Segmented gas feeding optimization control method for dual-wheel milling gas lift system

Abstract

The application discloses a sectional gas feeding optimization control method of a double-wheel milling gas lifting system, belongs to the technical field of cutoff wall construction, and is used for solving the problems of rigid control, poor adaptability, high energy consumption, easy blockage and insufficient power in super-deep working conditions of an existing gas lifting system gas feeding control. The application adopts global pre-training and migration learning to construct a construction correction model; a depth threshold value is adaptively adjusted according to geology, conventional depth is divided into three points for grading gas injection, and super-deep working conditions are divided into four points for grading gas injection; gas injection parameters are distributed through a quantitative formula, and PID closed-loop regulation and control are performed; and the functions of grading anti-blocking, sensor emergency, gas source emergency and leakage detection are integrated. The application has great deslagging efficiency, effectively saves energy, guarantees construction continuity and safety, and is suitable for various complex geological working conditions.

Description

Technical Field

[0001] This invention relates to the field of anti-seepage wall construction technology, specifically to a method for optimizing and controlling segmented air supply in a dual-wheel milling airlift system. Background Technology

[0002] In the field of anti-seepage wall construction technology, twin-wheel trenching machines, with their efficient ground cutting capabilities, have become the core equipment for trenching construction of underground continuous walls in large-scale projects such as deep foundation pits and super high-rise buildings. Their working principle involves a hydraulic system driving two milling wheels at the bottom of the cutter head to rotate, cutting and breaking up the strata. Then, an air-lift reverse circulation method is used to discharge the sediment at the bottom of the trench along with the drilling mud, enabling continuous trenching. Compared to traditional pump-suction slag removal methods, air-lift slag removal has advantages such as low power consumption, simple supporting equipment, low maintenance costs, good slag removal effect in deep trenches, and strong ability to handle large-diameter slag blocks. It has been widely used in various twin-wheel trenching machine slag removal systems.

[0003] As a key component of twin-wheel trenching machines, the air-lift slag removal system's air supply control directly determines slag removal efficiency, construction quality, and system energy consumption, while also affecting the safety of the construction process. However, current air supply control methods for twin-wheel trenching air-lift systems still have many technical shortcomings, making it difficult to meet the actual needs of complex construction conditions. These shortcomings are mainly reflected in the following aspects:

[0004] Firstly, existing air-lift systems mostly employ a single air injection point or a fixed multi-point synchronous air supply control method, failing to dynamically adjust the air supply strategy according to changes in the milling depth. During twin-wheel milling operations, there are significant differences in the pressure within the trench, the mud flow characteristics, and the air-lift slag removal mechanism at different stages: shallow, medium, and deep trenches. Excessive air injection pressure in the shallow trench stage can easily lead to mud splashing and energy waste, while insufficient air injection pressure in the deep trench stage will result in insufficient slag removal power and sediment accumulation. The fixed air supply mode cannot adapt to the air-lift slag removal characteristics at each stage, leading to unstable slag removal efficiency and difficulty in meeting the slag removal needs at different depths.

[0005] Secondly, the setting of air supply parameters lacks scientific basis and has poor adaptability. In existing technologies, key parameters such as injection pressure and injection volume rely heavily on the operator's experience, failing to fully consider the influence of variables such as geological parameters, mud performance parameters, and ambient temperature at the construction site, and cannot be dynamically corrected based on real-time working conditions during construction. Furthermore, while some air-lift systems attempt to use fixed models to calculate air supply parameters, these models are not trained and optimized using historical construction data, nor can they be updated online based on real-time construction data. This results in significant deviations between the calculated air supply parameters and the actual optimal parameters, leading to either incomplete slag removal and excessive sediment at the bottom of the trench, affecting the quality of the diaphragm wall formation, or excessive air injection and a substantial increase in energy consumption, failing to meet the requirements of energy-saving construction.

[0006] Secondly, the allocation of gas delivery parameters at each injection point is unreasonable. When using multi-point gas injection, existing technologies often employ average or fixed-ratio allocation methods, without considering differences in height and gas lift efficiency coefficients among the injection points. This results in insufficient gas delivery at some points and redundant gas delivery at others, which not only reduces the overall gas lift slag removal efficiency but also further increases system energy consumption. Additionally, local airflow turbulence may cause unstable mud flow within the slag discharge pipe, exacerbating the risk of sludge deposition.

[0007] In summary, current air supply control methods for dual-wheel milling air-lift systems suffer from technical defects such as poor adaptability, unscientific parameter settings, unreasonable allocation, and lagging blockage prevention. These defects result in low slag removal efficiency, high energy consumption, and poor construction stability, failing to meet the demands for efficient, energy-saving, and safe construction under complex conditions. Therefore, developing a segmented air supply optimization control method for dual-wheel milling air-lift systems that can adapt to different construction stages, dynamically optimize air supply parameters, and accurately predict potential blockages has become a pressing technical problem for those skilled in the art. Summary of the Invention

[0008] To address the technical problems existing in the prior art, this application provides a segmented air delivery optimization control method for a dual-wheel milling air lift system.

