distribution" and "local accumulation"), and are highly subjective, easily leading to misjudgments due to differences in operator experience. Furthermore, manual inspections require frequent on-site visits, increasing manpower burden and making it difficult to respond to abnormal packing material fluidization in real time, posing a risk of delay. When local accumulation or settling of the packing material occurs, it often only becomes apparent after the effluent quality has deteriorated, severely impacting treatment efficiency and system stability.
[0004] To compensate for the shortcomings of traditional methods, some projects have attempted to introduce single sensors (such as dissolved oxygen probes) for auxiliary monitoring. For example, the aeration rate is adjusted based on changes in dissolved oxygen (DO) concentration. However, such methods only reflect local parameters and cannot comprehensively capture the dynamic distribution characteristics of the packing material. DO concentration is affected by water quality fluctuations (such as temperature and organic load), which can easily lead to misadjustments. In addition, the installation and maintenance costs of sensors are high, and they require complex parameter calibration, limiting their widespread application in small and medium-sized MBBR projects.
[0005] Some literature proposes theoretical schemes for predicting the fluidization state of packing materials based on fluid dynamics models (such as CFD simulations), but these face the dual challenges of poor adaptability and high computational costs in practical engineering. The models require input of complex parameters such as packing density, hydraulic shear force, and bubble distribution, which are difficult to obtain in real-time on-site. Furthermore, the simulation process relies on high-performance computing resources, contradicting the core advantages of the MBBR process: "simple operation and convenient management." In addition, fluid dynamics models need to be individually calibrated for different reactor structures, lacking versatility and making it difficult to form a standardized solution.
[0006] In summary, existing fluidized bed control methods have significant shortcomings in terms of monitoring accuracy, response speed, and engineering applicability. There is an urgent need for an innovative technology that combines intelligence, automation, and real-time performance to achieve efficient and stable operation of the MBBR process. Summary of the Invention
[0007] The purpose of this invention is to provide an intelligent control method for the fluidization of packing material in an aerobic MBBR reactor. By combining the YOLO algorithm to identify the fluidization state of the packing material with multi-point monitoring and weight allocation strategies, this method breaks through the limitations of traditional qualitative analysis and provides a new path for the refined management of MBBR processes.
[0008] To achieve the above objectives, the present invention adopts the following technical solution:
[0009] A smart control method for fluidization of packing material in an aerobic MBBR reactor includes the following steps:
[0010] a. Set up four monitoring points in the aerobic MBBR reactor, ensuring that the four monitoring points cover the influent disturbance zone, the effluent stability zone, and the in-tank circulation zone;
[0011] b. Acquire real-time video streams of the packing material at the four monitoring sites respectively, use a pre-trained target detection model to classify and identify the packing material at each monitoring site, and output the fluidization state label of each site. The fluidization state label includes categories such as uniform distribution, local accumulation and excessive dispersion.
[0012] The training method for the target detection model is as follows:
[0013] b1. Model Construction: Construct an image dataset containing the fluidization state of the MBBR reactor packing material, and label it as uniform distribution, local accumulation, and excessive dispersion.
[0014] b2. Model training: YOLOv5 is trained using the CIoU Loss function; the input image size is 640×640 pixels, and the anchor box size is (10,13), (16,30), (33,23);
[0015] b3. Output results: The model output results include the coordinates of the filler bounding box and the probability of each category in the fluidization state label. The final detection results are filtered by NMS.
[0016] c. Combine the weight allocation method and the output results of the target detection model, and calculate the global fluidization anomaly index according to formula (1);
[0017] (1);
[0018] In equation (1): E is the global fluidization anomaly index; wi s is the weighting coefficient for the i-th site; i represents the monitoring site; s i Score the fluidization state of the i-th site;
[0019] d. Based on the global fluidization anomaly index calculated in step c, adjust the aeration to ensure that the packing fluidization is in a uniformly distributed fluidization state.
[0020] In the above-mentioned intelligent control method for fluidization of the packing material in an aerobic MBBR reactor, if the aerobic MBBR reactor is square, a circular area with a radius of 0.5m is delineated with the four corners of the top of the reactor as the center, and the intersection of the four corners with the center of the reactor and the circle is taken as the monitoring point.
