Shield slurry flocculation filter pressing system based on machine learning, optimization method and computer readable storage medium

Abstract

The invention relates to the technical field of geotechnical engineering, in particular to a shield slurry flocculation filter pressing system based on machine learning, an optimization method and a computer readable storage medium, and the optimization method comprises the following steps: obtaining and preprocessing shield slurry data; selecting characteristic data of shield slurry; carrying out standardization and normalization processing; the feature data is divided into a training set and a test set; an improved neural network model is adopted for training; obtaining an optimal neural network model parameter by adopting a particle swarm optimization algorithm; deploying the trained neural network model into a shield mud flocculation filter pressing system; and comparing the shield mud prediction treatment effect obtained by the trained neural network model with the actual treatment effect of the shield mud to obtain the optimal parameters of the shield mud flocculation filter pressing system, and implementing the optimal parameters. And periodically detecting the performance of the neural network model. According to the shield slurry flocculation filter pressing system, the filter pressing speed is reasonably controlled through a machine learning method, the best filter pressing effect is achieved through intelligent adjustment, solidification of shield slurry can be completed in a high-quality mode, and mud cakes obtained after filter pressing can be recycled.

Description

Technical Field

[0001] The present invention relates to the field of geotechnical engineering technology, and in particular to a shield mud flocculation and filtration system based on machine learning, an optimization method and a computer-readable storage medium. Background Art

[0002] With the continuous improvement of urbanization in my country, the mileage of urban subways has also increased accordingly. At present, a large number of urban subways are constructed by shield method. The reasonable discharge of construction waste mud and slag generated by shield method is very important. Among them, shield mud is the waste product formed by the mud produced by slurry balance shield and the slag produced by earth pressure balance shield after passing through the mud-water separation system and sand washing system. The chemical composition of shield mud is complex, not easy to settle, and improper treatment is prone to environmental pollution. There are a series of problems in the current shield mud treatment methods, including low treatment efficiency, difficult operation and maintenance, and long treatment cycle of shield mud disposal equipment. With the development of computer technology, machine learning methods provide new ideas for the intelligent control of shield mud.

[0003] Traditional shield tunneling mud is usually dehydrated by filter press and converted into mud cake, and then transferred and placed together with shield slag by transport vehicles. The transportation cost is high, the resource utilization rate is low, and a large amount of land resources are occupied. Some projects even have the phenomenon of illegal discharge and random burial, which seriously affects the urban environment, economic development and people's life safety. Although my country's treatment methods for engineering waste mud have developed from the most primitive direct drying method to more advanced mechanical treatment methods, flocculation sedimentation methods and chemical solidification methods, the actual resource utilization rate of shield mud in my country is less than 1%, and shield mud cannot be efficiently treated and reused. Therefore, shield tunneling mud is in urgent need of efficient treatment.

[0004] For example, the patent application document with Chinese invention patent application number CN201811337103.1 discloses a shield mud treatment system, including a shield mud extraction pump, a fully automatic flocculant liquid preparation and addition system, a mud reaction tank, a concentration tank, a clear water tank, and a secondary dewatering screen. The fully automatic flocculant liquid preparation and addition system can automatically add flocculants to complete the automation of shield mud treatment. This device cannot complete the intelligent control of the filter pressing speed, and cannot effectively improve the quality of shield mud filter pressing. In the prior art, there is a lack of a method to complete the automation of the shield mud filter pressing process to ensure the efficiency of shield mud treatment, while preventing the shield mud from being unable to be treated in time and causing environmental pollution. Summary of the invention

[0005] The purpose of the present invention is to solve the problems of low processing efficiency and low utilization rate of shield mud in the current shield mud flocculation and filtration system as described in the above background technology. A shield mud flocculation and filtration system based on machine learning, an optimization method and a computer-readable storage medium are proposed. The shield mud flocculation and filtration system uses a machine learning method to reasonably control the filtration speed and intelligently adjust to achieve the best filtration effect. Not only can the solidification of the shield mud be completed with high quality, but the mud cake after filtration can be reused for on-site roadbed, greening and backfill soil, etc.

