Sewage treatment control system and method based on mixing of mechanism model and AI model

Abstract

The invention relates to the technical field of sewage treatment, in particular to a sewage treatment control system and method based on mixing of a mechanism model and an AI model. The server side comprises a real-time water inflow data acquisition module, a mechanism control module, a first control module, a judgment module, an AI training module, an AI control module, a selection matching module and an equipment control module, and the selection matching module is used for selecting the equipment control module when the corresponding data volume is larger than a first preset quantity threshold value and smaller than a second preset quantity threshold value. Calculating a model coverage rate corresponding to each process control scheme at the current moment and in a preset time period before the current moment, determining a substitution rate of a control parameter prediction value of each control device output by the model according to the corresponding model coverage rate, and determining a control parameter prediction value of each control device according to the corresponding substitution rate. And selecting the control parameter prediction value corresponding to the control equipment, and replacing the control parameter prediction value with the control parameter calculation value corresponding to the control equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of sewage treatment, and in particular to a sewage treatment control system and method based on a hybrid of a mechanism model and an AI model. Background Art

[0002] Sewage treatment systems are an important part of modern environmental protection. They are designed to remove pollutants from sewage through a series of physical, chemical and biological processes to meet discharge standards or reuse requirements. These systems usually include pretreatment, primary treatment (such as grit chambers, primary sedimentation tanks), secondary treatment (such as aeration tanks, secondary sedimentation tanks) and tertiary treatment (such as filtration, disinfection, etc.) to ensure that the final effluent quality meets strict environmental regulations. In order to ensure the efficient operation of the sewage treatment system and meet strict emission standards, an accurate and intelligent control system is essential. An effective control system can not only optimize the operating parameters of each treatment unit, but also respond to changes in the external environment in a timely manner to ensure stable operation of the system.

[0003] Existing sewage treatment systems mainly rely on two methods. One is to use mechanism models for control. Mechanism models use mathematical equations to describe the biochemical reaction processes in sewage, such as carbon removal, nitrogen removal and phosphorus removal, and provide theoretically accurate predictions. However, their high cost, regular correction requirements and design-oriented characteristics limit their flexibility in actual operation. The other is to use AI models for control. AI algorithms can handle complex nonlinear relationships and optimize control strategies by learning large amounts of historical data. They are suitable for real-time adjustments. However, their accuracy and reliability are highly dependent on the quality and quantity of data, and they lack the ability to make decisions and interpret processes for sudden problems, and have poor adaptability in different scenarios.

[0004] In summary, any single method adopted in the existing technology cannot well adapt to various sewage treatment scenarios and emergencies.

[0005] Based on this, there is an urgent need for a hybrid sewage treatment control system and method based on a mechanism model and an AI model, which can realize the updating of the mechanism model and the training of the AI ​​control model in the continuous data accumulation corresponding to the sewage treatment system, and then continuously optimize itself, use more diversified sewage treatment needs, and improve the operating efficiency and reliability of the sewage treatment system. Summary of the invention

[0006] One of the objects of the present invention is to provide a sewage treatment control system and method based on a hybrid mechanism model and AI model, which can realize automatic updating of the mechanism model, training of the AI ​​control model and combination with the mechanism model in the continuous data accumulation corresponding to the sewage treatment system, and then continuously self-optimize to adapt to more diversified sewage treatment needs and improve the operating efficiency and reliability of the sewage treatment system.

[0007] In order to achieve the above-mentioned purpose, a sewage treatment control system based on a hybrid mechanism model and an AI model is provided, including a server;

[0008] The server includes:

[0009] The real-time water inflow data collection module is used to monitor the water inflow data in the sewage treatment system in real time and form corresponding real-time water inflow data;

[0010] The mechanism control module is used to retrieve the process control scheme of the sewage treatment system corresponding to the previous moment, and according to the real-time water inflow data corresponding to the current moment and the process control scheme of the sewage treatment system corresponding to the previous moment, based on the sewage treatment mechanism model of the previous moment, the process control scheme of the sewage treatment system corresponding to the previous moment and the sewage treatment mechanism model are corrected in turn with corresponding parameters, and the process control scheme and sewage treatment mechanism model after parameter correction are used as the process control scheme and sewage treatment mechanism model corresponding to the sewage treatment system at the current moment;

[0011] The first control module is used to calculate the control parameter calculation value of the control device corresponding to each intermediate control parameter in the sewage treatment system through each intermediate control parameter in the process control scheme corresponding to the sewage treatment system at the current moment;

[0012] A judgment module, for performing statistics on the amount of data of the historical process control scheme corresponding to the sewage treatment system and the historical equipment control parameter value corresponding to the historical process control scheme in the historical database when the corresponding real-time water inflow data is monitored at the current moment, and judging the relationship between the corresponding amount of data and the first preset quantity threshold and the second preset quantity threshold, respectively, wherein the first preset quantity threshold is less than the second preset quantity threshold;

[0013] An AI training module, for retrieving a historical process control scheme corresponding to the sewage treatment system and a historical equipment control parameter value corresponding to the historical process control scheme when the corresponding data volume is greater than a first preset quantity threshold and less than a second preset quantity threshold, and constructing a training strategy based on a preset AI control model, constructing and training the AI ​​control model, and forming an AI control model after training;

