Intelligent control system and method for cracks in impermeable concrete in sewage treatment ponds

Abstract

This invention discloses an intelligent control system and method for cracks in impermeable concrete in sewage treatment ponds. The system includes a sensing module, an execution module, and a control module. The method involves collecting real-time data, calculating the crack risk index R at each monitoring point, and dynamically generating a maintenance priority queue. An LSTM model is used to predict the future temperature field, and the optimal sequence of spraying, cooling, and insulation actions is solved based on the LSTM model prediction results. Simultaneously, a reinforcement training module is used to continuously optimize the strategy. Compared with existing technologies, this invention achieves quantitative assessment and predictive control of crack risk, reducing cracks by more than 80%, saving water and electricity by 30%-45%, and possessing adaptive environmental changes and emergency response capabilities. It is suitable for various large-volume impermeable concrete pond projects.

Description

Technical Field

[0001] This invention belongs to the field of building engineering and intelligent construction technology, specifically relating to an intelligent control system and method for cracks in impermeable concrete in sewage treatment ponds. Background Technology

[0002] Wastewater treatment ponds, due to their long-term contact with corrosive wastewater, have high requirements for leak-proofing, corrosion resistance, structural durability, and environmental protection. Therefore, wastewater treatment ponds are typically constructed using large-volume impermeable concrete. However, two core issues arise during the concrete hardening process:

[0003] 1. Cement hydration generates a large amount of heat, resulting in high internal temperature of concrete and rapid heat dissipation from the surface, forming a temperature gradient and generating thermal tensile stress. When the tensile stress exceeds the tensile strength of concrete, thermal cracks will occur.

[0004] 2. The surface moisture of concrete evaporates faster than the internal moisture, making it prone to shrinkage cracks. These cracks significantly reduce the impermeability of wastewater treatment ponds, allowing corrosive media in the wastewater to easily penetrate along the cracks, accelerating steel corrosion and structural failure.

[0005] To address the aforementioned problems, existing maintenance methods mainly include mulching and watering, timed spraying, pre-embedded cooling water pipes for water cooling, and laying insulation layers. However, these methods have the following drawbacks:

[0006] 1. The maintenance work relies on manual inspection and experience judgment, which cannot match the internal temperature of concrete and environmental changes in real time. Remedial measures are often only taken after cracks have appeared.

[0007] 2. Once parameters such as spray frequency, cooling water temperature, and insulation layer opening and closing are set, they cannot be adjusted and cannot adapt to weather changes such as day-night temperature differences, strong winds, and rainfall.

[0008] 3. Spraying, cooling, and heat preservation measures operate independently without a unified decision-making center, which can easily lead to over-maintenance (wasting water and electricity) or under-maintenance.

[0009] 4. Only temperature is monitored, but the actual strain of concrete is not directly monitored, so potential cracks cannot be predicted. In particular, there is a lack of targeted control for areas with concentrated confined stress, such as the junction of the pool wall and the bottom slab, and the area around the through-wall pipe sleeve.

[0010] Therefore, there is an urgent need for a control system that can sense multiple parameters in real time, make dynamic decisions, and execute collaboratively to suppress the generation of cracks. Summary of the Invention

[0011] The purpose of this invention is to provide an intelligent control system and method for cracks in impermeable concrete in sewage treatment ponds, so as to achieve predictive, adaptive and self-optimizing maintenance actions.

[0012] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0013] A smart crack control system for impermeable concrete in a wastewater treatment plant includes a sensing module, an execution module, and a control module. The sensing module includes a temperature sensor array, a humidity sensor, a vibrating wire strain gauge, and a weather station. The temperature sensor array is embedded inside and on the surface of the concrete to collect the internal temperature T of the concrete. in and surface temperature T surf The humidity sensor is attached to the concrete surface to collect the relative humidity (RH) of the concrete surface, and the vibrating wire strain gauge is embedded in the stress concentration area to collect the changes in tensile stress ε(t) and strain rate d in real time. ε / d t The on-site meteorological station collects information on ambient temperature T, wind speed v, solar radiation intensity, and rainfall.