[0021] In the above-mentioned intelligent control method for fluidization of the packing material in an aerobic MBBR reactor, if the aerobic MBBR reactor is circular, a monitoring point is taken at the intersection of the line connecting the inlet point, the outlet point, and the midpoint of the line connecting the inlet and outlet points with the center of the circle, extending to the wall of the reactor. The point is located 0.5m from the center of the circle.
[0022] In the above-mentioned intelligent control method for fluidization of packing material in an aerobic MBBR reactor, step b involves using an image acquisition device to obtain real-time video streams of the packing material at four monitoring points, with a monitoring frequency of once every 4 hours.
[0023] In the above-mentioned intelligent control method for fluidization of packing material in an aerobic MBBR reactor, in step c, s i The values are: uniform distribution equals 0, local accumulation equals 1, and excessive dispersion equals 2.
[0024] In the above-mentioned intelligent control method for fluidization of the packing material in an aerobic MBBR reactor, the specific distribution of the four monitoring points in step a is as follows: one monitoring point is set in the influent disturbance zone, one monitoring point is set in the effluent stabilization zone, and two monitoring points are set in the in-tank circulation zone.
[0025] In the above-mentioned intelligent control method for fluidization of packing material in an aerobic MBBR reactor, in step c, w i The values are as follows: Inlet disturbance zone: w1=0.3, because abnormal fluidization in this zone can easily cause packing blockage; Outlet stability zone: w2=0.4, because insufficient fluidization in this zone may lead to fluctuations in effluent quality; Inlet circulation zone: w3=0.15, w4=0.15, because the fluidization performance in this zone directly affects the overall efficiency of the reactor.
[0026] In the above-mentioned intelligent control method for fluidization of the packing material in an aerobic MBBR reactor, step d, the aeration adjustment method is as follows:
[0027] When the global fluidization anomaly index E < 0.425: increase the aeration intensity to 110% of the current value to improve the fluidization performance of the packing material;
[0028] When the global fluidization anomaly index E > 0.65: reduce the aeration intensity to 90% of the current value to reduce energy waste caused by excessive fluidization;
[0029] When 0.425≤E≤0.65: Maintain the current aeration intensity and no adjustment is required.
[0030] Compared with the prior art, the present invention brings the following beneficial technical effects:
[0031] (1) This invention provides an intelligent control method for the fluidization of packing material in an aerobic MBBR reactor. First, four monitoring points are set up in the aerobic MBBR reactor, and these four monitoring points are ensured to cover the influent disturbance zone, the effluent stability zone and the in-tank circulation zone. Then, real-time video streams of each monitoring point are collected periodically, and the packing material at each monitoring point is classified and identified using a pre-trained target detection model. Then, the global fluidization anomaly index is calculated by combining the weight allocation method and the output results of the target detection model with the formula. Finally, the aeration is adjusted according to the value of the global fluidization anomaly index.
[0032] (2) The present invention can realize dynamic monitoring of the fluidization of the packing material in the aerobic MBBR reactor, and the monitoring sites are set more reasonably, enabling real-time monitoring of different areas.
[0033] (3) Improved operation and maintenance safety: Through four monitoring points and weight allocation, the uniformity of packing distribution is improved to over 95%; Reduced aeration energy consumption: Combined with the global fluidization anomaly index, the aeration intensity is dynamically adjusted to avoid excessive air supply, which can save 10%-20% of aeration energy consumption; Improved treatment efficiency: Based on the uniformity of sulfidation, the biofilm thickness is maintained at a moderate level, which can achieve effluent ammonia nitrogen ≤0.5mg / L and effluent COD ≤20mg / L; Reduced manual intervention: Automated monitoring replaces manual inspection, reducing manpower input by more than 30%. Attached Figure Description
[0034] Figure 1 This is a water quality chart of the effluent over four consecutive years, as shown in Example 1 of the present invention. Detailed Implementation
[0035] This invention proposes an intelligent control method for the fluidization of packing material in an aerobic MBBR reactor. To make the advantages and technical solutions of this invention clearer and more explicit, the invention will be further described below with reference to specific embodiments.
[0036] The main technical concept of this invention lies in achieving dynamic monitoring and intelligent control of the fluidization of the packing material in an MBBR reactor. This is achieved through the following technical solution: the layout of monitoring points, the classification and identification of the packing material at each monitoring point using a pre-trained target detection model, the calculation of a global fluidization anomaly index, and the adjustment of aeration based on the specific value of the global fluidization anomaly index. As a whole, this technical solution enables intelligent control of the fluidization of the packing material in an MBBR reactor.