[0014] The AI ​​control module is used to input various intermediate control parameters in the process control scheme corresponding to the sewage treatment system at the current moment into the trained AI control model after completing the training of the corresponding AI control model, and output the control parameter prediction values ​​of various control devices corresponding to various intermediate control parameters in the sewage system;

[0015] A selection matching module is used to select the calculated control parameter value as the device control parameter value corresponding to each control device when the corresponding data amount is less than a first preset quantity threshold;

[0016] When the corresponding data volume is greater than the first preset quantity threshold and less than the second preset quantity threshold, the model coverage rate corresponding to each process control scheme in the preset time period before the current moment is calculated, and according to the corresponding model coverage rate, the replacement rate of the control parameter prediction value of each control device output by the model is determined, and according to the corresponding replacement rate, the control parameter prediction value of the corresponding control device is selected and replaced with the control parameter calculation value corresponding to the control device to form the corresponding device control parameter value of each control device;

[0017] When the corresponding data volume is greater than a second preset quantity threshold, selecting the control parameter prediction value of each control device as the device control parameter value corresponding to each control device;

[0018] The equipment control module is used to control various control devices in the sewage treatment system according to the equipment control parameters of the corresponding control devices.

[0019] Technical principles and effects of this scheme: In this scheme, first, by real-time monitoring of the influent data in the sewage treatment system, after monitoring the corresponding real-time influent data, the process control scheme of the sewage treatment system corresponding to the previous moment is retrieved, and based on the sewage treatment mechanism model of the previous moment, the process control scheme of the sewage treatment system corresponding to the previous moment and the sewage treatment mechanism model are corrected with corresponding parameters in turn, and the process control scheme and sewage treatment mechanism model after parameter correction are used as the process control scheme and sewage treatment mechanism model corresponding to the sewage treatment system at the current moment. Through this step, the corresponding process control scheme and sewage treatment mechanism model at the current moment are corrected in turn.

[0020] Afterwards, the control parameter calculation values ​​of the control equipment corresponding to each intermediate control parameter in the sewage treatment system are calculated by using each intermediate control parameter in the process control scheme corresponding to the sewage treatment system at the current moment;

[0021] At the same time, when the corresponding real-time water inflow data is monitored at the current moment, the historical process control scheme corresponding to the sewage treatment system in the historical database and the historical equipment control parameter value corresponding to the historical process control scheme are statistically analyzed, and the relationship between the corresponding data volume and the first preset quantity threshold and the second preset quantity threshold is determined, and the first preset quantity threshold is less than the second preset quantity threshold;

[0022] When the corresponding data volume is greater than the first preset quantity threshold and less than the second preset quantity threshold, retrieve the historical process control scheme corresponding to the sewage treatment system and the historical equipment control parameter value corresponding to the historical process control scheme, build a training strategy based on the preset AI control model, build and train the AI ​​control model, and form an AI control model after training;

[0023] After completing the training of the corresponding AI control model, each intermediate control parameter in the process control scheme corresponding to the sewage treatment system at the current moment is used as input data and input into the trained AI control model, and the control parameter prediction value of each control device corresponding to each intermediate control parameter in the sewage system is output;

[0024] Depending on the amount of data, it is decided whether to use the calculated control parameter values ​​or the control parameter values ​​predicted by the AI ​​control model. When the amount of data is small, the mechanism model results are used first to calculate the control parameter values ​​of the corresponding control equipment. When the amount of data is sufficient, the control method dominated by the AI ​​control model is gradually transitioned to predict the control parameter predicted values ​​of the control equipment through the AI ​​control model. Finally, according to the different amounts of data, the selected control parameter values ​​are used to accurately control each control equipment in the sewage treatment system to ensure the efficiency and stability of the entire treatment process.

[0025] In this scheme, the sewage treatment mechanism model describes the biochemical reaction process in sewage through mathematical equations, providing theoretically accurate predictions. Combined with the empirical value calculation theory and safety factor, the rationality and safety of the control parameters in the initial stage are ensured. In the early stage of system operation, when the data accumulation is insufficient, the sewage treatment mechanism model provides a reliable theoretical basis, avoids control errors caused by insufficient data, and ensures the stable startup of the system. At the same time, the process control scheme and the sewage treatment mechanism model are corrected in sequence to further improve the accuracy of the control parameter calculation values ​​corresponding to the corresponding control equipment in the early stage. With the increase in the amount of data, the AI ​​control model gradually takes over some control tasks. The AI ​​control model can learn the complex patterns in historical data and capture the nonlinear relationships that may be ignored by the sewage treatment mechanism model, thereby providing more accurate predictions of control parameters. Through the intervention of the AI ​​control model, the conservative estimates caused by empirical values ​​and safety factors are reduced, making the control parameters closer to actual needs, and improving the treatment efficiency and effluent quality. The hybrid control system is not only suitable for sewage treatment plants of different sizes and types, but also can be adaptively adjusted according to different seasons and different working conditions, enhancing the universality of the system. The AI ​​control model can quickly respond to changes in the external environment and adjust the control strategy in a timely manner to ensure that the system is always in the best working state. That is, it can realize the automatic update of the mechanism model, the training of the AI ​​control model and the combination with the mechanism model in the continuous data accumulation corresponding to the sewage treatment system, and then continuously self-optimize, use more diverse sewage treatment needs, and improve the operating efficiency and reliability of the sewage treatment system.