[0014] The execution module includes an automatic sprinkler system, a temperature-controlled water circulation system, and an insulation layer covering device. The automatic sprinkler system includes multiple atomizing nozzles with solenoid valves, which are evenly distributed on the sewage treatment tank through water supply pipes. The temperature-controlled water circulation system includes a cooling water pipe, a circulating water pump connected to the cooling water pipe, and a water tank with adjustable water temperature connected to the inlet of the circulating water pump. The cooling water pipe is embedded in the concrete. The insulation layer covering device includes an electric roller shutter and a covering film wrapped around the electric roller shutter, which can automatically unfold or retract the covering film.

[0015] The control module includes a controller that performs overall control of the collected data calculation and execution module. The controller incorporates a dynamic risk field assessment module, a trend prediction module, a reinforcement training module, and a data storage module. The dynamic risk field assessment module is based on real-time collected data such as the inner surface temperature difference ΔT and the cooling rate d. T / d t Surface humidity (RH), strain ratio (ε) i / ε u and ambient wind speed V windThe system calculates the crack risk index R at each monitoring point and generates a priority queue of maintenance actions based on the calculated risk index. The area with the highest risk index is given priority for execution. The trend prediction module uses an LSTM neural network to predict the changing trends of the concrete temperature field and crack risk index over a certain time domain. Within each control cycle, it solves for the action sequence that minimizes the objective function J and transmits it to the controller for execution. The reinforcement training module uses historical maintenance data and a deep Q-network algorithm trained with the cumulative crack risk reduction minus energy consumption penalty as the reward function. The training results are transmitted to the controller for further optimization of the trend prediction module. The data storage module stores real-time maintenance data.

[0016] By constructing an integrated intelligent crack control system that combines "perception, control, and execution," multi-source data such as temperature, humidity, strain, and meteorology are fused together. This system integrates risk assessment, LSTM prediction, reinforcement learning, and data storage modules, and coordinates automatic execution of spraying, temperature-controlled water circulation, and insulation layer. It overcomes the shortcomings of traditional single temperature measurement, manual experience, and decentralized maintenance, and achieves intelligent predictive crack control throughout the entire process.

[0017] Preferably, the internal surface temperature difference ΔT of the concrete is expressed by the formula ΔT=T in -T surf Calculations are performed to quantify the temperature gradient by the difference between the internal and surface temperatures of concrete, providing a unified and quantifiable core parameter for crack risk assessment and improving the accuracy of crack control.

[0018] Preferably, the dynamic risk field assessment module calculates the crack risk index R using the following formula:

[0019]

[0020] Where: R i Let ΔT be the crack risk index for the i-th monitoring point. max V max RH min V windmax ΔT is the preset reference value. i d represents the real-time internal surface temperature difference of the concrete. T / d t ε is the rate of cooling of the inner surface temperature of the concrete. i / ε u Let ε be the real-time tensile strain ratio at the i-th monitoring point. u For the ultimate tensile strain, RH i Let V be the surface relative humidity at the i-th monitoring point. wind For ambient wind speed, , , , , These are dynamic weighting coefficients, and The dynamic weighting coefficients are adaptively adjusted according to the concrete age and curing stage. The crack risk index Ri is weighted by five parameters, which integrate temperature difference, cooling rate, strain ratio, humidity, and wind speed. The weights are dynamically and adaptively adjusted to realize the quantification, visualization, and ranking of crack risk, providing core algorithm support for intelligent priority decision-making.

[0021] Preferably, when the strain rate d at a certain monitoring point ε / d t When the risk index R of the monitoring point exceeds the set threshold, i The amplification factor enables emergency intervention when strain exceeds the limit. The value range of the amplification factor is (1.5, 2.5). When the strain rate exceeds the threshold, high-priority emergency intervention is initiated to quickly stop the early initiation of cracks, making up for the shortcomings of traditional technology in that it cannot measure strain or provide early warning.

[0022] Preferably, the objective function of the trend prediction module is:

[0023]

[0024] Where: J is the objective function value of the model predictive control, including the opening degree of the solenoid valve of the atomizing nozzle, the cooling water temperature, the speed of the circulating water pump, and the expansion or contraction of the insulation layer; N P For the predicted time domain, This is the predicted value of the crack risk index for the k-th step in the future. Let t be the predicted energy consumption value for the k-th future step, t be the current time, k be the time step index, α and β be the weighting coefficients, and α+β=1.

