A self-adaptive intelligent dosing purification control system and method for contaminated groundwater

Abstract

The present invention relates to the technical field of contaminated groundwater treatment, and in particular to a self-adaptive intelligent dosing purification control system and method for contaminated groundwater. By acquiring groundwater quality parameters in real time, the pumping and dosing rates are intelligently adjusted according to monitoring data through a reinforcement learning model, so as to achieve real-time response to water quality changes and avoid unstable treatment effects and secondary pollution. The system uses a preset drug matching database to quickly identify excessive water quality parameters and intelligently match agents, accurately calculate the proportion of agent addition based on the degree of excess, and achieve precise dosing through a metering pump device. The reinforcement learning model takes water quality monitoring data, dosing treatment effects, pumping rates, dosing rates, and output and discharge rates as inputs, and outputs the optimal pumping and dosing control strategy through continuous learning and optimization. The dosing strategy is adjusted in real time according to water quality changes, effectively solving the problems of unstable treatment effects, waste of agents, and high costs in traditional contaminated groundwater treatment systems.

Description

Technical Field

[0001] The invention relates to the technical field of contaminated groundwater treatment, and in particular to a contaminated groundwater adaptive intelligent dosing purification control system and method. Background Art

[0002] Treatment of contaminated groundwater is a key link in environmental protection and sustainable resource utilization in modern society. With the rapid development of industrialization and urbanization, water pollution is becoming increasingly serious for plots of land where there is groundwater pollution and risk of spread in production enterprises. Traditional contaminated groundwater treatment systems usually adopt a fixed dosing strategy, that is, the dosing of agents is carried out according to experience or pre-set rules.

[0003] However, this method is difficult to adapt to the dynamic changes in the quality of contaminated groundwater, including: groundwater pollution often involves a variety of pollutants, including heavy metals, organic matter, ammonia nitrogen, etc., and the degree of pollution and the proportion of components vary with time and space. Fixed dosing strategies are difficult to accurately predict and respond to such complex and changeable water quality conditions, resulting in unstable treatment effects and difficulty in continuously ensuring that the effluent quality meets the standards, which in turn causes secondary pollution. Due to the lack of real-time monitoring and precise analysis of water quality, fixed dosing strategies tend to be conservative in dosing to ensure treatment results. However, this excessive dosing not only wastes pharmaceutical resources, but also increases treatment costs, bringing unnecessary economic burdens to companies. Summary of the invention

[0004] To solve the above problems, the present invention provides a contaminated groundwater adaptive intelligent dosing purification control system and method, which solves the problems of unstable treatment effect, waste of reagents and high cost in traditional contaminated groundwater treatment systems through real-time water quality monitoring, intelligent reagent matching and reinforcement learning optimization control.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] A contaminated groundwater adaptive intelligent dosing purification control system, comprising: a pumping module, a filtering module, a processing module, a purification detection module and an output module connected in sequence, and also comprising a water quality monitoring module, a drug dispensing and dosing module and an intelligent control module, wherein the drug dispensing and dosing module is respectively connected to the water quality monitoring module and the processing module, and the intelligent control module is respectively connected to the water quality monitoring module, the pumping module, the drug dispensing and dosing module and the purification detection module;

[0007] The pumping module is used to adjust the pumping rate and extract the groundwater at the current water level to the filtering module; the filtering module is used to filter the groundwater for impurities; the processing module is used to temporarily store the groundwater after impurity filtering and allow the dosing module to dosing the groundwater; the purification detection module is used to obtain the detection parameters of the water after dosing and determine whether it meets the discharge parameter standards; the output module is used to output and discharge the water that meets the discharge parameter standards; the water quality monitoring module is used to obtain the detection parameters of the current groundwater level;

[0008] The medicine dispensing and adding module is used to extract the exceeding items according to the detection parameters generated by the water quality monitoring module, match the contaminated groundwater treatment drugs according to the exceeding items, and mix several kinds of contaminated groundwater treatment drugs and output them to the processing module;

[0009] The intelligent control module is used to construct a reinforcement learning model based on the groundwater detection parameters of the current water level, the detection parameters of the water after drug treatment, the pumping rate, the drug addition rate and the output discharge rate, set the reward function through the difference between the detection parameters of the water after drug treatment and the discharge parameter standard, the output discharge rate and the drug addition cost, train the reinforcement learning model through the reward function, and output the control strategy of the pumping rate and the drug addition rate in real time through the reinforcement learning model that has completed the training.