[0026] Furthermore, the server also includes:

[0027] The water inlet data correction module is used to determine whether the real-time water inlet data corresponding to the current moment is abnormal based on the real-time water inlet data. If the judgment result is no, the real-time water inlet data at the current moment is judged to be accurate data. Otherwise, the real-time water inlet data monitored at the current moment is corrected, and the corrected real-time water inlet data is used as the real-time water inlet data at the current moment.

[0028] Beneficial effects: In this solution, the real-time water inlet data corresponding to the sewage treatment system may not be accurate during the monitoring process, and there may be data anomalies. Therefore, by judging the anomalies of the real-time water inlet data and automatically correcting them after the anomalies are judged, the accuracy and reliability of the real-time water inlet data collected by the sewage treatment system are greatly improved.

[0029] Further, the mechanism control module includes:

[0030] The first simulation module is used to simulate the sewage treatment of the sewage treatment system based on the real-time water inflow data corresponding to the current moment and the process control scheme corresponding to the sewage treatment system at the previous moment, and based on the sewage treatment mechanism model at the previous moment, and output the corresponding effluent water quality simulation data;

[0031] The process correction module is used to determine whether the corresponding effluent water quality meets the preset requirements based on the effluent water quality simulation data. If so, it is determined that the process control scheme corresponding to the sewage treatment system at the previous moment is a feasible scheme, and the process control scheme corresponding to the sewage treatment system at the previous moment is used as the process control scheme corresponding to the sewage treatment system at the current moment;

[0032] If not, then adjust the various intermediate control parameters in the process control scheme corresponding to the current moment of the sewage treatment system and re-input them into the sewage treatment mechanism model at the previous moment until the corresponding effluent water quality meets the preset requirements;

[0033] The model correction module is used to monitor the effluent water quality data corresponding to the sewage treatment system at the current moment after determining the process control scheme corresponding to the sewage treatment system at the current moment, and according to the effluent water quality data and the process control scheme corresponding to the sewage treatment system at the current moment, based on the preset model correction strategy, correct each model parameter in the sewage treatment mechanism model corresponding to the previous moment, and use the corrected sewage treatment mechanism model as the sewage treatment mechanism model corresponding to the current moment.

[0034] Beneficial effects: In this solution, based on the real-time influent data at the current moment and the process control scheme at the previous moment, the first simulation module can accurately simulate the sewage treatment process and output the corresponding effluent water quality simulation data. This function enables the system to foresee the treatment results before actual operation, so as to make adjustments in advance. By simulating the treatment effects under different control schemes, the system can find the optimal resource allocation method and reduce unnecessary energy and material consumption. According to the effluent water quality data obtained by simulation, it is quickly determined whether it meets the preset requirements. If it does, the process control scheme at the previous moment is confirmed as a feasible solution; otherwise, the parameters are adjusted immediately. This instant feedback mechanism ensures the response speed and flexibility of the system. By continuously adjusting until the preset standard is reached, the process correction module ensures the stability and consistency of the effluent water quality and reduces the probability of abnormal situations.

[0035] After determining the current process control scheme, the model correction module will automatically correct the parameters in the sewage treatment mechanism model according to the latest effluent water quality monitoring data and the corresponding control scheme. This process gives the system the ability to self-learn and continuously optimize. With the change of time and environment, the composition of sewage may change. The model correction module enables the system to gradually adapt to these changes and maintain long-term efficient operation.

[0036] Furthermore, the preset model correction strategy is:

[0037] After monitoring the effluent water quality data corresponding to the current moment, determine whether the corresponding effluent water quality data meets the preset water quality value. If so, determine that the sewage treatment mechanism model corresponding to the previous moment is a feasible model, and use the sewage treatment mechanism model as the sewage treatment mechanism model corresponding to the current moment;

[0038] If not, retrieve the historical effluent water quality data corresponding to the previous moment and the moment before that from the historical database, and retrieve the historical inlet water quality data corresponding to the effluent water quality data corresponding to the current moment, the previous moment and the moment before that;

[0039] According to the effluent water quality data at the current moment, the historical effluent water quality data corresponding to the previous moment and the moment before that, and the effluent water quality simulation data corresponding to each moment, the prediction error value of the sewage treatment mechanism model corresponding to the current moment is calculated, and it is determined whether the corresponding prediction error value is greater than the preset error threshold. If so, it is determined that the sewage treatment mechanism model at the previous moment needs to be adjusted, otherwise it does not need to be adjusted;

[0040] When the judgment result is that the sewage treatment mechanism model at the previous moment needs to be adjusted, the adjustment sensitivity of each model parameter in the model parameter group corresponding to the sewage treatment mechanism model at the previous moment is identified, and the adjustment sensitivities are sorted in descending order to form a sensitivity sorting table;

[0041] According to the sensitivity ranking table, select the model parameter with the highest adjustment sensitivity;

[0042] After selecting the corresponding intermediate control parameters, the model parameters are corrected and adjusted based on the preset adjustment amount. After the correction is completed, the sewage treatment mechanism model corresponding to the previous moment is updated, and the historical water inlet data corresponding to the effluent water quality data corresponding to the current moment and the process control scheme corresponding to the sewage treatment system at the current moment are input as input data into the updated sewage treatment mechanism model, and the corresponding effluent water quality re-simulation data is output;

[0043] Determine whether the simulation error between the effluent water quality re-simulation data and the effluent water quality monitoring data is less than the preset simulation error. If so, the corresponding corrected sewage treatment mechanism model is a feasible model;

[0044] If not, select the next model parameter with the highest adjustment sensitivity and repeat the correction and adjustment module until the effluent water quality re-simulation data and the effluent water quality monitoring data are within the preset simulation error. At this time, the corresponding corrected sewage treatment mechanism model is judged to be a feasible model.