[0037] The present invention will be further described below with reference to specific embodiments.
[0038] Example 1:
[0039] A wastewater treatment plant in northern China has a capacity of 10,000 m³ / d. Its aerobic reactor utilizes the MBBR process, with a 50% MBBR carrier filling rate. Each aerobic reactor is a square tank with dimensions of 8m x 4m x 5m (length x width x depth). The annual average influent water quality is as follows: COD = 285.3 mg / L, NH3-N = 49.96 mg / L, SS = 256 mg / L. The effluent meets the Class A standard of the "Discharge Standard of Pollutants for Municipal Wastewater Treatment Plants" (GB 18918-2002), with NH3-N ≤ 5 mg / L.
[0040] Technical solution:
[0041] Monitoring points: A circular area with a radius of 0.5m is delineated with the four corners of the top of the square pool as the center. The intersection points are the inlet disturbance zone, the outlet stable zone, the left side of the circulation zone, and the right side of the circulation zone, respectively.
[0042] Weighting: 0.3 for influent disturbance zone, 0.4 for effluent stability zone, 0.15 for the left side of circulation zone, and 0.15 for the right side of circulation zone;
[0043] Anomaly index calculation: si values are local accumulation = 0, uniform distribution = 0.5, and excessive dispersion = 1;
[0044] Combining the weight allocation method and the output results of the target detection model, the global fluidization anomaly index is calculated according to equation (1);
[0045] (1);
[0046] In equation (1): E is the global fluidization anomaly index; w i s is the weighting coefficient for the i-th site; i represents the monitoring site; s i Score the fluidization state of the i-th site;
[0047] Aeration adjustment plan:
[0048] When the global fluidization anomaly index E < 0.425: increase the aeration intensity to 110% of the current value to improve the fluidization performance of the packing material;
[0049] When the global fluidization anomaly index E > 0.65: reduce the aeration intensity to 90% of the current value to reduce energy waste caused by excessive fluidization;
[0050] When 0.425≤E≤0.65: Maintain the current aeration intensity and no adjustment is required.
[0051] Running result:
[0052] like Figure 1 As shown, after 4.5 years of continuous application, the average effluent water quality was: COD=23.4±1.22mg / L, NH3-N=0.21±0.17mg / L, SS=7.9±0.22mg / L.
[0053] The uniformity of the filler is 100%, and the labor input is reduced by 37%.
[0054] Comparative Example 1: Incorrect Weight Allocation
[0055] The MBBR pilot system was adopted, in which the aerobic reactor adopted the MBBR process, the MBBR carrier filling rate was 50%, and the single aerobic reactor had a size of 2m×1m×1.25m (square) with dimensions of length×width×depth. The influent water quality of the pilot test was consistent with that of Example 1, and the effluent complied with the Class A standard of the "Discharge Standard of Pollutants for Municipal Wastewater Treatment Plants" (GB 18918-2002), with NH3-N ≤5mg / L.
[0056] Technical solution:
[0057] Weighting: Inlet disturbance zone: w1=0.2, Outlet stable zone: w2=0.3, Pool circulation zone: w3=0.2.
[0058] Anomaly index calculation: si values are local accumulation = 0, uniform distribution = 0.5, and excessive dispersion = 1;
[0059] Combining the weight allocation method and the output results of the target detection model, the global fluidization anomaly index is calculated according to equation (1);
[0060] (1);
[0061] In equation (1): E is the global fluidization anomaly index; w i s is the weighting coefficient for the i-th site; i represents the monitoring site; s i The fluidization state score is given for the i-th site.
[0062] Aeration adjustment plan:
[0063] When the global fluidization anomaly index E < 0.425: increase the aeration intensity to 110% of the current value to improve the fluidization performance of the packing material;
[0064] When the global fluidization anomaly index E > 0.65: reduce the aeration intensity to 90% of the current value to reduce energy waste caused by excessive fluidization;
[0065] When 0.425≤E≤0.65: Maintain the current aeration intensity and no adjustment is required.
[0066] Operational results: After one year of continuous application, the average effluent water quality was as follows: COD = 29.5 ± 1.28 mg / L, NH3-N = 2.49 ± 1.29 mg / L, SS = 7.8 ± 0.37 mg / L, with the highest ammonia nitrogen value reaching 5.26 mg / L, indicating that the levels exceeded the standards.