[0010] Furthermore, the matching of contaminated groundwater treatment drugs according to the exceeded items, mixing several contaminated groundwater treatment drugs and outputting them to the processing module includes the following steps:

[0011] Extracting the exceeding items from the categories of detection parameters generated by the water quality monitoring module, wherein the categories of detection parameters include pH value, turbidity, heavy metal ion concentration, organic matter concentration, ammonia nitrogen concentration, total phosphorus concentration and total nitrogen concentration;

[0012] According to the extracted excessive items, the contaminated groundwater treatment drugs corresponding to the excessive items are traversed in the preset drug matching database;

[0013] According to the several types of contaminated groundwater treatment drugs obtained through traversal, the addition ratio of each contaminated groundwater treatment drug is calculated in combination with the degree of exceeding the standard, and each contaminated groundwater treatment drug is transported to the treatment module according to the addition ratio through a metering pump device.

[0014] Furthermore, the calculation of the addition ratio of each contaminated groundwater treatment drug includes the following steps:

[0015] Based on the processing capabilities of each contaminated groundwater treatment drug for the category items of the detection parameters, each contaminated groundwater treatment drug is preset with an initialization weight;

[0016] The ratio of the product of the degree of exceeding the standard for each item exceeding the standard and the initialization weight is calculated to obtain the addition ratio of each contaminated groundwater treatment drug.

[0017] Furthermore, the step of constructing a reinforcement learning model includes:

[0018] The detection parameters of the current groundwater level, the detection parameters of the water after drug treatment, the pumping rate, the drug addition rate and the output discharge rate are used as the state variables of the reinforcement learning model to construct the state space;

[0019] The adjustment of pumping rate and dosing rate is used as the action of the reinforcement learning model to construct the action space;

[0020] The reward function is constructed by the difference between the detection parameters of the water after the treatment and the discharge parameter standard, the output discharge rate and the cost of the treatment;

[0021] A deep Q network model is constructed based on the state space, action space and reward function, the pumping rate and dosing rate are adjusted, and the model parameters are updated according to the reward function until the model converges.

[0022] Furthermore, the formula of the reinforcement learning model is as follows:

[0023] ;

[0024] in, is the state parameter; is the action parameter; is the reward function; For the state parameter Take action parameters Q value; is the learning rate; is the discount factor; is the next state parameter; Next action parameters; Next state parameter Take the next action parameter The maximum Q value.

[0025] Furthermore, the formula of the reward function is as follows:

[0026] ;

[0027] in, is the reward function value; It is the detection parameter set of water after drug addition treatment; This is the test parameter set for water after treatment with chemicals. Item detection parameters; is the emission parameter standard set; Emission parameter standard set Standard values ​​of emission parameters; is the total number of emission parameters; The cost of the reagent used in the dosing process for the current preset time period; is the preset maximum drug cost; , and is the weight coefficient; is the output emission rate; The preset maximum output discharge rate.

[0028] Furthermore, the drug cost is calculated and generated by a preset drug addition rate per unit time and an addition ratio of each contaminated groundwater treatment drug.

[0029] Furthermore, it also includes an energy-saving module, which is used to suspend the operation of the system when the detection parameters of the current groundwater level meet the emission parameter standards and continue for a preset time.

[0030] A contaminated groundwater adaptive intelligent dosing purification control method is applied to any of the above-mentioned contaminated groundwater adaptive intelligent dosing purification control systems, comprising the following steps:

[0031] Extract groundwater at the current water level and filter out impurities;

[0032] Get the detection parameters of the current groundwater level;

[0033] Extract the exceeding items according to the detection parameters of the current groundwater level, match the contaminated groundwater treatment drugs according to the exceeding items, and add drugs to the groundwater after impurities are filtered after mixing several contaminated groundwater treatment drugs;

[0034] Obtain the test parameters of water after drug treatment;

[0035] A reinforcement learning model is constructed based on the groundwater detection parameters of the current water level, the detection parameters of the water after the dosing treatment, the pumping rate, the dosing rate and the output discharge rate. The reward function is set by the difference between the detection parameters of the water after the dosing treatment and the discharge parameter standard, the output discharge rate and the dosing cost. The reinforcement learning model is trained by the reward function. The control strategy of the pumping rate and the dosing rate is output in real time through the completed reinforcement learning model.

[0036] The water that meets the discharge parameter standards after chemical treatment is discharged.