[0009] To achieve the above objectives, the technical solution adopted in this application is: a segmented air supply optimization control method for a dual-wheel milling air lift system, comprising the following steps:

[0010] S1: Obtain the geological parameters, mud performance parameters, and ambient temperature parameters of the construction location, and input the above parameters into the pre-constructed construction correction model; the construction correction model is trained based on historical construction data, and the parameters are updated online every 5-10 minutes according to the collected real-time construction data, and the update is triggered immediately when the geological parameters change abruptly.

[0011] S2: Based on the current milling depth H, the construction process is divided into shallow trench stage, medium trench stage and deep trench stage. According to the air lift slag discharge characteristics of different construction stages, the corresponding air injection points arranged in advance along the slag discharge pipe axis are dynamically opened, and the initial air injection pressure and initial air injection volume are set for each opened air injection point.

[0012] S3: During construction, the density parameters of the circulating mud and the slag content parameters at the outlet of the slag discharge pipe are collected in real time. These real-time parameters are input into the construction correction model to calculate the optimal total gas injection pressure of the system under the current working conditions. and optimal total gas injection volume ;

[0013] S4: Based on the height difference of each injection point and the current air lift efficiency coefficient (calculated by the construction correction model based on real-time mud viscosity and gas density), the optimal total injection pressure is... and optimal total gas injection volume The gas is allocated to each injection point, and the target pressure and target flow rate of each injection point are obtained.

[0014] S5: Real-time acquisition of actual pressure and flow values ​​at each injection point, comparison with the corresponding target values, and dynamic adjustment of the valve opening at each injection point through a PID controller to ensure that the actual parameters track the target parameters;

[0015] S6: Pre-establish a standard characteristic curve library of gas injection pressure-slag discharge flow rate under different working conditions, collect the current gas injection pressure and slag discharge flow rate data in real time and generate real-time characteristic curves, compare the real-time characteristic curves with the standard characteristic curves under the corresponding working conditions; when the deviation between the two exceeds the preset threshold, it is determined to be a precursor to slag discharge pipe blockage and triggers graded anti-blockage measures.

[0016] Furthermore, in step S2, the criteria for dividing the construction stages are as follows:

[0017] When the milling groove depth H≤H1, it is the shallow groove stage, and only the lowest first air injection point is opened;

[0018] When the milling groove depth H1<H≤H2, it is the middle groove stage, and the first and second air injection points are opened at the same time.

[0019] When the milling groove depth H > H2, it is the deep groove stage, and the first, second, and third air injection points are opened simultaneously.

[0020] H1 and H2 are depth thresholds preset based on geological conditions and mud properties.

[0021] Furthermore, in step S3, the construction correction model adopts an improved BP neural network model, whose input layer nodes include geological parameters, mud density, slag content, milling depth and ambient temperature, and output layer nodes include optimal total gas injection pressure and optimal total gas injection volume.

[0022] Furthermore, in step S4, the formula for allocating the gas injection parameters is:

[0023] ;

[0024] ;

[0025] in, Let i be the target pressure value for the i-th injection point. For the target flow value, H represents the height of the i-th air injection point from the ground, and H represents the current milling groove depth. The set minimum safe immersion depth correction value; and These are the pressure distribution coefficient and flow distribution coefficient for the i-th gas injection point, respectively;

[0026] The pressure distribution coefficient and flow distribution coefficient All are dimensionless constants, and for all currently activated gas injection points, they satisfy the following conditions: , , where n is the total number of currently activated gas injection points.

[0027] Furthermore, in step S6, the graded prevention and blocking measures include:

[0028] Level 1 warning: When the deviation is between 10% and 20%, automatically increase the current gas injection pressure by 5% to 10% and increase the gas injection volume by 10% to 15%, and continue to observe for 30 seconds;

[0029] Level 2 warning: When the deviation is between 20% and 30%, turn off the milling head rotation, maintain the air lift slag removal state, and alternately open and close each air injection point to perform pulse clearing.

[0030] Level 3 warning: When the deviation exceeds 30%, construction is stopped, the milling head is raised to a safe height, and an audible and visual alarm is issued to prompt manual intervention.

[0031] Furthermore, it also includes step S7: real-time monitoring of the working status of each sensor, and when a sensor fault is detected, automatically switching to the experience control mode based on historical data and issuing a fault alarm.

[0032] Furthermore, it also includes gas source pressure monitoring and emergency response procedures:

[0033] The outlet pressure of the compressed air source is collected in real time. When the air source pressure is lower than the preset lower limit, the uppermost air injection point is automatically closed and the remaining air volume is concentrated and distributed to the lower air injection points.