[0067] Any parts not mentioned in this invention can be achieved by referring to existing technologies.
[0068] Those skilled in the art should recognize that the above embodiments are only used to illustrate this application and are not intended to limit this application. Any appropriate changes and variations made to the above embodiments within the essential spirit and scope of this application fall within the scope of protection claimed in this application.
Claims
1. A smart control method for fluidization of packing material in an aerobic MBBR reactor, characterized in that, The steps are as follows: a. Set up four monitoring points in the aerobic MBBR reactor, ensuring that the four monitoring points cover the influent disturbance zone, the effluent stability zone, and the in-tank circulation zone; b. Acquire real-time video streams of the packing material at the four monitoring sites respectively, use a pre-trained target detection model to classify and identify the packing material at each monitoring site, and output the fluidization state label of each site. The fluidization state label includes categories such as uniform distribution, local accumulation and excessive dispersion. The training method for the target detection model is as follows: b1. Model Construction: Construct an image dataset containing the fluidization state of the MBBR reactor packing material, and label it as uniform distribution, local accumulation, and excessive dispersion. b2. Model training: YOLOv5 is trained using the CIoU Loss function; the input image size is 640×640 pixels, and the anchor box size is (10,13), (16,30), (33,23); b3. Output results: The model output results include the coordinates of the filler bounding box and the probability of each category in the fluidization state label. The final detection results are filtered by NMS. c. Combine the weight allocation method and the output results of the target detection model, and calculate the global fluidization anomaly index according to formula (1); （1）； In equation (1): E is the global fluidization anomaly index; w i s is the weighting coefficient for the i-th site; i represents the monitoring site; s i Score the fluidization state of the i-th site; d. Based on the global fluidization anomaly index calculated in step c, adjust the aeration to ensure that the packing fluidization is in a uniformly distributed fluidization state.
2. The intelligent control method for fluidization of packing material in an aerobic MBBR reactor according to claim 1, characterized in that: If the aerobic MBBR reactor is square, then a circular area with a radius of 0.5m is delineated with the four corners of the top of the reactor as the center, and the intersection of the four corners with the center of the reactor and the circle is taken as the monitoring point.
3. The intelligent control method for fluidization of packing material in an aerobic MBBR reactor according to claim 1, characterized in that: If the aerobic MBBR reactor is circular, then the point where the line connecting the inlet point, the outlet point, and the midpoint of the line connecting the inlet and outlet points intersects with the center of the circle and is extended to the reactor wall, is taken as the monitoring point 0.5m from the center of the circle.
4. The intelligent control method for fluidization of packing material in an aerobic MBBR reactor according to claim 1, characterized in that: In step b, an image acquisition device is used to acquire real-time video streams of the packing material at four monitoring sites, with a monitoring frequency of once every 4 hours.
5. The intelligent control method for fluidization of packing material in an aerobic MBBR reactor according to claim 1, characterized in that: In step c, s i The values are: 0 for local accumulation, 0.5 for uniform distribution, and 1 for excessive dispersion.
6. The intelligent control method for fluidization of packing material in an aerobic MBBR reactor according to claim 1, characterized in that: In step a, the four monitoring sites are distributed as follows: one monitoring site is set in the influent disturbance zone, one monitoring site is set in the effluent stability zone, and two monitoring sites are set in the pool circulation zone.
7. The intelligent control method for fluidization of packing material in an aerobic MBBR reactor according to claim 1, characterized in that: In step c, w i The values are as follows: Inlet disturbance zone: w1=0.3, because abnormal fluidization in this zone can easily cause packing blockage; Outlet stable zone: w2=0.4, because insufficient fluidization in this zone may lead to fluctuations in the effluent quality. In the circulation zone of the tank: w3=0.15, w4=0.15, because the fluidization performance of this zone directly affects the overall efficiency of the reactor.
8. The intelligent control method for fluidization of packing material in an aerobic MBBR reactor according to claim 1, characterized in that: In step d, the aeration adjustment method is as follows: When the global fluidization anomaly index E < 0.425: increase the aeration intensity to 110% of the current value to improve the fluidization performance of the packing material; When the global fluidization anomaly index E > 0.65: reduce the aeration intensity to 90% of the current value to reduce energy waste caused by excessive fluidization; When 0.425≤E≤0.65: Maintain the current aeration intensity and no adjustment is required.