[0037] The beneficial effects of the present invention are as follows: the present invention uses a water quality monitoring module to monitor the groundwater level and water quality in real time and obtain key water quality parameters. Based on the monitoring data, the intelligent control module uses a reinforcement learning model to make intelligent decisions to achieve dynamic adjustment of the pumping rate and the dosing rate, thereby accurately responding to water quality changes and avoiding unstable treatment effects and secondary pollution risks. Through a preset drug matching database, the excessive water quality parameters are quickly identified and the corresponding agents are intelligently matched. At the same time, combined with the degree of exceeding the standard, the addition ratio of each agent is intelligently calculated, and accurate dosing is achieved through a metering pump device to avoid excessive dosing of agents and reduce treatment costs. In the reinforcement learning model, water quality monitoring data, dosing treatment effect, pumping rate, dosing rate and output discharge rate are used as inputs to output the optimal pumping and dosing control strategy. This adaptive control strategy can be adjusted in real time according to water quality changes, improve treatment efficiency, and effectively solve the problems of unstable treatment effect, waste of agents and high cost in traditional contaminated groundwater treatment systems, and achieve efficient, accurate and economical treatment of groundwater pollution. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a structural schematic diagram of a self-adaptive intelligent dosing purification control system for contaminated groundwater in the present invention.

[0039] Figure 2 It is a flow chart of the steps of a self-adaptive intelligent dosing purification control method for contaminated groundwater in the present invention. DETAILED DESCRIPTION

[0040] See also Figure 1-2 As shown, the present invention relates to a self-adaptive intelligent dosing and purification control system for contaminated groundwater, comprising: a pumping module, a filtering module, a processing module, a purification detection module and an output module connected in sequence, and also comprising a water quality monitoring module, a dispensing and dosing module and an intelligent control module, wherein the dispensing and dosing module is respectively connected to the water quality monitoring module and the processing module, and the intelligent control module is respectively connected to the water quality monitoring module, the pumping module, the dispensing and dosing module and the purification detection module;

[0041] The pumping module is used to adjust the pumping rate and extract the groundwater at the current water level to the filtering module; the filtering module is used to filter the groundwater for impurities; the processing module is used to temporarily store the groundwater after impurity filtering and allow the dosing module to dosing the groundwater; the purification detection module is used to obtain the detection parameters of the water after dosing and determine whether it meets the discharge parameter standards; the output module is used to output and discharge the water that meets the discharge parameter standards; the water quality monitoring module is used to obtain the detection parameters of the current groundwater level;

[0042] The medicine dispensing and adding module is used to extract the exceeding items according to the detection parameters generated by the water quality monitoring module, match the contaminated groundwater treatment drugs according to the exceeding items, and mix several kinds of contaminated groundwater treatment drugs and output them to the processing module;

[0043] The intelligent control module is used to construct a reinforcement learning model based on the groundwater detection parameters of the current water level, the detection parameters of the water after drug treatment, the pumping rate, the drug addition rate and the output discharge rate, set the reward function through the difference between the detection parameters of the water after drug treatment and the discharge parameter standard, the output discharge rate and the drug addition cost, train the reinforcement learning model through the reward function, and output the control strategy of the pumping rate and the drug addition rate in real time through the reinforcement learning model that has completed the training.

[0044] In some embodiments, first, groundwater is extracted to the treatment system on the ground through the pumping module, and groundwater extraction can be achieved through pumping devices such as pumping pumps. This module can not only monitor the groundwater level in real time, but also dynamically adjust the pumping rate to ensure that the system can stably treat groundwater at different water levels. The extracted groundwater first enters the filtration module, which includes a multi-layer filter and an adsorption medium, which can effectively remove suspended matter, silt and some organic matter in the water to ensure the effectiveness of subsequent dosing treatment. The filtered water enters the treatment module, where it is temporarily stored and waits for further dosing treatment. The treatment module is closely connected with the dispensing and dosing module, which generates the current water quality detection parameters based on the real-time data from the water level and water quality monitoring module. The system identifies the pollutant items exceeding the standard in the water by analyzing these parameters, and selects the corresponding agent from the preset database of contaminated groundwater treatment drugs. After the selected agent is accurately proportioned, it is added to the water body in the treatment module for treatment. This process ensures that different types of pollutants can be treated in a targeted manner, so that the treated water can meet the discharge standards. After the dosing treatment is completed, the water flows into the purification detection module for quality inspection. This module detects multiple parameters of the treated water through a test device, including but not limited to pH value, heavy metal ion concentration, organic matter content, etc. The system compares these detection parameters with the preset discharge standards to determine whether the treated water meets the discharge requirements. If the water quality meets the standard, the system will discharge the water through the output module; if the water quality does not meet the standard, the intelligent control module will automatically adjust the reagent ratio in the dosing module and treat it again. The intelligent control module is the core of the entire system and makes real-time decisions based on the reinforcement learning model. The module uses multi-dimensional data obtained from the water level and water quality monitoring module, the dosing module, and the purification detection module to dynamically adjust the pumping rate and dosing strategy. The reinforcement learning model can accurately predict the trend of water quality changes under different treatment conditions through training with a large amount of historical data, and adjust the control strategy according to the water quality data detected in real time. The training process of the model includes setting a reward function, which takes into account the treated water quality parameters, dosing costs, and treatment efficiency. The model learns how to minimize the use of reagents and energy consumption while ensuring that the water quality meets the standard through continuous optimization. Ultimately, through the intelligent control of the system, efficient purification of groundwater can be achieved under various complex water quality conditions. The system has a high degree of adaptability and optimization capabilities, and can continuously ensure that the discharged water quality meets the standards under different environmental and water quality conditions. The automation characteristics of the system reduce the need for manual intervention, greatly improve the stability and controllability of the treatment process, and are suitable for large-scale, continuous contaminated groundwater treatment scenarios.