[0006] The technical solution adopted by the present invention to solve the technical problem is: an optimization method for a shield mud flocculation filter press system, the optimization method comprising the following steps:

[0007] S1. Obtain shield mud data and perform preprocessing;

[0008] S2, select characteristic data of shield mud;

[0009] S3, standardizing and normalizing the feature data;

[0010] S4, dividing the processed feature data into a training set and a test set;

[0011] S5, using an improved neural network model for training;

[0012] S6. Using a particle swarm optimization algorithm to obtain the best neural network model parameters, and obtain a trained neural network model;

[0013] S7, deploying the trained neural network model to the shield mud flocculation and filtration system;

[0014] S8, comparing the predicted treatment effect of the shield mud obtained by the trained neural network model with the actual treatment effect of the shield mud, obtaining the optimal parameters of the shield mud flocculation filter press system and implementing them;

[0015] S9. Regularly test the performance of the neural network model and compare the predicted treatment effect of the shield mud obtained by the neural network model with the actual treatment effect of the shield mud.

[0016] Furthermore, in step S1, the preprocessing includes filling in missing data in the acquired shield mud data and deleting abnormal data in the acquired shield mud data.

[0017] Furthermore, the shield mud data includes the temperature and concentration of the shield mud, the dosage of the flocculant, the flocculation time, the stirring speed and the filter pressing speed.

[0018] Furthermore, in step S4, the processed feature data is divided into a training set and a test set in a ratio of 4:1.

[0019] Furthermore, in step S5, the neural network model includes multiple convolutional layers, multiple pooling layers, multiple activation functions and a fully connected layer.

[0020] The convolution layer is used to increase the receptive field of the neural network model and improve the feature extraction capability of shield mud data;

[0021] The pooling layer is used to reduce data dimension, reduce computational complexity, and improve the operating efficiency of the neural network model;

[0022] The fully connected layer is used to obtain the prediction results of the shield mud treatment effect;

[0023] The activation function is used to enhance the nonlinear fitting ability of the neural network model.

[0024] Furthermore, the neural network model includes a first convolutional layer, a first activation function, a first pooling layer, a second convolutional layer, a second activation function, a second pooling layer, a third convolutional layer, a third activation function, a third pooling layer and a fully connected layer connected in sequence, and the first activation function, the second activation function and the third activation function are ReLU activation functions.

[0025] Furthermore, in step S5, the training using the improved neural network model is specifically as follows:

[0026] S51, training the neural network model through historical data of shield mud and shield mud flocculation and filter pressing system operation parameters, and adjusting the neural network model parameters using a back propagation algorithm and an optimizer;

[0027] S52. Use training sets and validation sets for supervised learning to improve the generalization ability of the neural network model.

[0028] Furthermore, the treatment effect includes the moisture content, density, cohesion and tensile strength of the mud cake after the treatment.

[0029] A computer-readable storage medium stores an optimization program for a shield mud flocculation and filtration system. When the optimization program is executed by a processor, the steps of the optimization method described above are implemented.

[0030] A shield mud flocculation and filtration system based on machine learning, the shield mud flocculation and filtration system comprising a shield mud extraction pump, a primary flocculation system, an automatic flocculant dosing system, a clear water tank, a slurry storage tank, a slurry pump, a secondary dewatering screen, a plate and frame filter press, a mud cake collection box and a flocculation and filtration control system, the slurry outlet of the shield mud extraction pump is connected with the primary flocculation system through a pipeline, the drug outlet of the automatic flocculant dosing system is connected with the primary flocculation system through a pipeline, the upper water outlet of the primary flocculation system is connected with the clear water tank through a pipeline, the lower slurry outlet of the primary flocculation system is connected with the slurry storage tank through a pipeline, the slurry storage tank is connected with the slurry pump through a pipeline, the slurry pump is connected with the secondary dewatering screen through a pipeline, the secondary dewatering screen is connected with the plate and frame filter press, and the flocculation and filtration control system is electrically connected with the automatic flocculant dosing system and the plate and frame filter press;

[0031] The flocculation and filtration control system includes a storage module, a processing module and a monitoring module. The storage module includes the computer-readable storage medium as described above. The processing module is used to execute the program stored in the computer-readable storage medium. The monitoring module is used to obtain the operating parameters of the shield mud flocculation and filtration system and the shield mud data.