[0025] A method for intelligent control of cracks in impermeable concrete in sewage treatment ponds includes the following steps:

[0026] S1: Collect real-time data from each sensor and calculate the crack risk index R for each monitoring point according to the formula. i Generate a risk field cloud map within the pool;

[0027] S2: Sort all maintenance actions according to the risk index of the corresponding area from high to low, forming a priority queue for the action queue. If the strain rate d at a certain monitoring point... ε / d t If the threshold is exceeded, its risk index will be temporarily amplified and it will be placed at the top of the queue.

[0028] S3: Based on the rate of change of internal temperature of concrete d T / d t and strain rate d ε / d tThe coefficients are used to automatically determine the maintenance stage and adjust the dynamic weighting coefficients accordingly. , , , , ;

[0029] S4: Predict future time domain N using the LSTM model of the trend prediction module. P The temperature field, humidity field, and strain force evolution are used to obtain the minimum value of the objective function J and the action sequence.

[0030] S5: Execute the first action in the action sequence of step S4, wait for the next control cycle, and then return to step S1;

[0031] S6: The reinforcement training module retrains the LSTM model based on the data stored in the data storage module and sends the updated model parameters to the controller.

[0032] Preferably, the criteria for determining the curing stage in step S3 are as follows:

[0033] When the cooling rate d T / d t When the temperature is greater than 0 and the internal temperature has not yet reached its peak, it is determined to be in the heating period. At this time, the dynamic weighting... Set the weight to 0.5, and the sum of the remaining weights is 0.5;

[0034] When the cooling rate d T / d t <0 and |d T / d t When | > 0.5℃ / h, it is determined to be a cooling period, at which point the dynamic weighting... Set it to 0.6 and enable the strain amplification factor;

[0035] When the temperature difference between the inner and outer surfaces ΔT < 5℃ / h and the strain rate d ε / d t When the value is less than 1 με / h, it is determined to be in a stable period and switches to low-power intermittent maintenance mode.

[0036] Preferably, the prediction of the future time domain N in step S4 P The evolution of the control step size according to the rate of environmental change is as follows:

[0037] N in sunny, windy weather P The value is set to 2-4 hours, with a control step size of 5 minutes.

[0038] N on a cloudy, windless day P The value is 12h, and the control step is 30min.

[0039] Preferably, in step S4, the action sequence includes the opening degree of the atomizing nozzle solenoid valve, the water temperature in the cooling water pipe, the rotation speed of the circulating water pump, and the state of the covering film.

[0040] Compared with the prior art, the advantages of the present invention are as follows:

[0041] (1) By constructing a dynamic risk field, multiple parameters such as temperature gradient, cooling rate, strain, humidity, and wind speed are integrated into a single risk index R. i It also enables dynamic prioritization of maintenance actions, significantly improving the precision control of cracks;

[0042] (2) Using the LSTM model to predict the temperature field in the future, cooling or heat preservation can be actively increased before the temperature gradient increases sharply, avoiding control lag and eliminating the overshoot phenomenon of cooling rate.

[0043] (3) Compared with the traditional continuous maintenance mode, the method of on-demand spraying, frequency conversion cooling and automatic adjustment of the insulation layer has obvious energy-saving effect;

[0044] (4) Through dynamic weighting coefficients, it can be automatically adjusted according to the maintenance stage and environmental changes, making it suitable for use in complex environments;

[0045] (5) The reinforcement training module continuously optimizes the strategy using historical data. As the number of projects increases, the system regulation effect becomes better and better.

[0046] (6) Strain rate d ε / d t When the threshold is exceeded, the risk index is automatically amplified to achieve emergency response to "excessive strain," allowing for immediate intervention in the early stages of crack initiation and preventing crack propagation. Attached Figure Description

[0047] Figure 1 This is an overall architecture diagram provided for an embodiment of the present invention.

[0048] Figure 2 This is a diagram showing the internal structure of the base plate and side walls provided in an embodiment of the present invention.

[0049] Figure 3 The flowchart for dynamic risk field calculation provided in the embodiments of the present invention is shown. Detailed Implementation

[0050] The present invention will be further described below.

[0051] Example: An intelligent crack control system for impermeable concrete in a sewage treatment plant, see [link / reference]. Figures 1 to 3 It includes a sensing module, an execution module, and a control module. The sensing module includes a temperature sensor array, a humidity sensor, a vibrating wire strain gauge, and a weather station.