[0045] Furthermore, the matching of contaminated groundwater treatment drugs according to the exceeded items, mixing several contaminated groundwater treatment drugs and outputting them to the processing module includes the following steps:

[0046] Extracting the exceeding items from the categories of detection parameters generated by the water quality monitoring module, wherein the categories of detection parameters include pH value, turbidity, heavy metal ion concentration, organic matter concentration, ammonia nitrogen concentration, total phosphorus concentration and total nitrogen concentration;

[0047] According to the extracted excessive items, the contaminated groundwater treatment drugs corresponding to the excessive items are traversed in the preset drug matching database;

[0048] According to the several types of contaminated groundwater treatment drugs obtained through traversal, the addition ratio of each contaminated groundwater treatment drug is calculated in combination with the degree of exceeding the standard, and each contaminated groundwater treatment drug is transported to the treatment module according to the addition ratio through a metering pump device.

[0049] In some embodiments, when contaminated groundwater enters the system, the water quality monitoring module starts to collect water quality data in real time, and obtains key parameters including pH value, turbidity, heavy metal ion concentration, organic matter concentration, ammonia nitrogen concentration, total phosphorus concentration and total nitrogen concentration. These parameters reflect the actual concentration of various pollutants in the contaminated groundwater. The system first analyzes these data and extracts the exceeding items, that is, the categories of pollutants that exceed the discharge standard. Each type of exceeding item is a target that the system needs to focus on, and determining these exceeding items is the basis for subsequent dosing treatment. After determining the exceeding items, the system automatically accesses the preset drug matching database. The database stores effective contaminated groundwater treatment drugs for various pollutants and their corresponding treatment capacity data. The system matches the extracted exceeding items one by one, and searches the database for the treatment drugs corresponding to each exceeding item. For each exceeding item, the system may find multiple applicable drugs, which can reduce the concentration of the pollutant to a qualified level separately or jointly. Subsequently, the system calculates the specific addition ratio of each contaminated groundwater treatment drug according to the degree of exceeding the standard of each exceeding item, that is, the excess ratio of the concentration of the pollutant relative to the discharge standard. Once the addition ratio of each drug is determined, the system accurately delivers each contaminated groundwater treatment drug to the treatment module according to the calculated ratio through a metering pump device. The metering pump device has control capabilities, which can ensure that each drug is added to the contaminated groundwater at the required flow rate and amount, thereby achieving effective treatment of contaminated groundwater. The mixing and reaction process in the treatment module will further ensure the full reaction of the drug with the pollutants in the contaminated groundwater to achieve the effect of removing pollutants. Through the above steps, the system can flexibly respond to different types of contaminated groundwater pollution, and automatically adjust the type and amount of the drug added when the water quality changes. This embodiment shows how the system can achieve efficient purification of multiple pollutants through intelligent matching and precise measurement in the process of complex contaminated groundwater treatment, thereby ensuring that the final discharged water quality meets environmental protection standards. This process not only improves the treatment efficiency of the system, but also significantly reduces the use of agents and operating costs, and is suitable for large-scale industrial contaminated groundwater treatment scenarios.

[0050] Furthermore, the calculation of the addition ratio of each contaminated groundwater treatment drug includes the following steps:

[0051] Based on the processing capabilities of each contaminated groundwater treatment drug for the category items of the detection parameters, each contaminated groundwater treatment drug is preset with an initialization weight;

[0052] The ratio of the product of the degree of exceeding the standard for each item exceeding the standard and the initialization weight is calculated to obtain the addition ratio of each contaminated groundwater treatment drug.