[0045] Beneficial effects: This solution integrates the effluent water quality monitoring data of the current moment, the previous moment, and the moment before that, and combines the simulated effluent water quality data in the same time period with the historical inlet data corresponding to the effluent water quality monitoring data to calculate the prediction error value of the sewage treatment mechanism model at the previous moment. This process can not only accurately evaluate the feasibility of the sewage treatment mechanism model at that moment, but also provide a reliable basis for subsequent parameter adjustments.

[0046] In order to optimize the sewage treatment mechanism model more efficiently, we first identify the adjustment sensitivity of each model parameter and sort them from largest to smallest according to their impact on the effluent water quality. This method ensures that the parameters that have the greatest impact on the final water quality are adjusted first, thus achieving more accurate and efficient model optimization.

[0047] The selection module and the correction and adjustment module use a step-by-step optimization method to adjust the model parameters based on the sensitivity ranking table above. Each time, only one or several parameters with the greatest influence are fine-tuned, which avoids system instability that may be caused by modifying too many parameters at one time, while ensuring that each adjustment can bring significant improvements.

[0048] The iterative cycle module plays a key role in this process, ensuring that the entire optimization process continues until the simulated data of the effluent water quality reaches the preset standard. This closed-loop control mechanism not only ensures that the effluent water quality always meets the specified requirements, but also effectively reduces water quality fluctuations and maintains the stability and reliability of the system.

[0049] Furthermore, the AI ​​control model construction training strategy includes:

[0050] When the corresponding data volume is greater than a first preset quantity threshold and less than a second preset quantity threshold, retrieve the historical process control scheme corresponding to the sewage treatment system and the historical equipment control parameter value corresponding to the historical process control scheme;

[0051] Build an AI control model corresponding to the sewage treatment system;

[0052] Dividing a quantity interval corresponding to a first preset quantity threshold value and a second preset quantity threshold value into a plurality of sub-intervals;

[0053] When the corresponding data volume reaches a sub-interval, a random number B is randomly generated to determine whether the corresponding random number B is greater than the random threshold. If not, the corresponding data volume is determined not to be used. Otherwise, the corresponding data volume is determined to be used and the generation of random number B is stopped until the next sub-interval is reached.

[0054] When the judgment result is used for the corresponding data volume, the historical process control scheme corresponding to the sewage treatment system corresponding to the corresponding data volume and the historical equipment control parameter value corresponding to the historical process control scheme are used as the corresponding data set, and the data set is divided according to a preset ratio to form a training set and a verification set;

[0055] The historical process control scheme corresponding to the sewage treatment system in the training set is used as input data and input into the AI ​​control model trained in the previous sub-interval. The historical equipment control parameter values ​​corresponding to the historical process control scheme are used as the output data of the AI ​​control model to iteratively train the AI ​​control model.

[0056] After each training of the AI ​​control model is completed, the historical process control plan corresponding to the sewage treatment system in the verification set is input into the corresponding AI control model based on the trained AI control model, and the corresponding equipment parameter prediction values ​​are output. Based on the equipment parameter prediction values ​​and the historical equipment control parameter values ​​corresponding to the historical process control plan, the training effect of the corresponding AI control model is evaluated to determine whether the corresponding AI control model meets the standards. If so, the training is stopped and the corresponding AI control model is output as the corresponding trained AI control model; otherwise, the training continues.

[0057] Beneficial effect: The quantity interval between the first preset quantity threshold and the second preset quantity threshold is divided into multiple sub-intervals, so that the system can gradually introduce more historical data for training, avoiding the risk of overfitting caused by using too much data at one time.

[0058] By generating a random number B and judging whether it is greater than the random threshold, we decide whether to use the corresponding amount of data. This method not only increases the randomness of data selection, but also reduces unnecessary computational burden while ensuring data quality.

[0059] The AI ​​control model is iteratively trained using the historical process control solutions in the training set as input data and the corresponding historical equipment control parameter values ​​as output data. This end-to-end learning method can directly optimize the prediction accuracy of the model. After each training, the model performance is evaluated based on the validation set, and the decision on whether to continue training is made based on the evaluation results. This process ensures that the model continuously optimizes itself until the desired effect is achieved.

[0060] The present invention also provides a sewage treatment control method based on a hybrid of a mechanism model and an AI model, using the above-mentioned sewage treatment control system based on a hybrid of a mechanism model and an AI model. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a logical block diagram of a sewage treatment control system based on a hybrid mechanism model and an AI model in Example 1 of the present invention. DETAILED DESCRIPTION

[0062] The following is further described in detail through specific implementation methods:

[0063] Embodiment 1

[0064] Based on the hybrid sewage treatment control system of mechanism model and AI model, it is basically Figure 1 As shown, including the server;

[0065] The server includes:

[0066] The real-time water inflow data collection module is used to monitor the water inflow data in the sewage treatment system in real time and form corresponding real-time water inflow data;

[0067] The water inlet data correction module is used to determine whether the real-time water inlet data corresponding to the current moment is abnormal based on the real-time water inlet data. If the judgment result is no, the real-time water inlet data at the current moment is judged to be accurate data. Otherwise, the real-time water inlet data monitored at the current moment is corrected, and the corrected real-time water inlet data is used as the real-time water inlet data at the current moment. In this embodiment, when correcting the intermediate control parameters corresponding to the sewage treatment system at the previous moment, the corresponding parameters are corrected by the operator's parameter adjustment experience in combination with the real-time water inlet data at the current moment. For example, if a certain content in one of the real-time water inlet data is relatively high, the operator can know from experience that the corresponding adjustment of a certain intermediate control parameter can reduce the corresponding content. Of course, the system can also make automatic adjustments.