[0052] The temperature sensor array is embedded inside and on the surface of the concrete to collect the internal temperature T of the concrete. in and surface temperature T surf And according to the formula ΔT=T in -T surf Calculate the internal temperature T of the concrete. in and surface temperature T surf The internal surface temperature difference and cooling rate d T / d t (The first derivative of temperature with respect to time);

[0053] The humidity sensor is attached to the concrete surface to collect the relative humidity (RH) of the concrete surface; the vibrating wire strain gauge is embedded in the stress concentration area to collect the changes in tensile stress ε(t) and strain rate d in real time. ε / d t (The first derivative of strain with respect to time);

[0054] The on-site meteorological station collects information on ambient temperature (T), wind speed (v), solar radiation intensity, and rainfall.

[0055] The execution module includes an automatic sprinkler system, a temperature-controlled water circulation system, and an insulation layer covering device. The automatic sprinkler system includes multiple atomizing nozzles equipped with solenoid valves, which are evenly distributed across the wastewater treatment tank via water supply pipes. The temperature-controlled water circulation system includes cooling water pipes, a circulating water pump connected to the cooling water pipes, and a water tank with adjustable water temperature connected to the inlet of the circulating water pump; the cooling water pipes are embedded within the concrete. The insulation layer covering device includes an electric roller shutter and a covering film wrapped around the electric roller shutter, which can automatically unfold or retract the covering film.

[0056] The control module includes a controller, which performs overall control of the collected data calculation and execution module. The controller has a built-in dynamic risk field assessment module, trend prediction module, reinforcement training module, and data storage module.

[0057] The dynamic risk field assessment module calculates the cooling rate d based on the real-time collected inner surface temperature difference ΔT. T / d t Real-time acquisition of surface humidity (RH) and strain ratio (ε) i / ε u and ambient wind speed V wind Calculate the crack risk index R for each monitoring point using the following formula:

[0058]

[0059] Where: R i Let ΔT be the crack risk index for the i-th monitoring point. maxV max RH min V windmax ΔT is the preset reference value. i d represents the real-time internal surface temperature difference of the concrete. T / d t ε is the rate of cooling of the inner surface temperature of the concrete. i / ε u Let ε be the real-time tensile strain ratio at the i-th monitoring point. u For the ultimate tensile strain, RH i Let V be the surface relative humidity at the i-th monitoring point. wind For ambient wind speed, , , , , These are dynamic weighting coefficients, and The dynamic weighting coefficient is adaptively adjusted based on the concrete's age and curing stage. For example, it increases during the heating period. The cooling period is longer. Increased winds .

[0060] Then, a priority queue of maintenance actions is generated based on the calculated risk index, and the area with the highest risk index is given priority for execution; specifically as follows: when the strain rate d at a certain monitoring point... ε / d t When the risk index R of the monitoring point exceeds the set threshold, i The amplification factor enables emergency response when strain exceeds the limit. The amplification factor ranges from 1.5 to 2.5.

[0061] The trend prediction module predicts the changing trends of the concrete temperature field and crack risk index within a certain time domain based on an LSTM neural network. Within each control cycle, it solves for the action sequence that minimizes the objective function J. The objective function is:

[0062]

[0063] Where: J is the objective function value of the model predictive control, including the opening degree of the solenoid valve of the atomizing nozzle, the cooling water temperature, the speed of the circulating water pump, and the expansion or contraction of the insulation layer; N P For the predicted time domain, This is the predicted value of the crack risk index for the k-th step in the future. Let t be the predicted energy consumption value for the k-th future step, t be the current time, k be the time step index, α and β be the weighting coefficients, and α+β=1.

[0064] After finding the minimum action for J, the calculated data is transmitted to the controller for execution.

[0065] The enhanced training module trains a deep Q-network algorithm using historical maintenance data and a reward function that is the cumulative reduction in crack risk minus the energy consumption penalty. The training results are then transmitted to the controller for further optimization of the trend prediction module. The data storage module stores real-time maintenance data.

[0066] A method for intelligent control of cracks in impermeable concrete in sewage treatment ponds includes the following steps:

[0067] S1: Collect real-time data from each sensor and calculate the crack risk index R for each monitoring point according to the formula. i Generate a risk field cloud map within the pool;

[0068] S2: Sort all maintenance actions according to the risk index of the corresponding area from high to low, forming a priority queue for the action queue. If the strain rate d at a certain monitoring point... ε / d t If the threshold is exceeded, its risk index will be temporarily amplified and it will be placed at the top of the queue.

