Intelligent control method and system for slurry treatment of slurry shield tunnel

Abstract

The invention discloses an intelligent control method and system for muddy water treatment of a muddy water shield tunnel, and relates to the technical field of muddy water treatment.The method comprises the steps that physical characteristic data and chemical characteristic data of muddy water are collected in real time, preliminary treatment is conducted through edge computing nodes, and standardized muddy water characteristic data flow is output; carrying out preliminary screening operation on muddy water according to an intelligent separation strategy, and screening out medium-particle impurities; through self-adaptive dynamic vortex regulation and control, the muddy water rotational flow speed and the muddy water pressure are regulated, and medium particle impurities are subjected to rotational flow separation; carrying out deep separation treatment on medium-particle impurities subjected to cyclone separation, and outputting clear water and deposited mud cakes; clear water flows back to the shield tunneling machine through pumping equipment to be recycled, and deposited mud cakes are subjected to dehydration treatment. According to the invention, medium particle impurities are accurately separated according to the change of muddy water characteristics by utilizing a regulation and control method of dynamically adjusting the rotational flow speed and pressure, so that the muddy water separation efficiency is effectively improved, and the energy consumption is reduced.

Description

Technical Field

[0001] The invention relates to the technical field of muddy water treatment, and in particular to an intelligent control method and system for muddy water treatment in a muddy water shield tunnel. Background Art

[0002] With the continuous development of tunnel construction technology, slurry shield machines are increasingly used in underground projects. Slurry shield machines often encounter slurry treatment problems during construction, especially how to efficiently separate impurities in slurry and realize the recycling of slurry. Existing slurry treatment methods mainly rely on traditional physical and chemical separation methods, such as cyclone separation, sedimentation in sedimentation tanks, water filtration equipment, etc. Although these methods have improved the efficiency of slurry treatment to a certain extent, they still have certain limitations when facing complex environments and real-time changes in slurry characteristics. For example, traditional methods are often inefficient in dealing with tiny particles and changes in chemical composition in slurry, and cannot respond to dynamic changes in slurry characteristics in a timely manner. In addition, the system has poor adaptability and is difficult to meet the high-precision requirements for slurry quality in shield construction.

[0003] Although the introduction of intelligent technology in recent years has provided new solutions for mud and water treatment, existing intelligent control methods mostly focus on the application of a single technology and lack comprehensive consideration of the complexity and dynamic changes of mud and water. In addition, the existing technology still relies on a single sensor for mud and water characteristic data collection and processing, and the data processing accuracy is low, making it difficult to achieve real-time and accurate mud and water separation and optimization control. Therefore, how to efficiently and in real time process mud and water characteristic data and adjust the separation strategy in real time based on these data has become a key technical problem to improve the efficiency of slurry shield tunnel construction. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides an intelligent control method for mud water treatment in a mud water shield tunnel to solve the problem of insufficient real-time monitoring and dynamic regulation in the existing mud water treatment methods in mud water shield tunnels.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: In a first aspect, the present invention provides an intelligent control method for mud water treatment in a mud shield tunnel, which comprises collecting physical property data and chemical property data of mud water in real time, performing preliminary processing through edge computing nodes, and outputting a standardized mud water property data stream; Use dynamic knowledge graph to perform strategy matching and reasoning on standardized mud and water characteristic data streams to generate intelligent separation strategies; Perform preliminary screening of muddy water according to the intelligent separation strategy to screen out medium-sized particle impurities; Through adaptive dynamic vortex control, the mud and water cyclone speed and mud and water pressure are adjusted to perform cyclone separation on medium-sized particle impurities. The medium-sized particle impurities after cyclone separation are deeply separated and the clear water and sediment cake are output; The clean water is returned to the shield machine through pumping equipment for recycling, and the deposited mud cake is dehydrated.

[0007] As a preferred solution of the intelligent control method for mud water treatment in a mud water shield tunnel of the present invention, wherein: the physical property data includes density, viscosity and particle size distribution of mud water; The chemical property data include pH value, ion concentration and suspended solids concentration; The edge computing node performs preliminary processing and outputs a standardized mud and water characteristic data stream. The specific steps are as follows: The collected physical and chemical property data are transmitted to the edge computing node via a high-speed transmission protocol; Apply Kalman filter to the edge computing node to perform noise filtering and outlier detection processing to form a cleaned characteristic data stream; Apply minimum and maximum normalization to convert the cleaned feature data stream into a unified unit; The unit unified characteristic data stream is weighted and fused through the weighted average algorithm to output the standardized mud and water characteristic data stream.

[0008] As a preferred solution of the intelligent control method for mud water treatment in mud water shield tunnels described in the present invention, wherein: the dynamic knowledge graph is used to perform strategy matching and reasoning on the standardized mud water characteristic data stream to generate an intelligent separation strategy, and the specific steps are as follows: The physical property data and chemical property data in the standardized mud and water property data stream are used as nodes in the knowledge graph; The Pearson correlation coefficient is used to calculate the association strength between nodes and initialize the edge weights between nodes to form the basic structure of the knowledge graph; Whenever the standardized mud and water characteristic data is output, the weighted incremental update model is used to adjust the edge weights of the knowledge graph. The expression is: ; in, Represents physical properties node Chemical Properties Node At the next moment The edge weight when represents the learning rate, Indicates at time Time Physics Node The data value of Indicates the current time point Chemical Properties Node The data value of represents the time decay factor, represents the initial time point, From the initial time point To the current time the time interval that has elapsed; The swirl velocity, mud flow rate and reagent dosage are selected as process control parameters; Record the regular changes of standardized mud water characteristic data under different environments, and extract the association rules between standardized mud water characteristic data and process control parameters through association rule mining algorithm; For association rules, rule induction method is used to obtain symbolic rules; Use multi-layer perceptron and attention mechanism as the network architecture of deep neural network; The nonlinear relationship between the standardized mud-water characteristic data and the process control parameters is modeled through a deep neural network, and the mud-water separation operation parameters are output; The symbolic rules are combined with the mud-water separation operation parameters, and the adaptive fusion method is used to generate the basic guidance value of the intelligent separation strategy, which is expressed as: ; in, represents the basic guidance value of the intelligent separation strategy, represents the fusion coefficient, is the symbol rule, It is the mud-water separation operation parameter; Using the particle swarm optimization algorithm, the basic guidance values ​​of the intelligent separation strategy are mapped to the operating parameters of the mud-water separation equipment, and the mud-water separation equipment is controlled and adjusted; Collect the feedback parameters after control adjustment, and initialize the intelligent separation strategy according to the feedback parameters.

