Intelligent catch basin cloud model control method

Abstract

The invention discloses an intelligent intercepting well cloud model control method, and relates to the technical field of sewage treatment and drainage, and the control method utilizes an intelligent intercepting well cloud model control system to realize remote monitoring and intelligent control of the running state of an intercepting well. The system comprises a sensing acquisition unit, a communication unit, a cloud processing unit, a control unit and a remote server, and the cloud processing unit is used for receiving the monitoring data, selecting a corresponding operation mode according to a user demand, processing the monitoring data and generating a corresponding control instruction. According to the invention, various monitoring data in the catch basin can be accurately acquired in real time through the sensing acquisition unit, and an accurate data decision basis is provided for the cloud processing unit, so that the cloud processing unit can intelligently select an operation mode according to user requirements and environmental changes, and the safety of the catch basin is improved. Therefore, it is ensured that the intercepting well can still operate efficiently and stably under the complex and changeable working conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of sewage treatment and drainage, and in particular to a cloud-based model control method for an intelligent interception well. Background Art

[0002] As an important part of the urban drainage system, interception wells undertake important functions such as sewage interception and rainwater and sewage separation. However, with the acceleration of urbanization, the pressure on urban drainage systems is increasing. Traditional interception wells have gradually shown a series of problems in actual use. These problems affect their drainage efficiency and sewage interception results, which are specifically manifested in the following aspects:

[0003] (1) Manual control: Traditional interception wells are highly dependent on manual operation, which not only requires a large number of on-site operators, but also requires extremely high timeliness of regulation and professionalism of operators due to the wide distribution of smart interception wells and frequent fluctuations in water quality and quantity. This leads to high labor costs and makes it difficult to operate in real time according to actual conditions around the clock, making it difficult to ensure a good rainwater and sewage diversion effect.

[0004] (2) Rainfall-liquid level control: Although the PLC system can achieve a good dynamic response through the linkage of rainfall and liquid level sensors with gates and submersible sewage pumps (if configured), it is difficult to achieve accurate rainwater and sewage diversion by relying solely on liquid level control when the concentration of incoming water is uncertain. This may cause sewage overflow or a large amount of rainwater to mix into the sewage treatment plant, reducing the influent concentration of the sewage treatment plant and affecting the treatment effect.

[0005] (3) Rainfall-liquid level-water quality control: After online monitoring of water quality, linkage with rainfall, liquid level sensors, gates, and submersible sewage pumps (if configured) can further improve the accuracy of dynamic response. However, limited by the computing power of the PLC system and the information island phenomenon, the system cannot accurately judge complex water quality and quantity conditions, nor can it effectively link with pump stations, drainage systems, and sewage treatment plants, thus affecting the overall control effect.

[0006] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0007] In view of the problems in the related art, the present invention proposes a cloud-based model control method for an intelligent interception well to overcome the above-mentioned technical problems existing in the existing related art.

[0008] To this end, the specific technical solution adopted by the present invention is as follows:

[0009] A cloud model control method for an intelligent interception well, the control method uses an intelligent interception well cloud model control system to achieve remote monitoring and intelligent control of the operation status of the interception well, the system comprising: a sensing and collection unit, a communication unit, a cloud processing unit, a control unit and a remote server;

[0010] A sensing and collecting unit, used to collect monitoring data of the interception well in real time using sensors inside the interception well;

[0011] A communication unit, used to transmit the collected monitoring data to a cloud processing unit via an Internet of Things network;

[0012] The cloud processing unit is used to receive monitoring data, select the corresponding operation mode according to user needs, process the monitoring data, and generate corresponding control instructions;

[0013] A control unit, used to formulate and execute corresponding control strategies according to the control instructions generated by the cloud processing unit;

[0014] The remote server is used to monitor the operation status of the interception well in real time through the user interface and adjust the control strategy according to the operation status.

[0015] Furthermore, the cloud processing unit includes: a data storage and management module, a data preprocessing module, an operation mode selection module, a data analysis module and a control instruction generation module;

[0016] A data storage and management module is used to receive monitoring data and store the monitoring data in a database for user query and management;

[0017] A data preprocessing module is used to preprocess the monitoring data in the database;

[0018] An operation mode selection module is used to select a corresponding operation mode according to user needs. The operation modes include a water quantity restriction mode and a water quantity and water quality restriction mode.

[0019] The data analysis module is used to analyze the pre-processed monitoring data and historical monitoring data using the random forest algorithm to obtain monitoring analysis results;

[0020] The control instruction generation module is used to generate corresponding control instructions according to the selection of the operation mode and the results of monitoring and analysis.