[0069] S3: Based on the rate of change of internal temperature of concrete d T / d t and strain rate d ε / d t The coefficients are used to automatically determine the maintenance stage and adjust the dynamic weighting coefficients accordingly. , , , , ;

[0070] The criteria for determining the maintenance stage are as follows:

[0071] When the cooling rate d T / d t When the temperature is greater than 0 and the internal temperature has not yet reached its peak, it is determined to be in the heating period. At this time, the dynamic weighting... Set the weight to 0.5, and the sum of the remaining weights is 0.5;

[0072] When the cooling rate d T / d t <0 and |d T / d t When | > 0.5℃ / h, it is determined to be a cooling period, at which point the dynamic weighting... Set it to 0.6 and enable the strain amplification factor;

[0073] When the temperature difference between the inner and outer surfaces ΔT < 5℃ / h and the strain rate d ε / d t When the value is less than 1 με / h, it is determined to be in a stable period and switches to low-power intermittent maintenance mode.

[0074] S4: Predict future time domain N using the LSTM model of the trend prediction module. P The temperature field, humidity field, and strain force evolution are used to obtain the minimum value of the objective function J and the action sequence.

[0075] Among them, predicting the future time domain N P The evolution of the control step size according to the rate of environmental change is as follows:

[0076] N in sunny, windy weather P The value is set to 2-4 hours, with a control step size of 5 minutes.

[0077] N on a cloudy, windless day P The value is 12h, and the control step is 30min.

[0078] The optimal action sequence includes the opening degree of the atomizing nozzle solenoid valve, the water temperature in the cooling water pipe, the rotation speed of the circulating water pump, and the state of the covering film.

[0079] S5: Execute the first action in the action sequence of step S4, such as turning on the spray in a certain area for 3 minutes, adjusting the cooling water temperature to a certain value, and then returning to step S1 after waiting for the next control cycle (5-30 min).

[0080] S6: The reinforcement training module retrains the LSTM model based on the data stored in the data storage module and sends the updated model parameters to the controller.

[0081] Specific implementation examples:

[0082] Taking the construction of a sewage treatment pond as an example, the dimensions of the sewage treatment pond are 40×20×6m, the thickness of the bottom slab is 1.2m, and the thickness of the side wall is 0.8m.

[0083] The implementation steps are as follows:

[0084] Step 1: Sensor pre-embedding: Install temperature sensors inside and on the surface of the base plate and side walls; install vibrating wire strain gauges at the junction of the base plate and side walls; install serpentine cooling water pipes inside the base plate and U-shaped cooling water pipes vertically inside the side walls; and pour concrete after installation.

[0085] Step 2, System Deployment: After the concrete is poured and formed, multiple humidity sensors are installed on the base slab and side walls; an automatic sprinkler system is laid on the poured sewage treatment tank; the water tank, circulating water pump, and cooling water pipes are connected in sequence to form a water circulation system; the covering film is laid on the poured concrete using an electric roller shutter; the controller and on-site weather station are transported to the site and connected.

[0086] Step 3, Maintenance Phase:

[0087] (1) 0-48 hours after pouring (heating period):

[0088] The internal temperature of the concrete rapidly rises to 65℃. The dynamic risk field assessment module calculates that the index of the central area of ​​the base slab is the highest based on the temperature difference ΔT between the inner and outer surfaces of the concrete. The trend prediction module predicts that the temperature difference ΔT between the inner and outer surfaces of the concrete in the central area of ​​the base slab will exceed 25℃ within the next 4 hours.

[0089] At this time, the controller controls the automatic spraying system to spray the central area of ​​the base plate, retracts the covering film, and starts the cooling water circulation (cooling water temperature 20℃, flow rate 1.5m / s).

[0090] Four hours later, the actual internal surface temperature difference ΔT of the concrete in the central area was 23℃, which did not exceed the warning line of 25℃.

[0091] (2) 48-168 hours after pouring (cooling period)

[0092] The peak internal temperature of the concrete begins to decrease, with an initial cooling rate of 3.2℃ / d. At this point, the dynamic weighting automatically switches. The strain amplification factor is activated when the temperature rises to 0.6. The LSTM model of the trend prediction module predicts that the cooling rate will remain above 3.0℃ / d for the next 8 hours. The optimal action determined by the trend prediction module is to gradually increase the cooling water temperature from 20℃ to 32℃ at a rate of 2℃ per hour, while reducing the spraying frequency of the automatic spray system.