[0009] As a preferred solution of the intelligent control method for mud water treatment in mud water shield tunnels of the present invention, wherein: the mud water is preliminarily screened according to the intelligent separation strategy to screen out medium-sized particle impurities, and the specific steps are as follows: Input the intelligent separation strategy into the cyclone separator to set the screening parameters during the preliminary screening process; Based on the screening parameters, the cyclone separator uses the rotational motion to generate the centrifugal force of the water flow to separate the particles of different sizes and densities into layers; During the stratified sedimentation process, the sedimentation trajectories of different particles are recorded by a particle tracking algorithm; According to the sedimentation trajectory of the particles, the diversion valve opening and mud flow rate are adjusted to separate the mud and water and discharge the medium-sized particle impurities.

[0010] As a preferred solution of the intelligent control method for mud water treatment in mud water shield tunnels of the present invention, wherein: the mud water cyclone speed and mud water pressure are adjusted by adaptive dynamic vortex control to perform cyclone separation on medium-sized particle impurities, and the specific steps are as follows: Assume that there are multiple subtasks as the cyclone separation task for medium-sized impurities, and each subtask is in charge of a heterogeneous agent and a collaborative heterogeneous agent; The attention mechanism is used to dynamically adjust the collaboration weight between each heterogeneous agent, and the expression is: ; in, Indicates The output of heterogeneous agents is is the index variable of the heterogeneous agent, represents the index variable of the collaborative heterogeneous agents, Indicates The input data vector of heterogeneous agents, Indicates The input data vector of the collaborative heterogeneous agents, Indicates The linear regression function of the heterogeneous agents based on the input data vector, Indicates The linear regression function of the collaborative heterogeneous agents based on the input data vector; Through the reinforcement learning algorithm, the cyclone separation task corresponding to each heterogeneous agent is optimized; Define an immediate linear reward function for each heterogeneous agent and update the immediate linear reward function through the Bellman equation; Cyclone separation is performed through a cyclone separation task, and an eddy current field is formed during the cyclone separation process; Bayesian optimization is used to globally predict and optimize the eddy current field, and the expression is: ; in, represents the optimal control strategy parameters, represents the current control strategy parameter vector, Indicates search Make the objective function in the brackets reach the maximum value, Represents the separation efficiency function The expected value of represents the adjustment coefficient, Represents the separation efficiency function The variance of Apply the optimal control strategy parameters to the cyclone equipment control to dynamically adjust the muddy water cyclone speed and muddy water pressure; According to the adjusted mud-water cyclone speed and mud-water pressure, the separation efficiency is calculated using the separation efficiency function to continuously optimize the separation effect of medium-sized particle impurities.

[0011] As a preferred solution of the intelligent control method for mud water treatment in mud water shield tunnels of the present invention, the medium-sized particle impurities after cyclone separation are subjected to deep separation treatment to output clear water and sediment mud cake. The specific steps are as follows: The medium-sized impurities separated by cyclone are sent to the ultrasonic separation tank; The ultrasonic separation tank uses low-frequency ultrasonic waves to excite the sound waves, destroying the cohesion between fine particles, causing large particles to settle to the bottom of the tank to form a sediment layer, and water is discharged from the upper layer; The water is sent to the magnetic separation device, which uses a strong magnetic field to remove ferromagnetic impurities in the water and outputs clean water; The sediment in the sediment layer is sent to the sedimentation tank, and the water in the sediment is further removed through water flow stratification and particle sedimentation to form a sediment mud cake.

[0012] As a preferred solution of the intelligent control method for mud water treatment in the mud shield tunnel of the present invention, wherein: the clean water is returned to the shield machine for recycling through the pumping equipment, and the deposited mud cake is dehydrated. The specific steps are as follows: The clean water after ultrasonic and magnetic separation is transported to the shield machine through pumping equipment for recycling; The sediment cake is dehydrated by a belt filter press, and pressure is applied to squeeze the water out of the cake to output the sludge; The sludge is transported to the storage area for final disposal.

[0013] In a second aspect, the present invention provides an intelligent control system for mud water treatment in a mud shield tunnel, including a data acquisition module, a strategy reasoning module, a primary screening operation module, a cyclone separation module, a depth separation module and a reflux dehydration module; The data acquisition module is used to collect physical and chemical property data of muddy water in real time, perform preliminary processing through edge computing nodes, and output standardized muddy water property data stream; The strategy reasoning module is used to use a dynamic knowledge graph to perform strategy matching and reasoning on the standardized mud and water characteristic data stream to generate an intelligent separation strategy; The preliminary screening operation module is used to perform preliminary screening operations on muddy water according to the intelligent separation strategy to screen out medium-sized particle impurities; The cyclone separation module is used to adjust the muddy water cyclone speed and muddy water pressure through adaptive dynamic vortex control to perform cyclone separation on medium-sized particle impurities; The deep separation module is used to perform deep separation on the medium-sized particle impurities after cyclone separation, and output clear water and sediment cake; The reflux dehydration module is used to return clean water to the shield machine for recycling through pumping equipment, and the deposited mud cake is dehydrated.