[0021] Furthermore, the data preprocessing module includes: an outlier processing submodule, a missing data filling submodule and a normalization processing submodule;

[0022] An outlier processing submodule is used to detect outliers in monitoring data using an outlier detection algorithm and replace them with the average value of monitoring data at adjacent moments of the outliers;

[0023] A missing data filling submodule is used to fill in the missing data in the monitoring data after abnormal processing using a missing data filling algorithm;

[0024] The normalization processing submodule is used to normalize the padded monitoring data using a normalization algorithm.

[0025] Furthermore, using a missing data filling algorithm to fill in missing data in the monitoring data after abnormal processing includes:

[0026] S11. According to the completeness of the time series data, the monitoring data are divided into complete data sets and missing data sets;

[0027] S12, using a weighting algorithm to calculate the weights of the monitoring data attributes in the complete data set, and based on the weight analysis results, determine whether there are missing attributes in the complete data set. If not, the process ends; if so, execute step S13;

[0028] S13, according to the nth attribute of the missing data, calculate the grey correlation coefficient between the nth attribute and the nth attribute of any data point in the complete data set;

[0029] S14. Calculate the grey correlation value between the missing data and the complete data based on the grey correlation coefficient;

[0030] S15, sorting the grey correlation values ​​according to a preset standard, and selecting a number of complete data with the highest grey correlation values ​​as candidate nearest neighbors for filling the missing data;

[0031] S16, calculating the average value of several candidate nearest neighbors, obtaining the missing attributes of the missing data, and filling the missing attribute values. After the filling is completed, return to step S12 to continue processing new missing data.

[0032] Furthermore, the calculation formula of the grey relational coefficient is:

[0033]

[0034] In the formula, T represents the grey correlation coefficient between the nth attribute of missing data and the nth attribute of complete data;

[0035] j represents the number of complete data;

[0036] n represents the attribute of the monitoring data;

[0037] c indicates missing data;

[0038] d indicates complete data;

[0039] λ represents the adjustment parameter;

[0040] |c in -d jn | indicates missing data c i With complete data d j The absolute value difference of the nth attribute.

[0041] Furthermore, the data analysis module includes: a data set establishment submodule, a model initialization submodule, a model training and optimization submodule, an index calculation submodule, and an analysis result acquisition submodule;

[0042] The data set establishment and feature extraction submodule is used to obtain historical monitoring data from the database, establish a historical monitoring data set, and extract historical feature parameters from the historical monitoring data set;

[0043] The model initialization submodule is used to combine historical feature parameters with the random forest algorithm to initialize the random forest model;

[0044] The model training and optimization submodule is used to train and optimize the random forest model based on the historical monitoring data set using the adaptive learning rate optimization algorithm;

[0045] The index calculation submodule is used to input the pre-processed monitoring data into the optimized random forest model, output the water quality prediction results, and calculate the comprehensive water quality index;

[0046] The analysis result acquisition submodule is used to obtain the monitoring and analysis results based on the monitoring data and the comprehensive water quality index.

[0047] Furthermore, based on the historical monitoring data set, the adaptive learning rate optimization algorithm is used to train and optimize the random forest model, including:

[0048] S21, input the historical monitoring data set into the random forest model for training, and calculate the loss function value;

[0049] S22, back-propagating the loss function to calculate the gradient of the random forest model parameters relative to the loss function;

[0050] S23, using the adaptive learning rate optimization algorithm to calculate the square of the current gradient, and updating the cache variables of the adaptive learning rate optimization algorithm;

[0051] S24. Calculate a new learning rate according to the current state of the adaptive learning rate optimization algorithm, and use the new learning rate to update the parameters of the random forest model;

[0052] S25. Repeat steps S21 to S24 until a preset number of iterations is reached.

[0053] Furthermore, the pre-processed monitoring data is input into the optimized random forest model to output the water quality prediction results and calculate the comprehensive water quality index including:

[0054] S31, input the pre-processed monitoring data into the optimized random forest model, and output the water quality prediction result;

[0055] S32. Based on the prediction results, design the hierarchical structure of the analytic hierarchy process, which includes the target layer, the criterion layer and the solution layer;

[0056] S33, according to the preprocessed monitoring data, the elements in the criterion layer and the scheme layer are compared in pairs, and a judgment matrix between the layers is constructed based on the comparison results;

[0057] S34. Calculate the weights of the elements at the solution layer using the eigenvector algorithm, and calculate the comprehensive water quality index based on the weights of the elements at the solution layer and the monitoring data.