[0068] The mechanism control module is used to retrieve the process control scheme of the sewage treatment system corresponding to the previous moment, and according to the real-time water inlet data corresponding to the current moment and the process control scheme of the sewage treatment system corresponding to the previous moment, based on the sewage treatment mechanism model of the previous moment, the process control scheme of the sewage treatment system corresponding to the previous moment and the sewage treatment mechanism model are corrected in turn according to the corresponding parameters, and the process control scheme and sewage treatment mechanism model after parameter correction are used as the process control scheme and sewage treatment mechanism model corresponding to the sewage treatment system at the current moment; in this embodiment, the intermediate control parameters include the gas volume parameter, DO value, oxygen demand parameter, internal reflux ratio parameter, external reflux ratio parameter, sludge discharge parameter, flow parameter and dosage parameter corresponding to the sewage treatment system.

[0069] The mechanism control module includes:

[0070] The first simulation module is used to simulate the sewage treatment of the sewage treatment system based on the sewage treatment mechanism model at the previous moment according to the real-time water inflow data corresponding to the current moment and the process control scheme corresponding to the sewage treatment system at the previous moment, and output the corresponding effluent water quality simulation data; in this embodiment, before inputting into the sewage treatment mechanism model, when the real-time water inflow data at the current moment is accurate data, the real-time water inflow data is cleaned to ensure that the water inflow monitoring data enters the model in the format required by the algorithm. This step includes parameters for the distribution ratio of water inflow components (the percentage of each component). The sewage treatment plant with relatively complete online water inflow monitoring mainly monitors: COD / TN / TP / SS / NH4. The model needs to clean these data before use, and the ratio needs to be dynamically calibrated.

[0071] The process correction module is used to determine whether the corresponding effluent water quality meets the preset requirements based on the effluent water quality simulation data. If so, it is determined that the process control scheme corresponding to the sewage treatment system at the previous moment is a feasible scheme, and the process control scheme corresponding to the sewage treatment system at the previous moment is used as the process control scheme corresponding to the sewage treatment system at the current moment;

[0072] If not, the various intermediate control parameters in the process control scheme corresponding to the current moment of the sewage treatment system are adjusted and re-input into the sewage treatment mechanism model of the previous moment until the corresponding effluent water quality meets the preset requirements; in this embodiment, the corresponding adjustment of the various intermediate control parameters in the process control scheme corresponding to the current moment of the sewage treatment system is carried out using a preset comparison parameter adjustment model, and the comparison parameter adjustment model is constructed using the existing random forest algorithm.

[0073] The model correction module is used to monitor the effluent water quality data corresponding to the sewage treatment system at the current moment after determining the process control scheme corresponding to the sewage treatment system at the current moment, and according to the effluent water quality data and the process control scheme corresponding to the sewage treatment system at the current moment, based on the preset model correction strategy, correct each model parameter in the sewage treatment mechanism model corresponding to the previous moment, and use the corrected sewage treatment mechanism model as the sewage treatment mechanism model corresponding to the current moment. In this embodiment, the corresponding model parameters include the kinetic parameters and stoichiometric parameters of the mechanism model itself. These parameters change with the ambient temperature, season, and microbial growth conditions and need to be calibrated.

[0074] The preset model correction strategy is:

[0075] After monitoring the effluent water quality data corresponding to the current moment, determine whether the corresponding effluent water quality data meets the preset water quality value. If so, determine that the sewage treatment mechanism model corresponding to the previous moment is a feasible model, and use the sewage treatment mechanism model as the sewage treatment mechanism model corresponding to the current moment;

[0076] If not, retrieve the historical effluent water quality data corresponding to the previous moment and the moment before that from the historical database, and retrieve the historical inlet water quality data corresponding to the effluent water quality data corresponding to the current moment, the previous moment and the moment before that;

[0077] According to the effluent water quality data at the current moment, the historical effluent water quality data corresponding to the previous moment and the previous moment, and the effluent water quality simulation data corresponding to each moment, the prediction error value of the sewage treatment mechanism model corresponding to the current moment is calculated, and it is determined whether the corresponding prediction error value is greater than the preset error threshold. If so, it is determined that the sewage treatment mechanism model at the previous moment needs to be adjusted, otherwise no adjustment is required; in this embodiment, the prediction error value is calculated as the average corresponding to the sum of the absolute values ​​of the differences between the effluent water quality simulation data corresponding to each moment and the historical effluent water quality data.