[0032] Furthermore, the automatic flocculant dosing system includes a PLC control system, a flocculant silo, a control valve, a cylindrical heater, a screw feeder, a feeding pump and a flow meter. The PLC control system is electrically connected to the flocculant silo, the control valve, the cylindrical heater, the screw feeder, the feeding pump and the flow meter. The screw feeder is placed in the cylindrical heater, the feed port of the feeding pump is connected to the flocculant silo, the discharge port of the feeding pump is connected to the cylindrical heater through a pipeline, and the control valve and the flow meter are arranged on the pipeline.

[0033] In the above technical solution, the PLC control system uses an advanced programmable logic controller to be responsible for the control and coordination of the entire system. The PLC control system communicates with other components through electrical connections, receives and processes sensor data, and performs real-time control based on the output of the machine learning module.

[0034] Furthermore, the secondary dewatering screen includes a first sieve plate, a second sieve plate, a third sieve plate and three vibration motors. The aperture of the first sieve plate is 8 mm, the aperture of the second sieve plate is 5 mm, and the aperture of the third sieve plate is 3 mm. The first sieve plate, the second sieve plate and the third sieve plate are respectively connected to a vibration motor.

[0035] In the above technical solution, the gradual reduction of the pore size is conducive to the gradual filtration and dehydration, and the dehydration effect of the shield mud is improved. Each layer of the screen plate is equipped with a vibration motor to generate vibration force, so that the shield mud forms a thin layer on the screen plate, which is conducive to water separation and flocculent dehydration. Each vibration motor can adjust the vibration angle through an adjustable device to adapt to different shield mud characteristics and processing requirements.

[0036] The shield mud in the slurry storage tank enters the secondary dewatering screen through the slurry pump. The shield mud first passes through the first layer of sieve plates, and the large particle floccules are filtered out and dehydrated after the first layer; the shield mud that has not been dehydrated continues to pass through the second and third layers of sieve plates, and the gradually decreasing apertures help further dehydration; the vibration motors of each layer of sieve plates generate vibration force to evenly distribute the shield mud on the sieve plates, promote water separation and accelerate the sedimentation of floccules. Through sieve plates with different apertures, the shield mud gradually loses water when passing through each layer, and finally obtains relatively dry floccules. The angle of the vibration motor can be changed through the adjustable device to adjust the dehydration effect according to the properties of the floccules and the flow of the shield mud.

[0037] Furthermore, the agent used in the automatic flocculant dosing system can be selected from powder or liquid.

[0038] Furthermore, the primary flocculation system includes a flocculation sedimentation tank, an agitator and a reaction tank, the agitator is arranged in the reaction tank, the reaction tank is connected to the flocculation sedimentation tank, and a lower layer slurry outlet is provided at the bottom of the flocculation sedimentation tank.

[0039] The beneficial effects of the present invention are:

[0040] 1. The present invention uses a machine learning method to optimize the operating parameters of the shield mud flocculation and filter press system according to various parameters of the shield mud, thereby improving the treatment efficiency and treatment quality of the shield mud.

[0041] 2. The shield mud flocculation and filtration system of the present invention can obtain system operation parameters in real time through the monitoring module, timely discover potential safety hazards in the construction process, and reduce accident risks.