[0058] Furthermore, the calculation formula of the comprehensive water quality index is:

[0059]

[0060] In the formula, Q represents the comprehensive index of water quality and pollution;

[0061] g represents the number of water quality parameters;

[0062] i represents the index;

[0063] ω i represents the weight of the i-th water quality parameter;

[0064] x i represents the actual measured value of the i-th water quality parameter;

[0065] x i * represents the ideal value of the i-th water quality parameter.

[0066] Further, the control instruction generation module includes: a water quality and water quantity assessment submodule and a marking and instruction generation submodule;

[0067] The water quality and quantity assessment submodule is used to assess whether the water quality and quantity meet the standards based on the water quantity in the monitoring data and the calculated water quality and pollution comprehensive index, combined with the preset threshold;

[0068] The marking and instruction generation submodule is used to classify and mark the working status of the interception well according to the evaluation results of water quantity and water quality, and generate corresponding control instructions in combination with the selected operation mode.

[0069] The beneficial effects of the present invention are:

[0070] 1. The present invention can obtain various monitoring data inside the interception well in real time and accurately through the sensing and collection unit, and provide an accurate data decision-making basis for the cloud processing unit, which enables the cloud processing unit to intelligently select the operation mode according to user needs and environmental changes, thereby ensuring that the interception well can still maintain efficient and stable operation under complex and changeable working conditions.

[0071] 2. The present invention not only receives and processes the monitoring data from the sensing and collecting unit through the cloud processing unit, but also generates corresponding control instructions according to the processing results. These control instructions are transmitted to the control unit in real time through the communication unit, and the control unit formulates and executes corresponding control strategies according to the instructions. This close linkage between the cloud and the site realizes real-time adjustment and precise control of the operating status of the interception well, effectively improving the automation level and operating efficiency of the system.

[0072] 3. Through the remote server, the present invention allows users to monitor the operating status of the interception well in real time through the user interface at any time and anywhere, and adjust the control strategy according to the operating status, which not only improves the flexibility and convenience of the system, but also reduces the operation and maintenance costs and improves the operation and maintenance efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0074] Figure 1 It is a principle block diagram of a cloud-based model control system for an intelligent interception well according to an embodiment of the present invention;

[0075] Figure 2 It is a composition diagram of a cloud-based model control system for an intelligent interception well according to an embodiment of the present invention;

[0076] Figure 3 It is a flow chart of a cloud-based model control method for an intelligent interception well according to an embodiment of the present invention.

[0077] In the figure:

[0078] 1. Sensing and collecting unit; 2. Communication unit; 3. Cloud processing unit; 4. Control unit; 5. Remote server. DETAILED DESCRIPTION

[0079] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments. They can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in this field should be able to understand other possible implementation methods and advantages of the present invention.

[0080] According to an embodiment of the present invention, a cloud-based model control method for an intelligent interception well is provided.

[0081] The present invention will now be further described with reference to the accompanying drawings and specific embodiments. Figure 1-Figure 3 As shown, according to the intelligent interception well cloud model control method of the embodiment of the present invention, the control method uses the intelligent interception well cloud model control system to realize remote monitoring and intelligent control of the operation status of the interception well; the system includes: a sensing and collecting unit 1, a communication unit 2, a cloud processing unit 3, a control unit 4 and a remote server 5;

[0082] The sensing and collecting unit 1 is used to collect monitoring data of the interception well in real time using sensors inside the interception well.

[0083] It should be noted that the sensors inside the interception well include rainfall sensors, water level sensors, water quality sensors, etc.; the monitoring data include rainfall, water level, water quality, water volume, pH value, timestamp, etc.

[0084] The communication unit 2 is used to transmit the collected monitoring data to the cloud processing unit 3 through the Internet of Things network.

[0085] It should be noted that the sensor should be connected to the Internet of Things network, which can be achieved through wireless communication technologies such as Wi-Fi, Bluetooth, LoRa, NB-IoT, etc. The collected monitoring data is sent to the cloud processing unit 3 through the Internet of Things network, which usually involves using a specific communication protocol (such as MQTT, CoAP) to send the data to the cloud server.

[0086] The cloud processing unit 3 is used to receive monitoring data, select a corresponding operation mode according to user needs, process the monitoring data, and generate corresponding control instructions.

[0087] Specifically, the cloud processing unit 3 includes: a data storage and management module, a data preprocessing module, an operation mode selection module, a data analysis module and a control instruction generation module;

[0088] The data storage and management module is used to receive monitoring data and store the monitoring data in a database for user query and management.