[0078] When the judgment result is that the sewage treatment mechanism model at the previous moment needs to be adjusted, the adjustment sensitivity of each model parameter in the model parameter group corresponding to the sewage treatment mechanism model at the previous moment is identified, and the adjustment sensitivities are sorted in descending order to form a sensitivity sorting table;

[0079] According to the sensitivity ranking table, select the model parameter with the highest adjustment sensitivity;

[0080] After selecting the corresponding intermediate control parameters, the model parameters are corrected and adjusted based on the preset adjustment amount. After the correction is completed, the sewage treatment mechanism model corresponding to the previous moment is updated, and the historical water inlet data corresponding to the effluent water quality data corresponding to the current moment and the process control scheme corresponding to the sewage treatment system at the current moment are input as input data into the updated sewage treatment mechanism model, and the corresponding effluent water quality re-simulation data is output;

[0081] Determine whether the simulation error between the effluent water quality re-simulation data and the effluent water quality monitoring data is less than the preset simulation error. If so, the corresponding corrected sewage treatment mechanism model is a feasible model;

[0082] If not, select the next model parameter with the highest adjustment sensitivity and repeat the correction and adjustment module until the effluent water quality re-simulation data and the effluent water quality monitoring data are within the preset simulation error. At this time, the corresponding corrected sewage treatment mechanism model is judged to be a feasible model.

[0083] The first control module is used to calculate the control parameter calculation value of the control device corresponding to each intermediate control parameter in the sewage treatment system through each intermediate control parameter in the process control scheme corresponding to the sewage treatment system at the current moment; in this embodiment, the theoretical control quantity of each control device is calculated through each intermediate control parameter combined with the empirical value theory, and then the control parameter calculation value corresponding to each control device is calculated through a preset safety factor.

[0084] A judgment module, for performing statistics on the amount of data of the historical process control scheme corresponding to the sewage treatment system and the historical equipment control parameter value corresponding to the historical process control scheme in the historical database when the corresponding real-time water inflow data is monitored at the current moment, and judging the relationship between the corresponding amount of data and the first preset quantity threshold and the second preset quantity threshold, respectively, wherein the first preset quantity threshold is less than the second preset quantity threshold;

[0085] An AI training module, for retrieving a historical process control scheme corresponding to the sewage treatment system and a historical equipment control parameter value corresponding to the historical process control scheme when the corresponding data volume is greater than a first preset quantity threshold and less than a second preset quantity threshold, and constructing a training strategy based on a preset AI control model, constructing and training the AI ​​control model, and forming an AI control model after training;

[0086] The AI ​​control model construction training strategy includes:

[0087] When the corresponding data volume is greater than a first preset quantity threshold and less than a second preset quantity threshold, retrieve the historical process control scheme corresponding to the sewage treatment system and the historical equipment control parameter value corresponding to the historical process control scheme;

[0088] Build an AI control model corresponding to the sewage treatment system;

[0089] Dividing a quantity interval corresponding to a first preset quantity threshold value and a second preset quantity threshold value into a plurality of sub-intervals;

[0090] When the corresponding data volume reaches a sub-interval, a random number B is randomly generated to determine whether the corresponding random number B is greater than the random threshold. If not, the corresponding data volume is determined not to be used. Otherwise, the corresponding data volume is determined to be used and the generation of random number B is stopped until the next sub-interval is reached.

[0091] When the judgment result is used for the corresponding data volume, the historical process control scheme corresponding to the sewage treatment system corresponding to the corresponding data volume and the historical equipment control parameter value corresponding to the historical process control scheme are used as the corresponding data set, and the data set is divided according to a preset ratio to form a training set and a verification set;

[0092] The historical process control scheme corresponding to the sewage treatment system in the training set is used as input data and input into the AI ​​control model trained in the previous sub-interval. The historical equipment control parameter values ​​corresponding to the historical process control scheme are used as the output data of the AI ​​control model to iteratively train the AI ​​control model.

[0093] After each training of the AI ​​control model is completed, the historical process control plan corresponding to the sewage treatment system in the verification set is input into the corresponding AI control model based on the trained AI control model, and the corresponding equipment parameter prediction values ​​are output. Based on the equipment parameter prediction values ​​and the historical equipment control parameter values ​​corresponding to the historical process control plan, the training effect of the corresponding AI control model is evaluated to determine whether the corresponding AI control model meets the standards. If so, the training is stopped and the corresponding AI control model is output as the corresponding trained AI control model; otherwise, the training continues.

[0094] The AI ​​control module is used to input various intermediate control parameters in the process control scheme corresponding to the sewage treatment system at the current moment into the trained AI control model after completing the training of the corresponding AI control model, and output the control parameter prediction values ​​of various control devices corresponding to various intermediate control parameters in the sewage system;

[0095] A selection matching module is used to select the calculated control parameter value as the device control parameter value corresponding to each control device when the corresponding data amount is less than a first preset quantity threshold;

[0096] When the corresponding data volume is greater than the first preset quantity threshold and less than the second preset quantity threshold, the model coverage rate corresponding to each process control scheme in the preset time period before the current moment is calculated, and according to the corresponding model coverage rate, the replacement rate of the control parameter prediction value of each control device output by the model is determined, and according to the corresponding replacement rate, the control parameter prediction value of the corresponding control device is selected and replaced with the control parameter calculation value corresponding to the control device to form the corresponding device control parameter value of each control device;

[0097] When the corresponding data volume is greater than a second preset quantity threshold, selecting the control parameter prediction value of each control device as the device control parameter value corresponding to each control device;

[0098] The equipment control module is used to control each control device in the sewage treatment system according to the equipment control parameters of each corresponding control device. In this embodiment, when controlling each control device in the sewage treatment system, the DO value of the aerobic tank in the sewage treatment system is monitored, and it is determined whether the corresponding DO value meets the preset DO value. If not, based on the preset oxygen demand correction model, the oxygen demand parameters in each intermediate control parameter in the process control scheme calculated in the control module are corrected. In this embodiment, the oxygen demand correction model is also constructed using the existing random forest algorithm. According to the corresponding DO value, the required air volume is calculated and converted into the final operating frequency of the fan equipment. After executing the control parameter, whether the executed control parameter needs to be further adjusted according to the data feedback of the DO electrode in the aerobic tank.