[0053] It should be noted that it is necessary to preset initialization weights for each contaminated groundwater treatment drug for each category of detection parameters (such as pH value, turbidity, heavy metal ion concentration, etc.) based on the pre-established drug matching database and expert knowledge. These weights reflect the theoretical treatment capacity of the agent for a specific pollutant. The higher the weight, the stronger the treatment capacity of the agent for the pollutant. The setting of the initialization weight can be based on the chemical properties, reaction mechanism, laboratory or field test data of the agent, etc. For each exceeded item, the system calculates its degree of exceedance. The degree of exceedance can be defined as the difference between the actual detection value and the emission standard. For each exceeded item and each candidate agent, the system multiplies the degree of exceedance by the initialization weight of the agent for the exceeded item to obtain a product result. The system adds all the product results to obtain a sum. Then, each product result is divided by the sum to obtain the addition ratio of each agent. This calculation method ensures that the sum of the addition ratios of each agent is 1, and at the same time makes the addition ratio proportional to the degree of exceedance and the agent treatment capacity.

[0054] Furthermore, the step of constructing a reinforcement learning model includes:

[0055] The detection parameters of the current groundwater level, the detection parameters of the water after drug treatment, the pumping rate, the drug addition rate and the output discharge rate are used as the state variables of the reinforcement learning model to construct the state space;

[0056] The adjustment of pumping rate and dosing rate is used as the action of the reinforcement learning model to construct the action space;

[0057] The reward function is constructed by the difference between the detection parameters of the water after the treatment and the discharge parameter standard, the output discharge rate and the cost of the treatment;

[0058] A deep Q network model is constructed based on the state space, action space and reward function, the pumping rate and dosing rate are adjusted, and the model parameters are updated according to the reward function until the model converges.

[0059] In some embodiments, first, the system uses the detection parameters of the groundwater level, the detection parameters of the water after the dosing treatment, the pumping rate, the dosing rate and the output discharge rate as the state variables of the reinforcement learning model. These variables together constitute the state space of the system. The construction of the state space is the basis for the operation of the model, which comprehensively describes the operation status of the system at any time. The detection parameters of the groundwater level reflect the actual condition of the water source, while the detection parameters of the water after the dosing treatment directly affect whether the final discharge standard can be met. The pumping rate, the dosing rate and the output discharge rate represent the dynamic control parameters of the system. Through the real-time adjustment of these parameters, the system can respond immediately to changes in water quality. Then, the system uses the adjustment of the pumping rate and the dosing rate as the action of the reinforcement learning model to construct the action space. In the action space, each action represents the specific operation that the system can take in the current state, including different combinations of pumping rates and dosing rates. These actions will directly affect the water treatment process and the final water quality compliance. By selecting the optimal action combination, the system can continuously optimize the efficiency of water treatment. To achieve model learning and optimization, the system designed a reward function, which is calculated based on the difference between the detection parameters of the water treated with dosing and the discharge parameter standards, the output discharge rate, and the dosing cost. Specifically, when the detection parameters of the water treated with dosing are closer to the discharge standard and the dosing cost and the output discharge rate reach the optimal balance, the system will obtain a higher reward value. On the contrary, if the treatment effect is poor or the cost is too high, the reward value will decrease. The design of this reward function is to guide the system to gradually optimize the treatment strategy by adjusting the pumping rate and the dosing rate under different operating conditions, so as to minimize resource consumption while ensuring that the water quality meets the standard. On this basis, the system uses the state space, action space and reward function to construct a deep Q network model (DQN). The deep Q network model continuously adjusts the pumping rate and the dosing rate by training a large amount of historical data and real-time data. During the training process, the model updates its internal parameters by comparing the current Q value with the optimal Q value of the next state. As training progresses, the model’s predictive capabilities gradually improve and eventually converge to a stable state, enabling the system to output optimal pumping and dosing strategies in real time under a variety of complex water quality environments.

[0060] Furthermore, the formula of the reinforcement learning model is as follows:

[0061] ;

[0062] in, is the state parameter; is the action parameter; is the reward function; For the state parameter Take action parameters Q value; is the learning rate; is the discount factor; is the next state parameter; Next action parameters; Next state parameter Take the next action parameter The maximum Q value.