[0089] The data preprocessing module is used to preprocess the monitoring data in the database.

[0090] Specifically, the data preprocessing module includes: an outlier processing submodule, a missing data filling submodule and a normalization processing submodule;

[0091] The outlier processing submodule is used to detect outliers in the monitoring data using an outlier detection algorithm and replace them with the average value of the monitoring data at adjacent moments of the outlier.

[0092] It should be noted that the outlier detection algorithm is used to detect outliers in the monitoring data, and the outliers are replaced by the average value of the monitoring data at adjacent moments of the outliers, including:

[0093] Calculate the standardized score of each monitoring data, then,

[0094]

[0095] Where Z represents the standardized score of the monitoring data; x represents the monitoring data; μ represents the mean of the monitoring data; and σ represents the standard deviation of the monitoring data.

[0096] Set a threshold (usually 3 or -3). If the standardized score value of a certain monitoring data is greater than 3 or less than -3, the monitoring data is considered to be an outlier. Traverse all the monitoring data in the database and find all the monitoring data whose standardized score values ​​exceed the threshold as outliers. For each detected outlier, find the monitoring data at its adjacent moments, calculate the average value of the monitoring data at adjacent moments, and replace the outlier with the average value.

[0097] For example, suppose a monitoring data x i is an outlier, and its adjacent monitoring data are x i-1 and x i+1 , the replaced value is:

[0098]

[0099] In the formula, x i ′ represents the monitoring data after replacement; x i Represents any monitoring data; x i Represents monitoring data x i The previous monitoring data of x i Represents monitoring data x i The next monitoring data.

[0100] The missing data filling submodule is used to fill in the missing data in the monitoring data after abnormal processing using the missing data filling algorithm.

[0101] Specifically, using the missing data filling algorithm to fill in the missing data in the monitoring data after the abnormality processing includes:

[0102] S11. According to the completeness of time series data, the monitoring data are divided into complete data sets and missing data sets.

[0103] S12. Use the weighting algorithm to calculate the weights of the monitoring data attributes in the complete data set, and based on the weight analysis results, determine whether there are missing attributes in the complete data set. If not, the process ends. If so, execute step S13.

[0104] S13. According to the nth attribute of the missing data, the grey correlation coefficient between the nth attribute and the nth attribute of any data point in the complete data set is calculated.

[0105] Specifically, the calculation formula of the grey correlation coefficient is:

[0106]

[0107] In the formula, T represents the grey correlation coefficient between the nth attribute of missing data and the nth attribute of complete data;

[0108] j represents the number of complete data;

[0109] n represents the attribute of the monitoring data;

[0110] c indicates missing data;

[0111] d indicates complete data;

[0112] λ represents the adjustment parameter;

[0113] |c in -d jn | indicates missing data c i With complete data d j The absolute value difference of the nth attribute.

[0114] S14. Based on the grey correlation coefficient, the grey correlation value between the missing data and the complete data is calculated.

[0115] S15. Sort the grey correlation values ​​according to a preset standard, and select several complete data with the highest grey correlation values ​​as candidate nearest neighbors for filling the missing data.

[0116] It should be noted that the grey correlation values ​​are sorted in descending order, and the first several complete data with the highest grey correlation values ​​are selected.

[0117] S16, calculating the average value of several candidate nearest neighbors, obtaining the missing attributes of the missing data, and filling the missing attribute values. After the filling is completed, return to step S12 to continue processing new missing data.

[0118] The normalization processing submodule is used to normalize the padded monitoring data using a normalization algorithm.

[0119] It should be noted that the normalization process of the filled monitoring data using the normalization algorithm includes:

[0120] Calculate the maximum value (max) and minimum value (min) of each monitoring data, and for each monitoring data x, calculate its normalized value:

[0121]

[0122] Where x′ represents the normalized monitoring data; x represents the monitoring data.

[0123] The operation mode selection module is used to select the corresponding operation mode according to the user's needs. The operation modes include water quantity restriction mode and water quantity and water quality restriction mode.

[0124] It should be noted that the water quantity limitation mode refers to the mode that only considers the water quantity limitation, and the system only adjusts the valves and water pumps according to the water level data; the water quantity and water quality limitation mode refers to the mode that considers the water quantity and water quality limitations, and considers both water quantity and water quality at the same time, and performs more intelligent control based on the "Comprehensive Water Quality and Pollution Index".

[0125] The data analysis module is used to analyze the preprocessed monitoring data and historical monitoring data using the random forest algorithm to obtain monitoring analysis results.