[0014] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the intelligent control method for mud water treatment in a mud shield tunnel as described in the first aspect of the present invention is implemented.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the intelligent control method for mud water treatment in a mud shield tunnel as described in the first aspect of the present invention is implemented.

[0016] The beneficial effects of the present invention are as follows: the present invention realizes the real-time collection and processing of mud water characteristic data by introducing edge computing, dynamic knowledge graph and deep learning. When processing the physical properties (such as density, viscosity, etc.) and chemical properties (such as pH value, ion concentration, etc.) of mud water, it can quickly respond and adjust the separation strategy to improve the accuracy and real-time performance of data processing. By dynamically adjusting the control method of cyclone speed and pressure, medium-sized particle impurities can be accurately separated according to the changes in mud water characteristics. The mud water separation efficiency is effectively improved and energy consumption is reduced. At the same time, the environmental impact is reduced and the waste treatment cost is reduced through the return of mud water and the dehydration of mud residue. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0018] Figure 1 This is a flow chart of the intelligent control method for mud water treatment in a mud water shield tunnel in Example 1.

[0019] Figure 2 This is a module diagram of the intelligent control system for mud water treatment in a mud water shield tunnel in Example 1. DETAILED DESCRIPTION

[0020] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0021] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0022] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0023] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides an intelligent control method for mud water treatment in a mud water shield tunnel, comprising the following steps: S1. Collect the physical and chemical property data of mud and water in real time, perform preliminary processing through edge computing nodes, and output standardized mud and water property data stream.

[0024] Furthermore, the physical property data include density, viscosity and particle size distribution of muddy water; It should be noted that physical properties refer to the physical attributes of the mud-water system, which directly affect the fluidity and separation effect of the mud-water. Density and viscosity are closely related, and changes in the two can reflect the composition characteristics of the mud-water, while the particle size distribution directly determines the separation efficiency of solid impurities.

[0025] Chemical property data include pH, ion concentration, and suspended solids concentration; It should be noted that chemical properties reflect the acidity and alkalinity, ion concentration and dispersion of the muddy water, which affect the stability of solid particles in the muddy water and their reaction performance with additives. Through accurate acquisition and processing, the pertinence and flexibility of separation operations can be effectively improved.

[0026] The edge computing node performs preliminary processing and outputs the standardized mud and water characteristic data stream. The specific steps are as follows: The collected physical and chemical property data are transmitted to the edge computing node via a high-speed transmission protocol; Apply Kalman filter to the edge computing node to perform noise filtering and outlier detection processing to form a cleaned characteristic data stream; Preferably, processing data through edge computing nodes avoids the delay of transmitting all data to the cloud for processing, making the processing process more real-time, able to quickly respond to changes in mud and water characteristics, and improving the reaction speed of the entire system.

[0027] Apply minimum and maximum normalization to convert the cleaned feature data stream into a unified unit; The unit unified characteristic data stream is weighted and fused through the weighted average algorithm to output the standardized mud and water characteristic data stream.

[0028] Preferably, the weighted average algorithm can comprehensively consider the contribution of different characteristic data and ensure that the data that has the greatest impact on the separation operation is given priority by assigning different weights to each data source, thereby providing more accurate decision support.

[0029] S2. Use dynamic knowledge graph to perform strategy matching and reasoning on standardized mud and water characteristic data streams to generate intelligent separation strategies.

[0030] Furthermore, the physical property data and chemical property data in the standardized mud and water property data stream are used as nodes in the knowledge graph; The Pearson correlation coefficient is used to calculate the association strength between nodes and initialize the edge weights between nodes to form the basic structure of the knowledge graph; The best approach is to build a knowledge graph-based correlation structure of physical and chemical properties. Different from the traditional processing method based on a single property or linear model, the Pearson correlation coefficient is used to quantify the correlation between nodes, so that the complex relationship between different properties can be dynamically captured and utilized. By updating the edge weights with weighted increments, it can self-adjust after real-time data collection, and more accurately reflect the dynamic relationship between physical and chemical properties.

[0031] Whenever the standardized mud and water characteristic data is output, the weighted incremental update model is used to adjust the edge weights of the knowledge graph. The expression is: ; in, Represents physical properties node Chemical Properties Node At the next moment The edge weight when represents the learning rate, Indicates at time Time Physics Node The data value of Indicates the current time point Chemical Properties Node The data value of represents the time decay factor, represents the initial time point, From the initial time point To the current time the time interval that has elapsed; The swirl velocity, mud flow rate and reagent dosage are selected as process control parameters; Record the regular changes of standardized mud water characteristic data under different environments, and extract the association rules between standardized mud water characteristic data and process control parameters through association rule mining algorithm; The better solution is to extract symbolic rules through association rule mining and rule induction methods, which further improves the interpretability and flexibility of the separation strategy. This method can not only model the common changes in mud and water characteristics, but also adapt to the changes in different environments and generate a separation strategy that is more in line with the actual situation.

[0032] For association rules, rule induction method is used to obtain symbolic rules; Specifically, the regular relationship between standardized mud water characteristic data and process control parameters is extracted through association rule mining algorithm, and frequent patterns and associations under different environmental conditions such as mud concentration, particle distribution and pH value changes are identified. Association rules analyze the trend of standardized mud water characteristic data in changing environmental conditions, revealing the logical connection between characteristic data such as mud water density, viscosity, particle size distribution and swirl velocity and mud flow. The rule induction method summarizes and generalizes the extracted frequent patterns and association rules, extracts the core logical relationship between standardized mud water characteristic data and process control parameters such as swirl velocity and mud flow, and converts these logical relationships into symbolic expressions. Symbolic rules clearly describe the association between standardized mud water characteristic data and process control parameters in the form of formulas or symbols, which are used to generate intelligent separation strategies.