[0063] It should be noted that the state parameter , is the state of the system at a certain moment. The combination of the detection parameters of the current groundwater level, the detection parameters of the water after the treatment, the pumping rate, the dosing rate and the output discharge rate is the state of the system. Action parameters , is the specific action that the system can take in the current state, including adjusting the pumping rate or the dosing rate. The system affects the subsequent state by selecting the appropriate action. Reward function It is used to measure the effect of a system taking a specific action in a certain state. The design of the reward function is based on the deviation between the detection parameters of the treated water and the discharge standard, the cost of dosing, and the output discharge rate. It is the driving factor of reinforcement learning and helps the system optimize decisions through rewards or penalties. Indicates that the state parameter Take action parameters The expected cumulative reward that can be obtained. Q value is the core of reinforcement learning. The system gradually approaches the optimal strategy by updating the Q value. The learning rate is a parameter between 0 and 1 that controls the speed at which the system learns new information. A higher learning rate makes the system more sensitive to new experiences, while a lower learning rate means that the system relies more on past experiences. The discount factor, also between 0 and 1, is used to measure the impact of future rewards. A higher discount factor means that the system pays more attention to long-term benefits, while a lower discount factor means that the system pays more attention to short-term effects. Next state parameter , indicating that the system is taking action The new state reached. The next action parameter indicates the system is in a new state The following actions can be taken. The maximum expected reward is used to estimate the optimal future return, thereby guiding the update of the current Q value. In the process of reinforcement learning, the system continuously collects running data, calculates and updates the Q value. Whenever the system is in a certain state parameter It evaluates various possible action parameters , and selects an optimal action based on the current Q value. After taking this action, the system enters the next state parameter , and receive instant rewards , the system uses the above formula to adjust the corresponding Q value to make its prediction more accurate.

[0064] Furthermore, the formula of the reward function is as follows:

[0065] ;

[0066] in, is the reward function value; It is the detection parameter set of water after drug addition treatment; This is the test parameter set for water after treatment with chemicals. Item detection parameters; is the emission parameter standard set; Emission parameter standard set Standard values ​​of emission parameters; is the total number of emission parameters; The cost of the reagent used in the dosing process for the current preset time period; is the preset maximum drug cost; , and is the weight coefficient; is the output emission rate; The preset maximum output discharge rate.

[0067] It should be noted that the reward function value It is the comprehensive evaluation value obtained by the system after performing a specific action in a certain state. The higher the reward value, the better the decision of the system in this state, and vice versa, it means that the decision needs to be optimized. , which is the specific indicator of water quality after treatment, such as pH value, heavy metal concentration, and standardized value of turbidity. Since the magnitude of the parameters varies, each parameter needs to be standardized first, and the standardized values ​​in the value range are replaced to achieve the same dimension accumulation. This parameter is used to evaluate whether the treated water quality is close to or meets the discharge standard. is the set of emission parameter standards, representing the standardized values ​​of compliance standards for various water quality indicators. Emission parameter standard set Standard values ​​of emission parameters. This is the test parameter set for water after treatment with chemicals. The system evaluates the water quality compliance by comparing the difference between the water quality parameters after the treatment and the discharge standard. The cost of the reagent used in the treatment process during the current preset time period This parameter reflects the economic cost of the chemical agents consumed by the system during the treatment process. The lower the cost, the higher the reward value while ensuring that the water quality meets the standard. The preset maximum agent cost It is usually determined during the system design phase and represents the upper limit of the cost of the agent under specific processing conditions. This value is used to normalize the effect of the agent cost on the reward function. and , corresponding to the weights of water quality compliance and reagent cost respectively. These coefficients are used to balance the relative importance of the two objectives in the total reward. By adjusting the weight coefficients, the system can focus on water quality compliance or cost control in different operating environments. Output emission rate, It indicates the discharge rate of water after treatment by the system. A high discharge rate means that the system has high treatment efficiency and can treat more wastewater. is the preset maximum output emission rate, which represents the maximum emission rate that the system can achieve under optimal conditions. Through the above reward function, the system tends to choose those operating strategies that can both meet water quality standards and reduce the cost of reagents in each decision. With the continuous training and optimization of the reinforcement learning model, the system gradually learns to automatically adjust the pumping and dosing strategies under various complex operating conditions, so as to achieve the best treatment effect and economic benefits. This design ensures the efficiency, stability and sustainability of the contaminated groundwater treatment process.

[0068] Furthermore, the drug cost is calculated and generated by a preset drug addition rate per unit time and an addition ratio of each contaminated groundwater treatment drug.

[0069] Specifically, first, the system presets the dosing rate per unit time, which represents the total amount of drugs added to the contaminated groundwater per unit time during the treatment process. The cost of each contaminated groundwater treatment drug is determined by its addition ratio and dosing rate. The cost calculation not only takes into account the number of drugs, but also comprehensively considers the unit price and treatment effect of different drugs. In this way, the system can make the best choice based on economy and effectiveness in actual operation.

[0070] Furthermore, it also includes an energy-saving module, which is used to suspend the operation of the system when the detection parameters of the current groundwater level meet the emission parameter standards and continue for a preset time.