[0126] Specifically, the data analysis module includes: a data set establishment submodule, a model initialization submodule, a model training and optimization submodule, an index calculation submodule, and an analysis result acquisition submodule;

[0127] The data set establishment and feature extraction submodule is used to obtain historical monitoring data from the database, establish a historical monitoring data set, and extract historical feature parameters from the historical monitoring data set.

[0128] It should be noted that historical characteristic parameters include statistical characteristics, trend characteristics, periodic characteristics, abnormal characteristics, correlation characteristics, etc.

[0129] Among them, statistical characteristics include mean, median, maximum, minimum, standard deviation, and variance; trend characteristics include linear trend and nonlinear trend; periodic characteristics include cycle length and cycle amplitude; abnormal characteristics include outliers and abnormal patterns; and association characteristics include correlation analysis and causal inference.

[0130] The model initialization submodule is used to combine historical feature parameters with the random forest algorithm to initialize the random forest model.

[0131] The model training and optimization submodule is used to train and optimize the random forest model based on the historical monitoring data set using the adaptive learning rate optimization algorithm.

[0132] Specifically, based on the historical monitoring data set, the adaptive learning rate optimization algorithm is used to train and optimize the random forest model, including:

[0133] S21, input the historical monitoring data set into the random forest model for training, and calculate the loss function value;

[0134] S22, back-propagating the loss function to calculate the gradient of the random forest model parameters relative to the loss function;

[0135] S23, using the adaptive learning rate optimization algorithm to calculate the square of the current gradient, and updating the cache variables of the adaptive learning rate optimization algorithm;

[0136] S24. Calculate a new learning rate according to the current state of the adaptive learning rate optimization algorithm, and use the new learning rate to update the parameters of the random forest model;

[0137] S25. Repeat steps S21 to S24 until a preset number of iterations is reached.

[0138] It should be noted that random forest is an ensemble learning method based on decision trees. It builds multiple decision trees and combines the prediction results of these trees to improve the accuracy and stability of the model. The random forest model builds multiple decision trees by performing multiple random sampling (with replacement sampling) on ​​the original data set. Each tree is trained based on a different data subset and feature subset. When building a decision tree, random forest randomly selects feature subsets and sample subsets to avoid overfitting of the model. For classification problems, random forest determines the final prediction result by voting (majority voting); for regression problems, the final prediction is obtained by averaging the prediction results of all decision trees.

[0139] The index calculation submodule is used to input the preprocessed monitoring data into the optimized random forest model, output the water quality prediction results, and calculate the comprehensive water quality index.

[0140] Specifically, the pre-processed monitoring data is input into the optimized random forest model, the water quality prediction results are output, and the comprehensive water quality index is calculated, including:

[0141] S31. Input the preprocessed monitoring data into the optimized random forest model and output the water quality prediction results.

[0142] S32. Based on the prediction results, design the hierarchical structure of the hierarchical analysis method, which includes the target layer, the criterion layer and the solution layer.

[0143] S33. According to the preprocessed monitoring data, the elements in the criterion layer and the solution layer are compared in pairs, and a judgment matrix between each layer is constructed based on the comparison results.

[0144] S34. Calculate the weights of the elements at the solution layer using the eigenvector algorithm, and calculate the comprehensive water quality index based on the weights of the elements at the solution layer and the monitoring data.

[0145] Specifically, the calculation formula for the comprehensive water quality index is:

[0146]

[0147] In the formula, Q represents the comprehensive index of water quality and pollution;

[0148] g represents the number of water quality parameters;

[0149] i represents the index;

[0150] ω i represents the weight of the i-th water quality parameter;

[0151] x i represents the actual measured value of the i-th water quality parameter;

[0152] x i * represents the ideal value of the i-th water quality parameter.

[0153] The analysis result acquisition submodule is used to obtain the monitoring and analysis results based on the monitoring data and the comprehensive water quality index.

[0154] It should be noted that the monitoring data and the comprehensive water quality index at different time points are compared to analyze the changing trend of water quality. For example, the changes in pH value, water quality, water volume and other data, as well as their impact on the comprehensive water quality index, are observed. According to the size of the comprehensive water quality index, the water quality is comprehensively evaluated. The higher the comprehensive water quality index, the better the water quality; otherwise, it means that the water quality is poor.

[0155] The control instruction generation module is used to generate corresponding control instructions according to the selection of the operation mode and the results of monitoring and analysis.

[0156] Specifically, the control instruction generation module includes: a water quality and water quantity assessment submodule and a marking and instruction generation submodule;

[0157] The water quality and quantity assessment submodule is used to assess whether the water quality and quantity meet the standards based on the water quantity in the monitoring data and the calculated comprehensive water quality and pollution index, combined with the preset threshold.