[0033] For example, the rule that "the swirl velocity is proportional to the particle size of mud and water" is made more concise and abstract through symbolization, which is convenient for subsequent processing and reasoning. Finally, these symbolic rules are integrated into a symbolic rule set, which provides a basis for the subsequent generation of intelligent separation strategies and ensures that the operation guidance in the mud and water separation process is executable and consistent.

[0034] Use multi-layer perceptron and attention mechanism as the network architecture of deep neural network; The best method is to use multi-layer perceptron and attention mechanism to build a deep neural network, which can effectively process the complex nonlinear relationship between standardized mud water characteristic data and process control parameters. This method overcomes the limitation of traditional linear models that cannot accurately capture complex relationships, making it better able to adapt to changes in different mud water characteristics.

[0035] The nonlinear relationship between the standardized mud-water characteristic data and the process control parameters is modeled through a deep neural network, and the mud-water separation operation parameters are output; Specifically, the network architecture of the deep neural network adopts a combination of multi-layer perceptron and attention mechanism design, in which the multi-layer perceptron is used to capture complex nonlinear features, and the attention mechanism is used to dynamically focus on the characteristic dimensions in the standardized mud water characteristic data that have a greater impact on the separation effect. Standardized mud water characteristic data include physical characteristic data and chemical characteristic data such as mud water density, viscosity, particle size distribution, pH value, ion concentration and suspended matter concentration; process control parameters include swirl speed, mud flow rate and reagent dosage. By inputting standardized mud water characteristic data as the input layer of the deep neural network and using process control parameters as the target output layer, a large amount of historical data is used for training.

[0036] Furthermore, during the training process, the cross entropy loss function is used to measure the error between the prediction results and the actual operating parameters, and the weights of the deep neural network are updated through the back propagation algorithm. After the training is completed, the deep neural network can automatically predict the mud-water separation operating parameters such as swirl velocity, mud flow rate and reagent dosage that best match the current standardized mud-water characteristic data based on the real-time input standardized mud-water characteristic data. Ultimately, these mud-water separation operating parameters are used as an important basis for guiding the operation of the separation equipment to achieve intelligent mud-water separation control.

[0037] The symbolic rules are combined with the mud-water separation operation parameters, and the adaptive fusion method is used to generate the basic guidance value of the intelligent separation strategy, which is expressed as: ; in, represents the basic guidance value of the intelligent separation strategy, represents the fusion coefficient, is the symbol rule, It is the mud-water separation operation parameter; It should be noted that the fusion coefficient is determined by quantifying the contribution of symbolic rules and mud-water separation operation parameters. First, the symbolic rules reflect the conventional operation mode obtained through association rule mining and rule induction, while the mud-water separation operation parameters come from the modeling of data relationships by deep neural networks, representing the operation strategy generated based on standardized mud-water characteristic data.

[0038] It should also be noted that in order to effectively integrate the symbolic rules and mud-water separation operating parameters, the determination of the fusion coefficient depends on the feedback of the actual operating environment and standardized mud-water characteristic data. Through statistical analysis of historical data, the performance of the symbolic rules and mud-water separation operating parameters under different conditions is evaluated, and the relative contribution of each parameter to the separation effect is calculated. According to these relative contributions, the value of the fusion coefficient is adjusted using an optimization algorithm so that it can balance the weights between the symbolic rules and operating parameters under different operating conditions, ensuring that the adjustment follows regularity and has flexible real-time response capabilities. Finally, the fusion coefficient is updated through multiple iterations and fine-tuned in combination with feedback parameters to ensure efficiency and robustness.

[0039] Using the particle swarm optimization algorithm, the basic guidance values ​​of the intelligent separation strategy are mapped to the operating parameters of the mud-water separation equipment, and the mud-water separation equipment is controlled and adjusted; The particle swarm optimization algorithm is preferably used to adjust the equipment operating parameters in the intelligent separation strategy, achieving accurate mapping between the strategy and the equipment. The adaptive fusion method combines the symbolic rules with the mud-water separation operating parameters to provide a more flexible control strategy, so that the mud-water separation equipment can be adjusted according to real-time feedback, ensuring the efficiency and stability of the mud-water separation process.

[0040] Collect the feedback parameters after control adjustment, and initialize the intelligent separation strategy according to the feedback parameters.

[0041] S3. Perform preliminary screening of mud and water according to the intelligent separation strategy to screen out medium-sized particle impurities.

[0042] Furthermore, the intelligent separation strategy is input into the cyclone separator to set the screening parameters during the preliminary screening process; It should be noted that screening parameters are key variables that determine whether different particulate matter in muddy water can be effectively separated, including physical parameters such as the rotation speed, pressure and flow rate of the cyclone. Screening parameters are closely related to the core working principle of cyclone separation and can optimize the separation effect through precise regulation.

[0043] Based on the screening parameters, the cyclone separator uses the rotational motion to generate the centrifugal force of the water flow to separate the particles of different sizes and densities into layers; Preferably, in the traditional separation process, static screening is often relied on, while the present invention greatly improves the particle separation efficiency through the combination of cyclone and centrifugal force, especially when the particle size difference is not large, the high-efficiency power provided by the cyclone can effectively shorten the separation time and improve the separation effect.