[0093] At this point, the controller adjusts according to the optimal action;

[0094] After 8 hours, the actual cooling rate decreased to 1.5℃ / d, which meets the specification requirements.

[0095] (3) Night of the 3rd day (windy weather)

[0096] The weather station measured a wind speed of 8 m / s in the dynamic risk field. Automatically increases, while the risk index R... i Due to the increase in wind speed;

[0097] At this point, the controller controls the unfolding of the covering membrane and pauses the spraying to prevent the concrete surface from cooling too much due to evaporation;

[0098] The following morning, the wind speed dropped to 3 m / s, and the system automatically resumed spraying.

[0099] (4) Day 5 (Emergency Response and Early Warning)

[0100] A vibrating wire strain gauge at the junction of the wastewater treatment tank floor and sidewalls detected a sudden increase in strain rate to 8 με / h (threshold 5 με / h). At this point, the risk index R of this area... iBy temporarily multiplying by 1.8, this area jumps to the top of the action queue;

[0101] At this point, the controller immediately intensifies the spraying in the area and locally increases the thickness of the insulation layer (through multiple layers of electric roller shutters).

[0102] Two hours later, the strain rate dropped to 2 με / h, and the warning was lifted.

[0103] (5) After maintenance

[0104] The wastewater treatment tank showed no visible cracks, only minor surface textures.

[0105] Energy consumption statistics: Water consumption 28 tons, electricity consumption 320 kWh, routine maintenance water consumption approximately 45 tons, electricity consumption approximately 500 kWh;

[0106] The enhanced training module adjusted the spraying interval cycle after the 10th day of maintenance, further saving 8% of water.

[0107] The above provides a detailed description of the intelligent control system and method for cracks in impermeable concrete in sewage treatment ponds provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Modifications and improvements to the present invention are possible without exceeding the concept and scope specified in the appended claims. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A smart crack control system for impermeable concrete in sewage treatment ponds, characterized in that: The system includes a sensing module, an execution module, and a control module. The sensing module comprises a temperature sensor array, a humidity sensor, a vibrating wire strain gauge, and a weather station. The temperature sensor array is embedded inside and on the surface of the concrete to collect the internal temperature T of the concrete. in and surface temperature T surf The humidity sensor is attached to the concrete surface to collect the relative humidity (RH) of the concrete surface, and the vibrating wire strain gauge is embedded in the stress concentration area to collect the changes in tensile stress ε(t) and strain rate d in real time. ε / d t The on-site meteorological station collects information on ambient temperature T, wind speed v, solar radiation intensity, and rainfall. The execution module includes an automatic sprinkler system, a temperature-controlled water circulation system, and an insulation layer covering device. The automatic sprinkler system includes multiple atomizing nozzles with solenoid valves, which are evenly distributed on the sewage treatment tank through water supply pipes. The temperature-controlled water circulation system includes a cooling water pipe, a circulating water pump connected to the cooling water pipe, and a water tank with adjustable water temperature connected to the inlet of the circulating water pump. The cooling water pipe is embedded in the concrete. The insulation layer covering device includes an electric roller shutter and a covering film wrapped around the electric roller shutter, which can automatically unfold or retract the covering film. The control module includes a controller that performs overall control of the collected data calculation and execution module. The controller incorporates a dynamic risk field assessment module, a trend prediction module, a reinforcement training module, and a data storage module. The dynamic risk field assessment module is based on real-time collected data such as the inner surface temperature difference ΔT and the cooling rate d. T / d t Surface humidity (RH), strain ratio (ε) i / ε u and ambient wind speed V wind The system calculates the crack risk index R at each monitoring point and generates a priority queue of maintenance actions based on the calculated risk index. The area with the highest risk index is given priority for execution. The trend prediction module uses an LSTM neural network to predict the changing trends of the concrete temperature field and crack risk index over a certain time domain. Within each control cycle, it solves for the action sequence that minimizes the objective function J and transmits it to the controller for execution. The reinforcement training module uses historical maintenance data and a deep Q-network algorithm trained with the cumulative crack risk reduction minus energy consumption penalty as the reward function. The training results are transmitted to the controller for further optimization of the trend prediction module. The data storage module stores real-time maintenance data.

2. The intelligent crack control system for impermeable concrete in sewage treatment ponds according to claim 1, characterized in that: The internal surface temperature difference ΔT of the concrete is expressed by the formula ΔT=T in -T surf Perform the calculation.