[0158] It should be noted that water volume assessment includes comparing the water volume monitoring data at each time point with the preset water volume threshold to determine whether the water volume is within a reasonable range. For example, the water volume at time points T1, T2, and T4 is 80-120m 3 / h, which meets the standard; while the water volume at T3 and T5 time points is lower than and slightly higher than the threshold range respectively, and the reasons need further analysis.

[0159] Water quality assessment includes comparing the water quality index at each time point with the preset water quality index threshold to determine whether the water quality is qualified. For example, the water quality index at time points T1, T2, and T4 is higher than 60, and the water quality is qualified; while the WQI at time points T3 and T5 is lower than 60, and the water quality is unqualified.

[0160] Comprehensive assessment includes combining the water quantity and water quality assessment results to make a comprehensive judgment. If both the water quantity and water quality meet the standards, it is judged that both the water quality and quantity meet the standards; if either the water quantity or the water quality does not meet the standards, it is judged that neither the water quality nor the water quantity meets the standards.

[0161] The marking and instruction generation submodule is used to classify and mark the working status of the interception well according to the evaluation results of water quantity and water quality, and generate corresponding control instructions in combination with the selected operation mode.

[0162] It should be noted that it is assumed that the following water quality and quantity monitoring data are available and have been evaluated, as shown in Table 1;

[0163] Table 1 Water quality and water quantity monitoring data evaluation table

[0164]

[0165] According to the evaluation results, the working conditions of interception wells are divided into the following categories and given corresponding marks:

[0166] Both water quantity and quality are normal: The water quantity and quality meet the preset standards and are marked as "normal".

[0167] Water volume exceeds the preset threshold, but the water quality meets the standard, which is marked as "Water volume exceeds the standard".

[0168] Low Water: When the water level is below the preset threshold, it is marked as "Low Water" regardless of the water quality.

[0169] Unacceptable water quality: When the water quality is below the preset threshold, it is marked as “poor water quality” regardless of the water quantity.

[0170] In addition, the preset thresholds include water volume thresholds and water quality and pollution comprehensive index thresholds.

[0171] The water volume threshold includes a minimum threshold and a maximum threshold. The minimum threshold is set according to the minimum treatment capacity of the sewage treatment plant or the water source stability requirements. For example, it can be set to 70m 3 / h; The maximum threshold is set according to the maximum treatment capacity of the sewage treatment plant or the requirement to avoid overload operation. For example, it can be set to 120m 3 / h.

[0172] The threshold of the comprehensive water quality and pollution index is the qualified threshold. The qualified threshold of the comprehensive water quality and pollution index is set according to the water quality evaluation standard or relevant environmental protection laws and regulations. For example, it can be set to 60. When the comprehensive water quality and pollution index is higher than or equal to this value, the water quality is considered qualified; when it is lower than this value, the water quality is considered unqualified.

[0173] Water limit mode:

[0174] It is suitable for situations where the water volume exceeds the standard or is too low, and the water volume can be limited by adjusting the water inlet flow rate of the interception well.

[0175] Control command example:

[0176] At time point T2 (water volume exceeds the limit), the control command is to start the water volume restriction mode and adjust the water flow rate to the preset threshold of 100m 3 / h or less". For time points T3 and T5 (low water volume), the control instruction is "start water volume restriction mode (but here to maintain the minimum flow rate) to ensure that the water flow rate is not lower than the minimum threshold of 70m 3 / h (or set according to specific circumstances)".

[0177] Water quantity and quality restriction mode:

[0178] Applicable to situations where water quality is unacceptable, taking into account both water volume adjustment and water quality improvement measures.

[0179] Control command example:

[0180] For the T4 time point (normal water volume, unqualified water quality), the control instruction is "start the water volume and water quality restriction mode, strengthen the sewage treatment process, and adjust the water flow rate according to actual needs to maintain the treatment efficiency." For the T5 time point (although it has been marked as low water volume and unqualified water quality, they still need to be considered in this mode), the control instruction is "on the basis of maintaining the minimum water flow rate (such as not less than 70m 3 / h), initiate water quality improvement measures, such as increasing the dosage of chemicals, adjusting treatment process parameters, etc.

[0181] The control unit 4 is used to formulate and execute corresponding control strategies according to the control instructions generated by the cloud processing unit.

[0182] It should be noted that a control strategy is developed based on the evaluation results and the requirements of the control instructions. The control strategy includes adjusting the opening of the water inlet valve, changing the speed of the water pump, starting or stopping the dosing device, etc.