[0044] During the stratified sedimentation process, the sedimentation trajectories of different particles are recorded by a particle tracking algorithm; It should be noted that by real-time monitoring of the particle trajectory, the operating parameters of the hydrocyclone can be accurately adjusted to optimize the diversion valve opening and mud flow rate. This greatly improves the feedback control capability during the separation process, can accurately control the particle stratification during the separation process, avoids errors caused by manual experience, and makes the entire separation process more accurate and efficient.

[0045] According to the sedimentation trajectory of the particles, the diversion valve opening and mud flow rate are adjusted to separate the mud and water and discharge the medium-sized particle impurities.

[0046] S4. Through adaptive dynamic vortex control, the mud and water cyclone speed and mud and water pressure are adjusted to perform cyclone separation on medium-sized particle impurities.

[0047] Furthermore, multiple subtasks are set as the cyclone separation task of medium-sized particle impurities, and each subtask is responsible for heterogeneous agents and collaborative heterogeneous agents; It should be noted that the cyclone separation task of medium-sized impurities is divided into multiple subtasks, each of which is completed by heterogeneous agents and collaborative heterogeneous agents, and each subtask focuses on the separation target of impurities in a specific particle size range. Specifically, for each subtask, a heterogeneous agent is constructed to be responsible for the separation operation in a specific range, and is supported by collaborative heterogeneous agents to achieve the optimization of separation efficiency; For example, some heterogeneous agents are mainly oriented to the separation of larger particle impurities, while the collaborative heterogeneous agents provide real-time support for separation accuracy and dynamic adjustment; similarly, the heterogeneous agents responsible for the separation of smaller particle impurities also combine the auxiliary data provided by the collaborative heterogeneous agents for efficient processing. Each heterogeneous agent receives data input directly related to the separation task according to the characteristics of the task, including physical and chemical property parameters such as particle size, density and concentration. The heterogeneous agents use the regression model based on the random forest algorithm and the reinforcement learning algorithm, combined with the optimization suggestions provided by the collaborative heterogeneous agents, to calculate the cyclone separation process and output the separation effect.

[0048] The attention mechanism is used to dynamically adjust the collaboration weight between each heterogeneous agent, and the expression is: ; in, Indicates The output of heterogeneous agents is is the index variable of the heterogeneous agent, represents the index variable of the collaborative heterogeneous agents, Indicates The input data vector of heterogeneous agents, Indicates The input data vector of the collaborative heterogeneous agents, Indicates The linear regression function of the heterogeneous agents based on the input data vector, Indicates The linear regression function of the collaborative heterogeneous agents based on the input data vector; It should be noted that heterogeneous agents refer to a group of independent agents that differ in task execution, algorithm models, processing capabilities, etc. These agents may use different computing methods, learning algorithms, or processing strategies to solve different separation tasks. Specifically in mud-water separation, each heterogeneous agent can be responsible for mud-water separation tasks of different particle sizes, compositions, or physicochemical properties according to its expertise. For example, one agent may focus on the physical separation of larger particles, while another agent focuses on the separation of finer particles or those with more complex chemical properties. Through the design of heterogeneous agents, the advantages of each agent can be fully utilized in multi-task collaboration, thereby improving the overall separation efficiency.

[0049] It should also be noted that collaborative heterogeneous agents refer to multiple heterogeneous agents that collaborate to complete complex separation tasks in the same separation task. These agents not only perform tasks independently, but also need to optimize the overall separation effect through information exchange and collaboration. For example, one agent may be responsible for the initial separation of larger impurities, while another agent further separates smaller particles on this basis. By introducing the attention mechanism, collaborative heterogeneous agents can dynamically adjust the collaboration weights between each other according to task requirements, ensuring that each agent plays its strongest role at the right time. This collaborative approach enables more complex and dynamic separation tasks to be handled, improving separation effects and processing capabilities.

[0050] Preferably, in the traditional cyclone separation process, a single control system is usually relied on for adjustment, which may lead to low efficiency in separating impurities of different particles. The present invention divides the cyclone separation task into multiple subtasks, which are respectively responsible for different heterogeneous agents, and dynamically adjusts the collaboration weights between the agents through the attention mechanism. Each agent can adaptively collaborate with other agents according to the complexity of the task and environmental changes, thereby improving the overall task processing capability. The application of this dynamically adjusted collaboration strategy in multi-task processing can effectively improve the separation efficiency, especially for more accurate processing of medium-particle impurities.

[0051] Through the reinforcement learning algorithm, the cyclone separation task corresponding to each heterogeneous agent is optimized; Specifically, an initial cyclone separation task is set for each heterogeneous agent, and the reward value of the task is determined by environmental feedback information. Then, the strategy optimization method in the reinforcement learning algorithm is used to construct an immediate reward function for each agent, which gives corresponding rewards or penalties according to the separation effect and task completion. Each heterogeneous agent continuously tries different cyclone speeds, pressures and other operating parameters through interaction with the environment, and updates its strategy based on real-time feedback. This process updates the immediate linear reward function through the Bellman equation to form a recursive update mechanism, so that the agent can adjust its behavior strategy after each feedback, thereby gradually improving the efficiency and accuracy of cyclone separation. Finally, the application of the reinforcement learning algorithm enables each heterogeneous agent to adjust its separation task execution strategy according to real-time task requirements and environmental changes in the process of continuous optimization, so as to achieve accurate separation of medium-particle impurities.