[0071] In some embodiments, specifically, the energy-saving module is closely integrated with the water level and water quality monitoring module to monitor the changes in groundwater level and water quality in real time. When the detection parameters of groundwater, including but not limited to pH value, turbidity, heavy metal ion concentration, etc., have met the preset emission parameter standards, and this state lasts for a preset period of time, the energy-saving module will automatically trigger and temporarily stop the further processing operation of the system. This preset time is set according to the system design requirements and operating experience to ensure that after the water quality is stable and meets the standards, energy can be saved by suspending the system operation without affecting the continued compliance of the overall emission standards. The specific working process of the energy-saving module is as follows: When it is detected that the water quality parameters of the groundwater are stable within the emission standard range, the system will start a timer to record the duration of this state. If there is no significant fluctuation or sign of exceeding the standard in the water quality parameters within the preset time, the energy-saving module will issue an instruction to suspend the operation of the key processing units of the system, such as the pumping module, the dosing module and the purification detection module. At the same time, the system will maintain basic monitoring functions to ensure that if the water quality changes, the system can quickly resume operation to avoid any potential risk of exceeding the emission standard. During the system suspension, the energy-saving module still monitors the water level and water quality. If the water quality parameters are detected to fluctuate, or the detection parameters exceed the standard again, the system will immediately resume operation to re-perform the necessary treatment operations to ensure that the discharged water quality continues to meet the standard. In addition, the energy-saving module can also work with the intelligent control module to analyze the cumulative time of system suspension and evaluate the energy-saving effect of the system under different operating conditions, providing data support for further optimization of energy-saving strategies.

[0072] The present invention also includes a method for controlling the self-adaptive intelligent dosing and purification of contaminated groundwater, comprising the following steps:

[0073] Extract groundwater at the current water level and filter out impurities;

[0074] Get the detection parameters of the current groundwater level;

[0075] Extract the exceeding items according to the detection parameters of the current groundwater level, match the contaminated groundwater treatment drugs according to the exceeding items, and add drugs to the groundwater after impurities are filtered after mixing several contaminated groundwater treatment drugs;

[0076] Obtain the test parameters of water after drug treatment;

[0077] A reinforcement learning model is constructed based on the groundwater detection parameters of the current water level, the detection parameters of the water after the dosing treatment, the pumping rate, the dosing rate and the output discharge rate. The reward function is set by the difference between the detection parameters of the water after the dosing treatment and the discharge parameter standard, the output discharge rate and the dosing cost. The reinforcement learning model is trained by the reward function. The control strategy of the pumping rate and the dosing rate is output in real time through the completed reinforcement learning model.

[0078] The water that meets the discharge parameter standards after chemical treatment is discharged.

[0079] The above implementation modes are merely descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary engineering and technical personnel in the field shall fall within the protection scope determined by the claims of the present invention.

Claims

1. An adaptive intelligent dosing and purification control system for contaminated groundwater, characterized in that: include: The pumping module, the filtering module, the processing module, the purification detection module and the output module are connected in sequence, and also include a water quality monitoring module, a medicine dispensing and dosing module and an intelligent control module, wherein the medicine dispensing and dosing module is respectively connected to the water quality monitoring module and the processing module, and the intelligent control module is respectively connected to the water quality monitoring module, the pumping module, the medicine dispensing and dosing module and the purification detection module; The pumping module is used to adjust the pumping rate and extract the groundwater at the current water level to the filtering module; the filtering module is used to filter the groundwater for impurities; the processing module is used to temporarily store the groundwater after impurity filtering and allow the dosing module to dosing the groundwater; the purification detection module is used to obtain the detection parameters of the water after dosing and determine whether it meets the discharge parameter standards; the output module is used to output and discharge the water that meets the discharge parameter standards; the water quality monitoring module is used to obtain the detection parameters of the current groundwater level; The medicine dispensing and adding module is used to extract the exceeding items according to the detection parameters generated by the water quality monitoring module, match the contaminated groundwater treatment drugs according to the exceeding items, and mix several kinds of contaminated groundwater treatment drugs and output them to the processing module; The intelligent control module is used to build a reinforcement learning model based on the groundwater detection parameters of the current water level, the detection parameters of the water after the dosing treatment, the pumping rate, the dosing rate and the output discharge rate, set a reward function through the difference between the detection parameters of the water after the dosing treatment and the discharge parameter standard, the output discharge rate and the dosing cost, train the reinforcement learning model through the reward function, and output the control strategy of the pumping rate and the dosing rate in real time through the reinforcement learning model that has completed the training; The steps of constructing the reinforcement learning model include: The detection parameters of the current groundwater level, the detection parameters of the water after drug treatment, the pumping rate, the drug addition rate and the output discharge rate are used as the state variables of the reinforcement learning model to construct the state space; The adjustment of pumping rate and dosing rate is used as the action of the reinforcement learning model to construct the action space; The reward function is constructed by the difference between the detection parameters of the water after the treatment and the discharge parameter standard, the output discharge rate and the cost of the treatment; Build a deep Q network model based on the state space, action space and reward function, adjust the pumping rate and dosing rate, and update the model parameters according to the reward function until the model converges; The formula of the reinforcement learning model is as follows: ； in, are state parameters, including detection parameters of the current groundwater level, detection parameters of water after treatment, pumping rate, dosing rate and output discharge rate; Action parameters include adjusting the pumping rate or the dosing rate; is the reward function; For the state parameter Take action parameters Q value; is the learning rate; is the discount factor; is the next state parameter; Next action parameters; Next state parameter Take the next action parameter The maximum Q value when The formula of the reward function is as follows: ； in, is the reward function value; It is the detection parameter set of water after drug addition treatment; This is the test parameter set for water after treatment with chemicals. Item detection parameters; is the emission parameter standard set; Emission parameter standard set Standard values ​​of emission parameters; is the total number of emission parameters; The cost of the reagent used in the dosing process for the current preset time period; is the preset maximum drug cost; , and is the weight coefficient; is the output emission rate; The preset maximum output discharge rate.