[0183] The remote server 5 is used to monitor the operation status of the interception well in real time through a user interface and adjust the control strategy according to the operation status.

[0184] It should be noted that the real-time monitoring data of the interception well, including key information such as water quantity, water quality, and water level, is displayed through the user interface of the remote server 5; a graphical interface is provided to intuitively display the operating status of the interception well, such as the valve opening and closing status, the water pump operating status, etc. The remote server 5 continuously monitors the operating status of the interception well, including changes in water quantity, fluctuations in water quality, and water level rise and fall, and by setting a threshold or alarm mechanism, when the operating status exceeds the normal range, an alarm is promptly issued to notify the user.

[0185] According to the control strategy set by the user, the remote server 5 can automatically or manually adjust the control parameters of the interception well. The user can input a new control strategy through the user interface. After the remote server receives and verifies it, it sends the adjusted control strategy to the cloud processing unit 3.

[0186] In summary, by means of the above technical solution of the present invention, through the sensing and collecting unit 1, various monitoring data inside the interception well can be obtained in real time and accurately, providing an accurate data decision basis for the cloud processing unit 3, which enables the cloud processing unit 3 to intelligently select the operation mode according to user needs and environmental changes, thereby ensuring that the interception well can still maintain efficient and stable operation under complex and changeable working conditions. Through the cloud processing unit 3, not only the monitoring data from the sensing and collecting unit 1 is received and processed, but also corresponding control instructions are generated according to the processing results. These control instructions are transmitted to the control unit 4 in real time through the communication unit 2, and the control unit 4 formulates and executes corresponding control strategies according to the instructions. This close linkage between the cloud and the site realizes real-time adjustment and precise control of the operation state of the interception well, and effectively improves the automation level and operation efficiency of the system. Through the remote server 5, the user can monitor the operation state of the interception well in real time through the user interface at any time and anywhere, and adjust the control strategy according to the operation state, which not only improves the flexibility and convenience of the system, but also reduces the operation and maintenance costs and improves the operation and maintenance efficiency.

[0187] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A cloud model control method for intelligent interception wells, characterized in that: The control method uses an intelligent interception well cloud model control system to achieve remote monitoring and intelligent control of the operation status of the interception well; The system includes: a sensing and collecting unit, a communication unit, a cloud processing unit, a control unit and a remote server; The sensing and collecting unit is used to collect monitoring data of the interception well in real time using sensors inside the interception well; The communication unit is used to transmit the collected monitoring data to the cloud processing unit through the Internet of Things network; The cloud processing unit is used to receive monitoring data, select a corresponding operation mode according to user needs, process the monitoring data, and generate corresponding control instructions; The control unit is used to formulate and execute corresponding control strategies according to the control instructions generated by the cloud processing unit; The remote server is used to monitor the operation status of the interception well in real time through a user interface and adjust the control strategy according to the operation status.

2. According to claim 1, a cloud-based model control method for intelligent interception wells is characterized in that: The cloud processing unit includes: a data storage and management module, a data preprocessing module, an operation mode selection module, a data analysis module and a control instruction generation module; The data storage and management module is used to receive monitoring data and store the monitoring data in a database for user query and management; The data preprocessing module is used to preprocess the monitoring data in the database; The operation mode selection module is used to select a corresponding operation mode according to user needs, and the operation modes include a water quantity restriction mode and a water quantity and water quality restriction mode; The data analysis module is used to analyze the pre-processed monitoring data and historical monitoring data using a random forest algorithm to obtain monitoring analysis results; The control instruction generation module is used to generate corresponding control instructions according to the selection of the operation mode and the results of the monitoring and analysis.

3. The intelligent interception well cloud model control method according to claim 2 is characterized in that: The data preprocessing module includes: an outlier processing submodule, a missing data filling submodule and a normalization processing submodule; The outlier processing submodule is used to detect outliers in the monitoring data using an outlier detection algorithm and replace them with the average value of the monitoring data at adjacent moments of the outliers; The missing data filling submodule is used to fill in the missing data in the monitoring data after the abnormality processing by using the missing data filling algorithm; The normalization processing submodule is used to perform normalization processing on the padded monitoring data using a normalization algorithm.