[0052] Define an immediate linear reward function for each heterogeneous agent and update the immediate linear reward function through the Bellman equation; Cyclone separation is performed through a cyclone separation task, and an eddy current field is formed during the cyclone separation process; Bayesian optimization is used to globally predict and optimize the eddy current field, and the expression is: ; in, represents the optimal control strategy parameters, represents the current control strategy parameter vector, Indicates search Make the objective function in the brackets reach the maximum value, Represents the separation efficiency function The expected value of represents the adjustment coefficient, Represents the separation efficiency function The variance of It should be noted that the determination process of the adjustment coefficient is carried out through the exploration and utilization balance mechanism in optimization. In this optimization process, first, by calculating the expected value of the separation efficiency, the role of the adjustment coefficient is to control the trade-off between exploration (trying new strategies) and utilization (enhancing known optimal strategies) in the optimization process. Larger adjustment coefficient values ​​tend to improve exploration and increase the extensive exploration of the strategy space to find more efficient operating strategies; while smaller adjustment coefficient values ​​focus more on optimizing the current strategy using existing information to maintain the stability of the separation efficiency. Specifically, the selection of the adjustment coefficient value needs to be adjusted according to the variance of the separation efficiency in multiple experiments. By comparing the performance and stability of different strategies, the adjustment coefficient value that can maintain the optimal separation effect under various operating conditions is selected. Finally, the adjustment coefficient value is carefully adjusted through feedback and analysis of the optimization results, combined with the maximization strategy of the objective function, to ensure that the ideal eddy field control effect can be achieved in practical applications, thereby optimizing the efficiency and accuracy of the mud-water separation process.

[0053] Apply the optimal control strategy parameters to the cyclone equipment control to dynamically adjust the muddy water cyclone speed and muddy water pressure; According to the adjusted mud-water cyclone speed and mud-water pressure, the separation efficiency is calculated using the separation efficiency function to continuously optimize the separation effect of medium-sized particle impurities.

[0054] S5. Perform deep separation on the medium-sized particle impurities after cyclone separation, and output clear water and sediment cake.

[0055] Furthermore, the medium-sized impurities separated by cyclone are sent to an ultrasonic separation tank; The ultrasonic separation tank uses low-frequency ultrasonic waves to excite the sound waves, destroying the cohesion between fine particles, causing large particles to settle to the bottom of the tank to form a sediment layer, and water is discharged from the upper layer; It should be noted that the ultrasonic separation pool uses low-frequency ultrasonic waves to excite sound waves. The ultrasonic generator emits low-frequency sound waves with a frequency of 20kHz to 100kHz. When the sound waves propagate in the water, they excite the particles in the liquid through vibration and cavitation effects, destroying the cohesion between the fine particles. Specifically, the ultrasonic wave forms periodic compression and expansion fluctuations in the water, resulting in local cavitation and the generation of tiny bubbles. These bubbles release a strong impact force when they burst, further tearing the aggregation structure of the fine particles, thereby effectively reducing the adhesion of the particles. As the cohesion between the particles is destroyed, the larger particles gradually settle to the bottom of the pool due to their larger mass and faster sedimentation rate, forming an obvious sedimentation layer. At the same time, the smaller particles in the water continue to remain suspended with the water body due to incomplete sedimentation, and finally the clean water is discharged from the upper layer of the pool. This process ensures that the dispersibility between the particles is effectively improved through the repeated action of ultrasonic waves, thereby promoting the sedimentation of larger particles and the separation of clean water.

[0056] The water is sent to the magnetic separation device, which uses a strong magnetic field to remove ferromagnetic impurities in the water and outputs clean water; The sediment in the sediment layer is sent to the sedimentation tank, and the water in the sediment is further removed through water flow stratification and particle sedimentation to form a sediment mud cake.

[0057] Specifically, after the sediment in the sediment layer is sent into the sedimentation tank through the conveying equipment, the sediment is affected by the water flow in the tank. Through the water flow layered structure, particles of different particle sizes are respectively settled to the bottom of the tank under the guidance of the water flow. Larger particles settle first due to gravity, while smaller particles sink slowly in the lower speed water flow. The water flow speed and flow direction in the sedimentation tank are precisely controlled to optimize the sedimentation rate between particles, thereby forming clear water layers and sediment layers. As the sediment gradually settles to the bottom of the tank, water is continuously separated during the sedimentation process, further removing the water in the sediment, and finally forming a high-concentration sediment cake. The moisture content in the sediment cake is greatly reduced, providing ideal conditions for subsequent solid waste disposal or utilization.

[0058] S6. The clean water is returned to the shield machine through pumping equipment for recycling, and the sediment cake is dehydrated.

[0059] Furthermore, the clean water after ultrasonic and magnetic separation is transported to the shield machine for recycling through pumping equipment; Specifically, the muddy water is physically vibrated by an ultrasonic treatment device to cause large particles of impurities in the water to detach and aggregate, and then the ferromagnetic impurities in the water are separated by a magnetic separation device. After these treatments, the water quality of the clean water is significantly improved and the impurity content is greatly reduced. The clean water after ultrasonic and magnetic separation enters the water storage pipe, and then the clean water is transported to the circulation system of the shield machine through the pumping equipment. In this process, the pumping equipment is responsible for controlling the flow rate and pressure of the clean water to ensure that the water source can be stably supplied to the working area of ​​the shield machine, providing the shield machine with sufficient water for soil excavation and mud-water separation operations. The whole process ensures the efficient recovery and recycling of clean water, minimizes the consumption of external water sources, and improves the working efficiency and environmental protection of the shield machine.

[0060] The best solution is to return the clean water after ultrasonic and magnetic separation to the shield machine through pumping equipment, which reflects the high attention paid to mud water recovery and resource utilization in the technology. Different from the traditional single mud water discharge or treatment method, this step optimizes the recycling of mud water during the operation of the shield machine, greatly reduces the dependence on fresh water sources, and effectively reduces the pressure on the consumption of external water resources.