2. According to claim 1, a contaminated groundwater adaptive intelligent dosing purification control system is characterized in that: The method of matching the contaminated groundwater treatment drugs according to the exceeded items, mixing several contaminated groundwater treatment drugs and outputting them to the processing module comprises the following steps: Extracting the exceeding items from the categories of detection parameters generated by the water quality monitoring module, wherein the categories of detection parameters include pH value, turbidity, heavy metal ion concentration, organic matter concentration, ammonia nitrogen concentration, total phosphorus concentration and total nitrogen concentration; According to the extracted excessive items, the contaminated groundwater treatment drugs corresponding to the excessive items are traversed in the preset drug matching database; According to the several types of contaminated groundwater treatment drugs obtained through traversal, the addition ratio of each contaminated groundwater treatment drug is calculated in combination with the degree of exceeding the standard, and each contaminated groundwater treatment drug is transported to the treatment module according to the addition ratio through a metering pump device.

3. The self-adaptive intelligent dosing and purification control system for contaminated groundwater according to claim 2 is characterized in that: The calculation of the addition ratio of each contaminated groundwater treatment drug comprises the following steps: Based on the processing capabilities of each contaminated groundwater treatment drug for the category items of the detection parameters, each contaminated groundwater treatment drug is preset with an initialization weight; The ratio of the product of the degree of exceeding the standard for each item exceeding the standard and the initialization weight is calculated to obtain the addition ratio of each contaminated groundwater treatment drug.

4. The self-adaptive intelligent dosing and purification control system for contaminated groundwater according to claim 1 is characterized in that: The drug cost is calculated and generated by presetting the drug addition rate per unit time and the addition ratio of each contaminated groundwater treatment drug.

5. The self-adaptive intelligent dosing and purification control system for contaminated groundwater according to claim 1 is characterized in that: It also includes an energy-saving module, which is used to suspend the operation of the system when the detection parameters of the current groundwater level meet the emission parameter standards and continue for a preset time.

6. A method for controlling the self-adaptive intelligent dosing and purification of contaminated groundwater, applied to a control system for self-adaptive intelligent dosing and purification of contaminated groundwater as claimed in any one of claims 1 to 5, characterized in that: The following steps are involved: Extract groundwater at the current water level and filter out impurities; Get the detection parameters of the current groundwater level; Extract the exceeding items according to the detection parameters of the current groundwater level, match the contaminated groundwater treatment drugs according to the exceeding items, and add drugs to the groundwater after impurities are filtered after mixing several contaminated groundwater treatment drugs; Obtain the test parameters of water after drug treatment; A reinforcement learning model is constructed based on the groundwater detection parameters of the current water level, the detection parameters of the water after the dosing treatment, the pumping rate, the dosing rate and the output discharge rate. The reward function is set by the difference between the detection parameters of the water after the dosing treatment and the discharge parameter standard, the output discharge rate and the dosing cost. The reinforcement learning model is trained by the reward function. The control strategy of the pumping rate and the dosing rate is output in real time through the completed reinforcement learning model. The water that meets the discharge parameter standards after chemical treatment is discharged.

Images