[0042] 3. The optimization method of the shield mud flocculation and filtration system of the present invention analyzes a large amount of shield mud treatment data through a machine learning algorithm, optimizes the waste liquid treatment plan, and improves the system performance and adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of the optimization method of the shield mud flocculation filter press system of the present invention;

[0044] Figure 2 This is a schematic diagram of the structure of the improved convolutional neural network model of the present invention;

[0045] Figure 3 It is a schematic structural diagram of the shield mud flocculation and filtration system of the present invention. DETAILED DESCRIPTION

[0046] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0047] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0048] The present invention is further described in conjunction with specific examples. The following examples are only for explaining the present invention, but do not constitute a limitation of the present invention. The test samples and test processes used in the following examples include the following (if the specific experimental conditions are not indicated in the examples, they are usually in accordance with conventional conditions or the conditions recommended by the reagent company; the reagents, consumables, etc. used in the following examples, if not otherwise specified, can all be obtained from commercial channels).

[0049] One of the purposes of the present invention is to provide an optimization method for a shield mud flocculation filter press system. The flow chart of the optimization method is as follows: Figure 1 As shown, the specific steps are as follows:

[0050] S1. Obtain shield mud data and perform preprocessing;

[0051] S2, select characteristic data of shield mud;

[0052] S3, standardizing and normalizing the feature data;

[0053] S4, dividing the processed feature data into a training set and a test set;

[0054] S5, using an improved neural network model for training;

[0055] S6. Using a particle swarm optimization algorithm to obtain the best neural network model parameters, and obtain a trained neural network model;

[0056] S7, deploying the trained neural network model to the shield mud flocculation and filtration system;

[0057] S8, comparing the predicted treatment effect of the shield mud obtained by the trained neural network model with the actual treatment effect of the shield mud, obtaining the optimal parameters of the shield mud flocculation filter press system and implementing them;

[0058] S9. Regularly test the performance of the neural network model and compare the predicted treatment effect of the shield mud obtained by the neural network model with the actual treatment effect of the shield mud.

[0059] Among them, in step S1, the preprocessing is to fill the missing data in the acquired shield mud data and delete the abnormal data in the acquired shield mud data.

[0060] In this embodiment, the shield mud data are the temperature, concentration, flocculant dosage, stirring speed of the agitator, filtration speed of the plate and frame filter press and flocculation time of the shield mud; the treatment effect is the moisture content, density, cohesion and tensile strength of the mud cake after the treatment.

[0061] In steps S1-S5, shield mud data is collected in real time through sensors, and the shield mud data is preprocessed, including processing missing values ​​and outliers; feature engineering is performed on the processed data to extract meaningful features to ensure that the obtained feature data meets the requirements of neural network model training; the feature data is divided into a training set and a test set in a ratio of 4:1 for neural network model training, wherein the neural network model adopts a sequentially connected input layer, a first convolutional layer, a first activation function, a first pooling layer, a second convolutional layer, a second activation function, a second pooling layer, a third convolutional layer, a third activation function, a third pooling layer and a fully connected layer structure, the first activation function, the second activation function and the third activation function are ReLU (Rectified Linear Unit) activation functions, and its structure diagram is as shown in FIG. Figure 2 As shown in the figure, the convolution layer uses an improved convolution kernel structure to increase the receptive field of the model and improve the feature extraction capability of shield mud data; the pooling layer uses maximum pooling or average pooling to reduce the data dimension, reduce the computational complexity, and improve the operation efficiency of the model; the fully connected layer combines and transforms the extracted features to obtain the final flocculation effect prediction result; the activation function is used to enhance the nonlinear fitting capability of the model.

[0062] The specific steps of this training method are as follows:

[0063] S51. The neural network model is trained through the historical data of shield mud and shield mud flocculation and filtration system operation parameters, and the back propagation algorithm and optimizer are used to adjust the parameters of the neural network model.

[0064] S52. Use training sets and validation sets for supervised learning to improve the generalization ability of the neural network model.