[0061] The sediment cake is dehydrated by a belt filter press, and pressure is applied to squeeze the water out of the cake to output the sludge; It should be noted that the belt filter press squeezes water out of the mud cake by applying pressure. This step can efficiently separate the water and thus achieve the dehydration of the mud cake. In the mud treatment process, conventional dehydration methods may face the problems of low efficiency and small processing volume, while the belt filter press can provide higher dehydration efficiency through continuous operation.

[0062] The sludge is transported to the storage area for final disposal.

[0063] The present embodiment also provides an intelligent control system for mud and water treatment in a mud and water shield tunnel, comprising: a data acquisition module, a strategy reasoning module, a primary screening operation module, a cyclone separation module, a deep separation module and a reflux dehydration module; the data acquisition module is used to collect physical property data and chemical property data of mud and water in real time, perform preliminary processing through edge computing nodes, and output a standardized mud and water characteristic data stream; the strategy reasoning module is used to use a dynamic knowledge graph to perform strategy matching and reasoning on the standardized mud and water characteristic data stream to generate an intelligent separation strategy; the primary screening operation module is used to perform preliminary screening operations on the mud and water according to the intelligent separation strategy to screen out medium-sized particle impurities; the cyclone separation module is used to adjust the mud and water cyclone speed and mud and water pressure through adaptive dynamic vortex control, and perform cyclone separation on medium-sized particle impurities; the deep separation module is used to perform deep separation processing on the medium-sized particle impurities after cyclone separation, and output clean water and sediment mud cake; the reflux dehydration module is used to return the clean water to the shield machine through a pumping device for recycling, and the sediment mud cake is dehydrated.

[0064] This embodiment also provides a computer device, which is suitable for the case of an intelligent control method for mud water treatment in a mud shield tunnel, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the intelligent control method for mud water treatment in a mud shield tunnel as proposed in the above embodiment.

[0065] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0066] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the intelligent control method for mud water treatment in a mud shield tunnel proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0067] In summary, the present invention realizes the real-time collection and processing of mud water characteristic data by introducing edge computing, dynamic knowledge graph and deep learning. When processing the physical properties (such as density, viscosity, etc.) and chemical properties (such as pH value, ion concentration, etc.) of mud water, it can quickly respond and adjust the separation strategy to improve the accuracy and real-time performance of data processing. By dynamically adjusting the control method of cyclone speed and pressure, medium-sized particle impurities can be accurately separated according to the changes in mud water characteristics. The mud water separation efficiency is effectively improved and energy consumption is reduced. At the same time, the environmental impact is reduced and the waste treatment cost is reduced through the return of mud water and the dehydration of mud residue.

[0068] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An intelligent control method for mud water treatment in a mud water shield tunnel, characterized in that: include, Collect physical and chemical property data of mud and water in real time, perform preliminary processing through edge computing nodes, and output standardized mud and water property data stream; Use dynamic knowledge graph to perform strategy matching and reasoning on standardized mud and water characteristic data streams to generate intelligent separation strategies; Perform preliminary screening of muddy water according to the intelligent separation strategy to screen out medium-sized particle impurities; Through adaptive dynamic vortex control, the mud and water cyclone speed and mud and water pressure are adjusted to perform cyclone separation on medium-sized particle impurities. The medium-sized particle impurities after cyclone separation are deeply separated and the clear water and sediment cake are output; The clean water is returned to the shield machine through pumping equipment for recycling, and the deposited mud cake is dehydrated.

2. The intelligent control method for mud water treatment in a mud water shield tunnel according to claim 1, characterized in that: The physical property data include density, viscosity and particle size distribution of muddy water; The chemical property data include pH value, ion concentration and suspended solids concentration; The edge computing node performs preliminary processing and outputs a standardized mud and water characteristic data stream. The specific steps are as follows: The collected physical and chemical property data are transmitted to the edge computing node via a high-speed transmission protocol; Apply Kalman filter to the edge computing node to perform noise filtering and outlier detection processing to form a cleaned characteristic data stream; Apply minimum and maximum normalization to convert the cleaned feature data stream into a unified unit; The unit unified characteristic data stream is weighted and fused through the weighted average algorithm to output the standardized mud and water characteristic data stream.

3. The intelligent control method for mud water treatment in a mud water shield tunnel according to claim 2, characterized in that: The dynamic knowledge graph is used to perform strategy matching and reasoning on the standardized mud and water characteristic data stream to generate an intelligent separation strategy. The specific steps are as follows: The physical property data and chemical property data in the standardized mud and water property data stream are used as nodes in the knowledge graph; The Pearson correlation coefficient is used to calculate the association strength between nodes and initialize the edge weights between nodes to form the basic structure of the knowledge graph; Whenever the standardized mud and water characteristic data is output, the weighted incremental update model is used to adjust the edge weights of the knowledge graph. The expression is: ； in, Represents physical properties node Chemical Properties Node At the next moment The edge weight when represents the learning rate, Indicates at time Time Physics Node The data value of Indicates the current time point Chemical Properties Node The data value of represents the time decay factor, represents the initial time point, From the initial time point To the current time the time interval that has elapsed; The swirl velocity, mud flow rate and reagent dosage are selected as process control parameters; Record the regular changes of standardized mud water characteristic data under different environments, and extract the association rules between standardized mud water characteristic data and process control parameters through association rule mining algorithm; For association rules, rule induction method is used to obtain symbolic rules; Use multi-layer perceptron and attention mechanism as the network architecture of deep neural network; The nonlinear relationship between the standardized mud-water characteristic data and the process control parameters is modeled through a deep neural network, and the mud-water separation operation parameters are output; The symbolic rules are combined with the mud-water separation operation parameters, and the adaptive fusion method is used to generate the basic guidance value of the intelligent separation strategy, which is expressed as: ； in, represents the basic guidance value of the intelligent separation strategy, represents the fusion coefficient, is the symbol rule, It is the mud-water separation operation parameter; Using the particle swarm optimization algorithm, the basic guidance values ​​of the intelligent separation strategy are mapped to the operating parameters of the mud-water separation equipment, and the mud-water separation equipment is controlled and adjusted; Collect the feedback parameters after control adjustment, and initialize the intelligent separation strategy according to the feedback parameters.