[0099] This embodiment also discloses a sewage treatment control method based on a hybrid of a mechanism model and an AI model, using the above-mentioned sewage treatment control system based on a hybrid of a mechanism model and an AI model.

[0100] The above is only an embodiment of the present invention. The common sense such as the known specific structure and characteristics in the scheme will not be described in detail here. The ordinary technicians in the relevant field know all the common technical knowledge in the technical field to which the invention belongs before the application date or the priority date, can obtain all the existing technologies in the field, and have the ability to apply the conventional experimental means before that date. The ordinary technicians in the relevant field can improve and implement this scheme in combination with their own abilities under the enlightenment given by this application. Some typical known structures or known methods should not become obstacles for the ordinary technicians in the relevant field to implement this application. It should be pointed out that for the technicians in this field, without departing from the structure of the present invention, several deformations and improvements can be made, which should also be regarded as the protection scope of the present invention, which will not affect the effect of the implementation of the present invention and the practicality of the patent. The protection scope required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to explain the content of the claims.

Claims

1. A sewage treatment control system based on a hybrid mechanism model and AI model, characterized by: Including the server side; The server includes: The real-time water inflow data collection module is used to monitor the water inflow data in the sewage treatment system in real time and form corresponding real-time water inflow data; The mechanism control module is used to retrieve the process control scheme of the sewage treatment system corresponding to the previous moment, and according to the real-time water inflow data corresponding to the current moment and the process control scheme of the sewage treatment system corresponding to the previous moment, based on the sewage treatment mechanism model of the previous moment, the process control scheme of the sewage treatment system corresponding to the previous moment and the sewage treatment mechanism model are corrected in turn with corresponding parameters, and the process control scheme and sewage treatment mechanism model after parameter correction are used as the process control scheme and sewage treatment mechanism model corresponding to the sewage treatment system at the current moment; The first control module is used to calculate the control parameter calculation value of the control device corresponding to each intermediate control parameter in the sewage treatment system through each intermediate control parameter in the process control scheme corresponding to the sewage treatment system at the current moment; A judgment module, for performing statistics on the amount of data of the historical process control scheme corresponding to the sewage treatment system and the historical equipment control parameter value corresponding to the historical process control scheme in the historical database when the corresponding real-time water inflow data is monitored at the current moment, and judging the relationship between the corresponding amount of data and the first preset quantity threshold and the second preset quantity threshold, respectively, wherein the first preset quantity threshold is less than the second preset quantity threshold; An AI training module, for retrieving a historical process control scheme corresponding to the sewage treatment system and a historical equipment control parameter value corresponding to the historical process control scheme when the corresponding data volume is greater than a first preset quantity threshold and less than a second preset quantity threshold, and constructing a training strategy based on a preset AI control model, constructing and training the AI ​​control model, and forming an AI control model after training; The AI ​​control module is used to input various intermediate control parameters in the process control scheme corresponding to the sewage treatment system at the current moment into the trained AI control model after completing the training of the corresponding AI control model, and output the control parameter prediction values ​​of various control devices corresponding to various intermediate control parameters in the sewage system; A selection matching module is used to select the calculated control parameter value as the device control parameter value corresponding to each control device when the corresponding data amount is less than a first preset quantity threshold; When the corresponding data volume is greater than the first preset quantity threshold and less than the second preset quantity threshold, the model coverage rate corresponding to each process control scheme in the preset time period before the current moment is calculated, and according to the corresponding model coverage rate, the replacement rate of the control parameter prediction value of each control device output by the model is determined, and according to the corresponding replacement rate, the control parameter prediction value of the corresponding control device is selected and replaced with the control parameter calculation value corresponding to the control device to form the corresponding device control parameter value of each control device; When the corresponding data volume is greater than a second preset quantity threshold, selecting the control parameter prediction value of each control device as the device control parameter value corresponding to each control device; The equipment control module is used to control various control devices in the sewage treatment system according to the equipment control parameters of the corresponding control devices.

2. The sewage treatment control system based on the hybrid mechanism model and AI model according to claim 1 is characterized by: The server also includes: The water inlet data correction module is used to determine whether the real-time water inlet data corresponding to the current moment is abnormal based on the real-time water inlet data. If the judgment result is no, the real-time water inlet data at the current moment is judged to be accurate data. Otherwise, the real-time water inlet data monitored at the current moment is corrected, and the corrected real-time water inlet data is used as the real-time water inlet data at the current moment.