[0065] In this embodiment, an improved neural network model is used to learn the complex patterns and features in the shield mud flocculation and filtration process, using multiple improved convolutional layers, pooling layers, and fully connected layers to better capture the spatial relationships and features in the data.

[0066] In step S6, the PSO particle swarm optimization algorithm is used to optimize the neural network model and adjust the hyperparameters of the model, such as learning rate, convolution kernel size, etc., to optimize the performance of the model. The model can be evaluated and improved by cross-validation and other methods.

[0067] In steps S7 and S8, the trained neural network model is used to compare the model prediction results with the actual results, and the optimal parameter combination is selected according to the actual results. After the selection is completed, the optimal filter pressing speed, flocculant dosage, and agitator stirring speed are obtained to achieve the optimal shield mud treatment result.

[0068] The second object of the present invention is to provide a computer-readable storage medium, in which an optimization program for the operating parameters of a shield mud flocculation and filtration system can be stored. The computer-readable storage medium can optionally be a storage device such as a memory card or a hard disk, and the optimization program can be executed by a processor.

[0069] The third object of the present invention is to provide a shield mud flocculation and filtration system based on machine learning, the shield mud flocculation and filtration system is as follows: Figure 3 As shown, the shield mud flocculation and filtration system consists of a shield mud extraction pump, a primary flocculation system, an automatic flocculant dosing system, a clear water tank, a slurry storage tank, a slurry pump, a secondary dewatering screen, a plate and frame filter press, a mud cake collection box, a flocculation and filtration control system and a number of sensors. The slurry outlet of the shield mud extraction pump is connected with the primary flocculation system through a pipeline, the drug outlet of the automatic flocculant dosing system is connected with the primary flocculation system through a pipeline, the upper water outlet of the primary flocculation system is connected with the clear water tank through a pipeline, the lower slurry outlet of the primary flocculation system is connected with the slurry storage tank through a pipeline, the slurry storage tank is connected with the slurry pump through a pipeline, the slurry pump is connected with the secondary dewatering screen through a pipeline, the secondary dewatering screen is connected with the plate and frame filter press, and the flocculation and filtration control system is electrically connected with the automatic flocculant dosing system and the plate and frame filter press.

[0070] In this embodiment, sensors with different functions, such as thermometers and concentration meters, are installed in various components of the shield mud flocculation and filtration system to obtain data such as the temperature, concentration, moisture content, density, cohesion and tensile strength of the shield mud. The sensor is electrically connected to the flocculation and filtration control system to transmit the acquired data to the flocculation and filtration control system for real-time display and monitoring.

[0071] The flocculation and filtration control system consists of a CPU processor, a mainboard, a memory, a solid-state drive, a speed sensor, a Wi-Fi communication module, a power supply, a chassis shell and a display. The CPU processor, the memory, the solid-state drive, the Wi-Fi communication module and the power supply are all installed in the chassis shell, and the display is electrically connected to the mainboard. The solid-state drive is used as a computer-readable storage medium to store the program of the optimization method of the shield mud flocculation and filtration system and the Windows operating system. The CPU processor is used as a processing module to run the above program. The system operating parameters and shield mud data obtained by the sensor are transmitted through the Wi-Fi communication module, processed by the CPU processor and displayed on the display.

[0072] In this embodiment, the CPU processor adopts Intel Core i9-12900K, the memory adopts 64GB, the solid-state drive adopts Samsung 970EVO Plus, the speed sensor adopts LMS1xx series sensor, the Wi-Fi communication module adopts ESP8266 series, the power supply adopts 18650 lithium battery, the chassis shell adopts aluminum shell, and the Windows operating system adopts Windows 10 operating system.

[0073] The flocculant automatic dosing system consists of a PLC control system, a flocculant silo, a control valve, a cylindrical heater, a screw feeder, a feeding pump and a flow meter. The PLC control system is electrically connected to the flocculant silo, the control valve, the cylindrical heater, the screw feeder, the feeding pump and the flow meter. The screw feeder is placed in the cylindrical heater, the feed port of the feeding pump is connected to the flocculant silo, the discharge port of the feeding pump is connected to the cylindrical heater through a pipeline, and the control valve and the flow meter are arranged on the pipeline.