4. The intelligent control method for mud water treatment in a mud water shield tunnel as claimed in claim 3, characterized in that: The preliminary screening operation of muddy water is carried out according to the intelligent separation strategy to screen out medium-sized particle impurities. The specific steps are as follows: Input the intelligent separation strategy into the cyclone separator to set the screening parameters during the preliminary screening process; Based on the screening parameters, the cyclone separator uses the rotational motion to generate the centrifugal force of the water flow to separate the particles of different sizes and densities into layers; During the stratified sedimentation process, the sedimentation trajectories of different particles are recorded by a particle tracking algorithm; According to the sedimentation trajectory of the particles, the diversion valve opening and mud flow rate are adjusted to separate the mud and water and discharge the medium-sized particle impurities.

5. The intelligent control method for mud water treatment in a mud water shield tunnel as claimed in claim 4, characterized in that: The adaptive dynamic vortex control is used to adjust the mud water cyclone speed and mud water pressure to perform cyclone separation on medium-sized particle impurities. The specific steps are as follows: Assume that there are multiple subtasks as the cyclone separation task for medium-sized impurities, and each subtask is in charge of a heterogeneous agent and a collaborative heterogeneous agent; The attention mechanism is used to dynamically adjust the collaboration weight between each heterogeneous agent, and the expression is: ； in, Indicates The output of heterogeneous agents is is the index variable of the heterogeneous agent, represents the index variable of the collaborative heterogeneous agents, Indicates The input data vector of heterogeneous agents, Indicates The input data vector of the collaborative heterogeneous agents, Indicates The linear regression function of the heterogeneous agents based on the input data vector, Indicates The linear regression function of the collaborative heterogeneous agents based on the input data vector; Through the reinforcement learning algorithm, the cyclone separation task corresponding to each heterogeneous agent is optimized; Define an immediate linear reward function for each heterogeneous agent and update the immediate linear reward function through the Bellman equation; Cyclone separation is performed through a cyclone separation task, and an eddy current field is formed during the cyclone separation process; Bayesian optimization is used to globally predict and optimize the eddy current field, and the expression is: ； in, represents the optimal control strategy parameters, represents the current control strategy parameter vector, Indicates search Make the objective function in the brackets reach the maximum value, Represents the separation efficiency function The expected value of represents the adjustment coefficient, Represents the separation efficiency function The variance of Apply the optimal control strategy parameters to the cyclone equipment control to dynamically adjust the muddy water cyclone speed and muddy water pressure; According to the adjusted mud-water cyclone speed and mud-water pressure, the separation efficiency is calculated using the separation efficiency function to continuously optimize the separation effect of medium-sized particle impurities.

6. The intelligent control method for mud water treatment in a mud water shield tunnel according to claim 5, characterized in that: The medium-sized particle impurities after cyclone separation are subjected to deep separation treatment to output clear water and sediment cake. The specific steps are as follows: The medium-sized impurities separated by cyclone are sent to the ultrasonic separation tank; The ultrasonic separation tank uses low-frequency ultrasonic waves to excite the sound waves, destroying the cohesion between fine particles, causing large particles to settle to the bottom of the tank to form a sediment layer, and water is discharged from the upper layer; The water is sent to the magnetic separation device, which uses a strong magnetic field to remove ferromagnetic impurities in the water and outputs clean water; The sediment in the sediment layer is sent to the sedimentation tank, and the water in the sediment is further removed through water flow stratification and particle sedimentation to form a sediment mud cake.

7. The intelligent control method for mud water treatment in a mud water shield tunnel according to claim 6, characterized in that: The clean water is returned to the shield machine through the pumping equipment for recycling, and the sediment cake is dehydrated. The specific steps are as follows: The clean water after ultrasonic and magnetic separation is transported to the shield machine through pumping equipment for recycling; The sediment cake is dehydrated by a belt filter press, and pressure is applied to squeeze the water out of the cake to output the sludge; The sludge is transported to the storage area for final disposal.

8. An intelligent control system for mud water treatment in a slurry shield tunnel, based on the intelligent control method for mud water treatment in a slurry shield tunnel according to any one of claims 1 to 7, characterized in that: Including data acquisition module, strategy reasoning module, primary screening operation module, cyclone separation module, deep separation module and reflux dehydration module; The data acquisition module is used to collect physical and chemical property data of muddy water in real time, perform preliminary processing through edge computing nodes, and output standardized muddy water property data stream; The strategy reasoning module is used to use a dynamic knowledge graph to perform strategy matching and reasoning on the standardized mud and water characteristic data stream to generate an intelligent separation strategy; The preliminary screening operation module is used to perform preliminary screening operations on muddy water according to the intelligent separation strategy to screen out medium-sized particle impurities; The cyclone separation module is used to adjust the muddy water cyclone speed and muddy water pressure through adaptive dynamic vortex control to perform cyclone separation on medium-sized particle impurities; The deep separation module is used to perform deep separation on the medium-sized particle impurities after cyclone separation, and output clear water and sediment cake; The reflux dehydration module is used to return clean water to the shield machine for recycling through pumping equipment, and the deposited mud cake is dehydrated.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the intelligent control method for mud water treatment in a mud shield tunnel described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the intelligent control method for mud water treatment in a mud shield tunnel according to any one of claims 1 to 7 are implemented.

Images