4. The intelligent interception well cloud model control method according to claim 3 is characterized in that: The method of using a missing data filling algorithm to fill in the missing data in the monitoring data after the abnormality processing includes: S11. According to the completeness of the time series data, the monitoring data are divided into complete data sets and missing data sets; S12, using a weighting algorithm to calculate the weights of the monitoring data attributes in the complete data set, and based on the weight analysis results, determine whether there are missing attributes in the complete data set. If not, the process ends; if so, execute step S13; S13, according to the nth attribute of the missing data, calculate the grey correlation coefficient between the nth attribute and the nth attribute of any data point in the complete data set; S14. Calculate the grey correlation value between the missing data and the complete data based on the grey correlation coefficient; S15, sorting the grey correlation values ​​according to a preset standard, and selecting a number of complete data with the highest grey correlation values ​​as candidate nearest neighbors for filling the missing data; S16, calculating the average value of several candidate nearest neighbors, obtaining the missing attributes of the missing data, and filling the missing attribute values. After the filling is completed, return to step S12 to continue processing new missing data.

5. The intelligent interception well cloud model control method according to claim 4 is characterized in that: The calculation formula of the grey correlation coefficient is: In the formula, T represents the grey correlation coefficient between the nth attribute of missing data and the nth attribute of complete data; j represents the number of complete data; n represents the attribute of the monitoring data; c indicates missing data; d indicates complete data; λ represents the adjustment parameter; |c in -d jn | indicates missing data c i With complete data d j The absolute value difference of the nth attribute.

6. The intelligent interception well cloud model control method according to claim 2 is characterized in that: The data analysis module includes: a data set establishment submodule, a model initialization submodule, a model training and optimization submodule, an index calculation submodule and an analysis result acquisition submodule; The data set establishment and feature extraction submodule is used to obtain historical monitoring data from the database, establish a historical monitoring data set, and extract historical feature parameters from the historical monitoring data set; The model initialization submodule is used to initialize the random forest model by combining historical feature parameters with the random forest algorithm; The model training and optimization submodule is used to train and optimize the random forest model based on the historical monitoring data set using an adaptive learning rate optimization algorithm; The index calculation submodule is used to input the pre-processed monitoring data into the optimized random forest model, output the water quality prediction results, and calculate the comprehensive water quality index; The analysis result acquisition submodule is used to obtain monitoring analysis results based on monitoring data and the comprehensive water quality index.

7. The intelligent interception well cloud model control method according to claim 6 is characterized in that: The training and optimization of the random forest model using an adaptive learning rate optimization algorithm based on the historical monitoring data set includes: S21, input the historical monitoring data set into the random forest model for training, and calculate the loss function value; S22, back-propagating the loss function to calculate the gradient of the random forest model parameters relative to the loss function; S23, using the adaptive learning rate optimization algorithm to calculate the square of the current gradient, and updating the cache variables of the adaptive learning rate optimization algorithm; S24. Calculate a new learning rate according to the current state of the adaptive learning rate optimization algorithm, and use the new learning rate to update the parameters of the random forest model; S25. Repeat steps S21 to S24 until a preset number of iterations is reached.

8. The intelligent interception well cloud model control method according to claim 7 is characterized in that: The pre-processed monitoring data is input into the optimized random forest model, the water quality prediction result is output, and the comprehensive water quality index is calculated, including: S31, input the pre-processed monitoring data into the optimized random forest model, and output the water quality prediction result; S32. Based on the prediction results, design a hierarchical structure of the analytic hierarchy process, wherein the hierarchical structure includes a target layer, a criterion layer, and a solution layer; S33, according to the preprocessed monitoring data, the elements in the criterion layer and the scheme layer are compared in pairs, and a judgment matrix between the layers is constructed based on the comparison results; S34. Calculate the weights of the elements at the solution layer using the eigenvector algorithm, and calculate the comprehensive water quality index based on the weights of the elements at the solution layer and the monitoring data.

9. The intelligent interception well cloud model control method according to claim 8, characterized in that: The calculation formula of the water quality comprehensive index is: In the formula, Q represents the comprehensive index of water quality and pollution; g represents the number of water quality parameters; i represents the index; ω i represents the weight of the i-th water quality parameter; x i represents the actual measured value of the i-th water quality parameter; x i * represents the ideal value of the i-th water quality parameter.

10. The intelligent interception well cloud model control method according to claim 9, characterized in that: The control instruction generation module includes: a water quality and water quantity assessment submodule and a marking and instruction generation submodule; The water quality and quantity assessment submodule is used to assess whether the water quality and quantity meet the standards based on the water quantity in the monitoring data and the calculated water quality and pollution comprehensive index, combined with the preset threshold value; The marking and instruction generation submodule is used to classify and mark the working status of the interception well according to the evaluation results of water quantity and water quality, and generate corresponding control instructions in combination with the selected operation mode.

Images