[0074] Among them, the PLC control system uses an advanced programmable logic controller, which is responsible for the control and coordination of the entire system; the PLC control system communicates with other components through electrical connections, receives and processes sensor data, and performs real-time control according to the output of the neural network model (i.e., the machine learning model); the flocculant silo is used to store flocculants to ensure that the system can supply flocculants in time to optimize the flocculation effect; the control valve is installed on the pipeline, and the PLC control system controls the switch according to the real-time monitoring data to adjust the amount of flocculant added; the cylindrical heater is located in the system and is connected to the PLC control system through an electrical connection to heat the shield mud to improve the flocculation effect and fluid performance; the screw feeder is placed in the cylindrical heater and is connected to the PLC control system through an electrical connection to add flocculants to the shield mud to ensure uniform mixing of the flocculants; the feeding pump is connected to the PLC control system through an electrical connection and to the cylindrical heater, and is responsible for transporting the mixture of shield mud and flocculant to the subsequent processing stage; the flow meter is installed in the pipeline and is connected to the PLC control system through an electrical connection to monitor the flow of shield mud and flocculant in real time, and provide it to the machine learning model for learning and control.

[0075] In this embodiment, a thermometer is also installed, which is connected to the PLC control system through an electrical connection and is installed in the area where the shield mud and flocculant are mixed to monitor temperature changes to support the optimization of the heating and flocculation effects of the shield mud.

[0076] The secondary dewatering screen consists of a first sieve plate, a second sieve plate, a third sieve plate and three vibration motors. The aperture of the first sieve plate is 8mm, the aperture of the second sieve plate is 5mm, and the aperture of the third sieve plate is 3mm. The first sieve plate, the second sieve plate and the third sieve plate are respectively connected to a vibration motor.

[0077] The gradual reduction of pore size helps to filter and dehydrate the shield slurry step by step, improving the dehydration effect of the shield slurry. Each layer of the screen is equipped with a vibration motor to generate vibration force, which causes the shield slurry to form a thin layer on the screen, which is conducive to water separation and flocculant dehydration. Each vibration motor can adjust the vibration angle through an adjustable device to adapt to different shield slurry characteristics and processing requirements.

[0078] The primary flocculation system comprises a flocculation sedimentation tank, an agitator and a reaction tank. The agitator is arranged in the reaction tank. The reaction tank is connected to the flocculation sedimentation tank. A lower layer slurry outlet is arranged at the bottom of the flocculation sedimentation tank.

[0079] In this embodiment, the flocculation sedimentation box has a capacity of 1000 liters, is made of stainless steel 304, has a rectangular box structure, and is designed with a slope at the bottom to facilitate the discharge of sediment. The maximum operating temperature is 80°C. The agitator is a suspended agitator with a power of 1.5kW, adjustable, and a maximum of 500 rpm. It is made of corrosion-resistant alloy material. The reaction tank has a capacity of 800 liters, is made of polypropylene, and has a cylindrical tank body, which is convenient for mixing and reaction, and is corrosion-resistant and high-temperature resistant. The lower slurry outlet can be manually controlled and adjusted. It is made of corrosion-resistant stainless steel, and the discharge speed is adjustable to ensure that the sediment is fully settled.

[0080] The operation process of the primary flocculation system of the present invention is as follows:

[0081] 1. The staff starts the flocculation system to ensure the normal operation of all equipment;

[0082] 2. Set the key parameters of the primary flocculation system on the human-machine interface, such as flocculant concentration, agitator speed, etc.;

[0083] 3. Start the agitator and flocculant automatic dosing system to ensure that the flocculant is evenly added to the shield mud;

[0084] 4. The shield mud and flocculant undergo flocculation reaction in the reaction tank, and the suspended matter gradually forms floccules;

[0085] 5. The flocculated shield mud enters the flocculation sedimentation tank, and the floccules settle to the bottom;

[0086] 6. Control the discharge rate of sediment through the lower slurry outlet to ensure that the flocs can be effectively precipitated;

[0087] 7. The flocculation filter press control system monitors the flocculation effect and system operation status in real time through the data transmitted by the sensor, and adjusts the system parameters according to the real-time data through the stored program and the neural network model.