3. The sewage treatment control system based on the hybrid mechanism model and AI model according to claim 2 is characterized by: The mechanism control module includes: The first simulation module is used to simulate the sewage treatment of the sewage treatment system based on the real-time water inflow data corresponding to the current moment and the process control scheme corresponding to the sewage treatment system at the previous moment, and based on the sewage treatment mechanism model at the previous moment, and output the corresponding effluent water quality simulation data; The process correction module is used to determine whether the corresponding effluent water quality meets the preset requirements based on the effluent water quality simulation data. If so, it is determined that the process control scheme corresponding to the sewage treatment system at the previous moment is a feasible scheme, and the process control scheme corresponding to the sewage treatment system at the previous moment is used as the process control scheme corresponding to the sewage treatment system at the current moment; If not, then adjust the various intermediate control parameters in the process control scheme corresponding to the current moment of the sewage treatment system and re-input them into the sewage treatment mechanism model at the previous moment until the corresponding effluent water quality meets the preset requirements; The model correction module is used to monitor the effluent water quality data corresponding to the sewage treatment system at the current moment after determining the process control scheme corresponding to the sewage treatment system at the current moment, and according to the effluent water quality data and the process control scheme corresponding to the sewage treatment system at the current moment, based on the preset model correction strategy, correct each model parameter in the sewage treatment mechanism model corresponding to the previous moment, and use the corrected sewage treatment mechanism model as the sewage treatment mechanism model corresponding to the current moment.

4. The sewage treatment control system based on the hybrid mechanism model and AI model according to claim 3 is characterized by: The preset model correction strategy is: After monitoring the effluent water quality data corresponding to the current moment, determine whether the corresponding effluent water quality data meets the preset water quality value. If so, determine that the sewage treatment mechanism model corresponding to the previous moment is a feasible model, and use the sewage treatment mechanism model as the sewage treatment mechanism model corresponding to the current moment; If not, retrieve the historical effluent water quality data corresponding to the previous moment and the moment before that from the historical database, and retrieve the historical inlet water quality data corresponding to the effluent water quality data corresponding to the current moment, the previous moment and the moment before that; According to the effluent water quality data at the current moment, the historical effluent water quality data corresponding to the previous moment and the moment before that, and the effluent water quality simulation data corresponding to each moment, the prediction error value of the sewage treatment mechanism model corresponding to the current moment is calculated, and it is determined whether the corresponding prediction error value is greater than the preset error threshold. If so, it is determined that the sewage treatment mechanism model at the previous moment needs to be adjusted, otherwise it does not need to be adjusted; When the judgment result is that the sewage treatment mechanism model at the previous moment needs to be adjusted, the adjustment sensitivity of each model parameter in the model parameter group corresponding to the sewage treatment mechanism model at the previous moment is identified, and the adjustment sensitivities are sorted in descending order to form a sensitivity sorting table; According to the sensitivity ranking table, select the model parameter with the highest adjustment sensitivity; After selecting the corresponding intermediate control parameters, the model parameters are corrected and adjusted based on the preset adjustment amount. After the correction is completed, the sewage treatment mechanism model corresponding to the previous moment is updated, and the historical water inlet data corresponding to the effluent water quality data corresponding to the current moment and the process control scheme corresponding to the sewage treatment system at the current moment are input as input data into the updated sewage treatment mechanism model, and the corresponding effluent water quality re-simulation data is output; Determine whether the simulation error between the effluent water quality re-simulation data and the effluent water quality monitoring data is less than the preset simulation error. If so, the corresponding corrected sewage treatment mechanism model is a feasible model; If not, select the next model parameter with the highest adjustment sensitivity and repeat the correction and adjustment module until the effluent water quality re-simulation data and the effluent water quality monitoring data are within the preset simulation error. At this time, the corresponding corrected sewage treatment mechanism model is judged to be a feasible model.

5. The sewage treatment control system based on the hybrid mechanism model and AI model according to claim 4 is characterized by: The AI ​​control model construction training strategy includes: When the corresponding data volume is greater than a first preset quantity threshold and less than a second preset quantity threshold, retrieve the historical process control scheme corresponding to the sewage treatment system and the historical equipment control parameter value corresponding to the historical process control scheme; Build an AI control model corresponding to the sewage treatment system; Dividing a quantity interval corresponding to a first preset quantity threshold value and a second preset quantity threshold value into a plurality of sub-intervals; When the corresponding data volume reaches a sub-interval, a random number B is randomly generated to determine whether the corresponding random number B is greater than the random threshold. If not, the corresponding data volume is determined not to be used. Otherwise, the corresponding data volume is determined to be used and the generation of random number B is stopped until the next sub-interval is reached. When the judgment result is used for the corresponding data volume, the historical process control scheme corresponding to the sewage treatment system corresponding to the corresponding data volume and the historical equipment control parameter value corresponding to the historical process control scheme are used as the corresponding data set, and the data set is divided according to a preset ratio to form a training set and a verification set; The historical process control scheme corresponding to the sewage treatment system in the training set is used as input data and input into the AI ​​control model trained in the previous sub-interval. The historical equipment control parameter values ​​corresponding to the historical process control scheme are used as the output data of the AI ​​control model to iteratively train the AI ​​control model. After each training of the AI ​​control model is completed, the historical process control plan corresponding to the sewage treatment system in the verification set is input into the corresponding AI control model based on the trained AI control model, and the corresponding equipment parameter prediction values ​​are output. Based on the equipment parameter prediction values ​​and the historical equipment control parameter values ​​corresponding to the historical process control plan, the training effect of the corresponding AI control model is evaluated to determine whether the corresponding AI control model meets the standards. If so, the training is stopped and the corresponding AI control model is output as the corresponding trained AI control model; otherwise, the training continues.

6. A sewage treatment control method based on a hybrid mechanism model and an AI model, characterized by: A sewage treatment control system based on a hybrid of a mechanism model and an AI model using any one of claims 1 to 5 above.

Images