3. The intelligent crack control system for impermeable concrete in sewage treatment ponds according to claim 1, characterized in that: The dynamic risk field assessment module calculates the crack risk index R using the following formula: Where: R i Let ΔT be the crack risk index for the i-th monitoring point. max V max RH min V windmax ΔT is the preset reference value. i d represents the real-time internal surface temperature difference of the concrete. T / d t ε is the rate of cooling of the inner surface temperature of the concrete. i / ε u Let ε be the real-time tensile strain ratio at the i-th monitoring point. u For the ultimate tensile strain, RH i Let V be the surface relative humidity at the i-th monitoring point. wind For ambient wind speed, , , , , These are dynamic weighting coefficients, and The dynamic weighting coefficient is adaptively adjusted according to the concrete age and curing stage.

4. The intelligent crack control system for impermeable concrete in sewage treatment ponds according to claim 3, characterized in that: When the strain rate d at a certain monitoring point ε / d t When the risk index R of the monitoring point exceeds the set threshold, i The amplification factor enables emergency response when strain exceeds the limit. The amplification factor ranges from 1.5 to 2.

5.

5. The intelligent crack control system for impermeable concrete in sewage treatment ponds according to claim 1, characterized in that: The optimization objective function of the trend prediction module is: Where: J is the objective function value of the model predictive control, including the opening degree of the solenoid valve of the atomizing nozzle, the cooling water temperature, the speed of the circulating water pump, and the expansion or contraction of the insulation layer; N P For the predicted time domain, This is the predicted value of the crack risk index for the k-th step in the future. Let t be the predicted energy consumption value for the k-th future step, t be the current time, k be the time step index, α and β be the weighting coefficients, and α+β=1.

6. A method for intelligent control of cracks in impermeable concrete in a wastewater treatment tank as described in any one of claims 1-5, characterized in that, Includes the following steps: S1: Collect real-time data from each sensor and calculate the crack risk index R for each monitoring point according to the formula. i Generate a risk field cloud map within the pool; S2: Sort all maintenance actions according to the risk index of the corresponding area from high to low, forming a priority queue for the action queue. If the strain rate d at a certain monitoring point... ε / d t If the threshold is exceeded, its risk index will be temporarily amplified and it will be placed at the top of the queue. S3: Based on the rate of change of internal temperature of concrete d T / d t and strain rate d ε / d t The coefficients are used to automatically determine the maintenance stage and adjust the dynamic weighting coefficients accordingly. , , , , ; S4: Predict future time domain N using the LSTM model of the trend prediction module. P The temperature field, humidity field, and strain force evolution are used to obtain the minimum value of the objective function J and the action sequence. S5: Execute the first action in the action sequence of step S4, wait for the next control cycle, and then return to step S1; S6: The reinforcement training module retrains the LSTM model based on the data stored in the data storage module and sends the updated model parameters to the controller.

7. The intelligent crack control method for impermeable concrete in sewage treatment ponds according to claim 6, characterized in that: The criteria for determining the maintenance stage in step S3 are as follows: When the cooling rate d T / d t When the temperature is greater than 0 and the internal temperature has not yet reached its peak, it is determined to be in the heating period. At this time, the dynamic weighting... Set the weight to 0.5, and the sum of the remaining weights is 0.5; When the cooling rate d T / d t <0 and |d T / d t When | > 0.5℃ / h, it is determined to be a cooling period, at which point the dynamic weighting... Set it to 0.6 and enable the strain amplification factor; When the temperature difference between the inner and outer surfaces ΔT < 5℃ / h and the strain rate d ε / d t When the value is less than 1 με / h, it is determined to be in a stable period and switches to low-power intermittent maintenance mode.

8. The intelligent crack control method for impermeable concrete in sewage treatment ponds according to claim 6, characterized in that: Predicting the future time domain N in step S4 P The evolution of the control step size based on the rate of environmental change is as follows: N in sunny, windy weather P The value is set to 2-4 hours, with a control step size of 5 minutes. N on a cloudy, windless day P The value is 12h, and the control step is 30min.

9. The intelligent crack control method for impermeable concrete in sewage treatment ponds according to claim 6, characterized in that: In step S4, the action sequence includes the opening degree of the atomizing nozzle solenoid valve, the water temperature in the cooling water pipe, the rotation speed of the circulating water pump, and the state of the covering film.

Images