[0088] In this embodiment, the flocculation and filtration control system may be provided with corresponding main console software to control and operate the shield mud flocculation and filtration system. The operation flow is as follows:

[0089] Step 1: The staff starts the power supply, completes the initialization, opens the main console software, and ensures that all necessary hardware devices are connected properly and the main console is in an operational state.

[0090] Step 2: In the console software interface, input the relevant parameters of the current construction geological conditions (including soil density, tensile strength, cohesion, internal friction angle), including but not limited to stratum type, hydrological conditions, groundwater level, etc. The staff needs to confirm the accuracy of the geological conditions so that the optimization program can perform personalized optimization of shield mud treatment according to the actual situation.

[0091] Step 3: The staff sets the treatment plan according to the shield mud treatment requirements, and sets the shield mud flocculation and filtration treatment plan in the system interface, including key parameters such as flocculant concentration, filtration time, and wastewater discharge standards.

[0092] Adjust the treatment plan according to project requirements and environmental characteristics to achieve the best treatment effect.

[0093] Step 4: The staff starts the system, the shield mud extraction pump and the flocculant automatic dosing system start working, and the real-time monitoring function of the system is started to ensure that the system can continuously monitor the key parameters of the shield mud treatment process, such as slurry concentration, filter pressing effect, etc. If any abnormal situation occurs during the operation, the staff needs to promptly discover and record any abnormal situation in the monitoring interface.

[0094] Step 5: The flocculation filter press control system starts working. The flocculation filter press control system uses the optimization program to conduct in-depth analysis of the real-time monitoring data. It studies the treatment effects under different geological conditions, gradually optimizes the filter press speed and other parameters, and improves the system's adaptability and performance.

[0095] Step 6: Abnormal handling. If the system detects abnormal conditions during the shield mud treatment process, the system will issue an alarm. In the abnormal handling interface provided by the system, the operator can view detailed information (including the location of the abnormal situation and the cause of the abnormal accident) and take appropriate measures to stabilize the shield mud treatment process.

[0096] Step 7: After a single task is completed, the staff checks the system performance and adjusts the processing plan based on historical data and machine learning results to optimize the system performance. A manual optimization option is provided in the system interface to meet the needs of special situations.

[0097] Step 8: The system regularly generates detailed reports on shield mud treatment, including treatment effects, resource utilization, wastewater discharge, etc. Record historical data and establish a database to provide a reliable reference for future decision-making and improvement.

[0098] Step 9: After the equipment completes all shield mud processing tasks, regular maintenance of the system is required, including cleaning each device, checking the accuracy of sensors, updating system software, etc. During maintenance, the system can switch to backup equipment to ensure a continuous construction process.

[0099] Step 10: After the task is completed, the staff will shut down the machine learning control system according to the operating procedures. Conduct a comprehensive inspection of the equipment, record the key parameters and data of the construction, and prepare for the next construction.

[0100] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0101] The above is a detailed introduction to a machine learning-based shield mud flocculation and filtration system, an optimization method, and a computer-readable storage medium provided by the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present invention, several improvements and modifications can be made to the present invention, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.

Claims

1. An optimization method for shield mud flocculation and filter press system based on machine learning, characterized in that: The optimization method comprises the following steps: S1. Obtain shield mud data and perform preprocessing; S2, select characteristic data of shield mud; S3, standardizing and normalizing the feature data; S4, dividing the processed feature data into a training set and a test set; S5, using an improved neural network model for training; S6. Using a particle swarm optimization algorithm to obtain the best neural network model parameters, and obtain a trained neural network model; S7, deploying the trained neural network model to the shield mud flocculation and filtration system; S8, comparing the predicted treatment effect of the shield mud obtained by the trained neural network model with the actual treatment effect of the shield mud, obtaining the optimal parameters of the shield mud flocculation filter press system and implementing them; S9. Regularly test the performance of the neural network model and compare the predicted treatment effect of the shield mud obtained by the neural network model with the actual treatment effect of the shield mud.

2. The optimization method according to claim 1, characterized in that: In step S1, the preprocessing includes filling in missing data in the acquired shield mud data and deleting abnormal data in the acquired shield mud data.

3. The optimization method according to claim 1, characterized in that: The shield mud data include the temperature and concentration of the shield mud, the dosage of the flocculant, the flocculation time, the stirring speed and the filter pressing speed.

4. The optimization method according to claim 1, characterized in that: In step S4, the processed feature data is divided into a training set and a test set in a ratio of 4:

1.

5. The optimization method according to claim 1, characterized in that: In step S5, the neural network model includes multiple convolutional layers, multiple pooling layers, multiple activation functions and a fully connected layer. The convolution layer is used to increase the receptive field of the neural network model and improve the feature extraction capability of shield mud data; The pooling layer is used to reduce data dimension, reduce computational complexity, and improve the operating efficiency of the neural network model; The fully connected layer is used to obtain the prediction results of the shield mud treatment effect; The activation function is used to enhance the nonlinear fitting ability of the neural network model.

6. The optimization method according to claim 5, characterized in that: The neural network model includes a first convolutional layer, a first activation function, a first pooling layer, a second convolutional layer, a second activation function, a second pooling layer, a third convolutional layer, a third activation function, a third pooling layer and a fully connected layer connected in sequence, and the first activation function, the second activation function and the third activation function are ReLU activation functions.

7. The optimization method according to claim 1, characterized in that: In step S5, the training using the improved neural network model is specifically as follows: S51, training the neural network model through historical data of shield mud and shield mud flocculation and filter pressing system operation parameters, and adjusting the neural network model parameters using a back propagation algorithm and an optimizer; S52. Use training sets and validation sets for supervised learning to improve the generalization ability of the neural network model.

8. The optimization method according to claim 1, characterized in that: The treatment effect includes the moisture content, density, cohesion and tensile strength of the mud cake after the treatment.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores an optimization program for a shield mud flocculation and filtration system. When the optimization program is executed by a processor, the steps of the optimization method according to any one of claims 1 to 8 are implemented.

10. A shield mud flocculation and filtration system based on machine learning, characterized in that: The shield mud flocculation and filtration system comprises a shield mud extraction pump, a primary flocculation system, an automatic flocculant dosing system, a clear water tank, a slurry storage tank, a slurry pump, a secondary dewatering screen, a plate and frame filter press, a mud cake collection box and a flocculation and filtration control system. The slurry outlet of the shield mud extraction pump is connected with the primary flocculation system through a pipeline, the drug outlet of the automatic flocculant dosing system is connected with the primary flocculation system through a pipeline, the upper water outlet of the primary flocculation system is connected with the clear water tank through a pipeline, the lower slurry outlet of the primary flocculation system is connected with the slurry storage tank through a pipeline, the slurry storage tank is connected with the slurry pump through a pipeline, the slurry pump is connected with the secondary dewatering screen through a pipeline, the secondary dewatering screen is connected with the plate and frame filter press, and the flocculation and filtration control system is electrically connected with the automatic flocculant dosing system and the plate and frame filter press; The flocculation and filtration control system includes a storage module, a processing module and a monitoring module. The storage module includes the computer-readable storage medium as described in claim 9. The processing module is used to execute the program stored in the computer-readable storage medium. The monitoring module is used to obtain the operating parameters of the shield mud flocculation and filtration system and the shield mud data.

Images