[0034] At the same time, reduce the milling head feed speed and issue an alarm for insufficient air supply pressure.

[0035] Furthermore, it also includes a leak detection procedure for the gas injection line:

[0036] The flow deviation rate of each gas injection point is calculated in real time. When the deviation rate between the actual flow rate and the target flow rate of a certain gas injection point exceeds 30% and lasts for more than 10 seconds, it is determined that the gas injection pipeline has leaked.

[0037] Automatically close the regulating valve at the leaking gas injection point, reallocate the corresponding gas injection parameters to other open gas injection points, and issue a leakage alarm.

[0038] Further, the pre-training steps of the construction correction model are as follows:

[0039] Collect historical construction data from multiple different construction sites and different geological conditions to construct a global construction data set;

[0040] Pre-train a basic model based on the global construction data set;

[0041] At a new construction site, use the construction data of the first N meters at this site to fine-tune the basic model through transfer learning to obtain a construction correction model adapted to this site.

[0042] Further, the depth thresholds H1 and H2 are adaptively adjusted according to geological conditions:

[0043] When the milling groove depth H ≤ 60m:

[0044] If the construction formation is a soft soil formation, H1 = 15m, H2 = 30m;

[0045] If the construction formation is a sandy cobble formation, H1 = 10m, H2 = 25m;

[0046] If the construction formation is a hard rock formation, H1 = 8m, H2 = 20m;

[0047] When 200m < H ≤ 300m for the milling groove depth:

[0048] Arrange 4 gas injection points along the axial direction of the slag discharge pipe, and the spacing between the gas injection points is 35 - 40m;

[0049] The depth thresholds are successively 、 、H3, where is the depth from single point to two points, is the depth from two points to three points, and H3 is the depth from three points to four points;

[0050] The reference values for the sandy cobble formation are: = 40m, = 80m, H3 = 120m;

[0051] The reference values for the hard rock formation are: = 30m, = 65m, H3 = 100m;

[0052] Beneficial effects:

[0053] 1. This invention divides the construction process into three stages—shallow, medium, and deep—based on the current milling depth H. It dynamically activates the corresponding air injection points at each stage, taking into account the air-lift slag removal characteristics, and sets appropriate initial air injection pressures and volumes. This completely solves the problem in existing technologies where single air injection or fixed multi-point air delivery cannot adapt to different depth conditions. In the shallow trench stage, excessively high injection pressure avoids mud splashing and energy waste. In the deep trench stage, sufficient slag removal power is ensured, preventing slag accumulation. In the medium trench stage, a balance between slag removal efficiency and energy consumption is achieved, thus consistently improving slag removal efficiency throughout the entire construction process, ensuring that the slag at the bottom of the trench meets standards, improving the construction quality of the diaphragm wall, and allowing for adjustments based on changes in working depth and geological conditions.

[0054] 2. This invention introduces a construction correction model trained on historical construction data and updated online based on real-time construction data, taking into account initial parameters such as geology, mud properties, and ambient temperature; then, through step S3, it collects real-time parameters such as circulating mud density and slag content at the slag discharge pipe outlet, and calculates the optimal total gas injection pressure under the current working conditions using the construction correction model. and optimal total gas injection volume This significantly reduces reliance on operator experience and solves the problem of fixed models being unable to adapt to changes in working conditions. The online updating capability of the construction correction model can respond to fluctuations in working conditions during construction in real time, ensuring that the gas supply parameters are always within the optimal range. This avoids incomplete slag removal caused by insufficient gas injection and eliminates energy waste caused by excessive gas injection, achieving a synergistic improvement in slag removal efficiency and energy-saving benefits.

[0055] 3. This invention combines the height difference between each injection point and the current gas lift efficiency coefficient to determine the optimal total injection pressure. and optimal total gas injection volume By scientifically allocating gas to each injection point, the target pressure and flow rate values ​​for each point are obtained, solving the problems of insufficient gas supply and redundant gas supply at some points caused by average or fixed-ratio allocation in existing technologies. Reasonable parameter allocation can avoid local airflow turbulence, ensure that each injection point works synergistically, improve the overall working efficiency of the gas lift system, further reduce system energy consumption, and at the same time reduce the risk of sludge deposition in the sludge discharge pipe, extending the service life of system components.

[0056] 4. This invention, through step S5, collects the actual pressure and flow parameters of each injection point in real time, compares them with the target parameters, and then dynamically adjusts the opening of the regulating valve at each point using a PID controller. This ensures that the actual parameters accurately track the target parameters, forming a complete closed-loop control mechanism. This mechanism solves the shortcomings of existing technologies where injection parameters cannot be corrected in real time and are prone to deviating from the optimal value. It can quickly respond to parameter fluctuations caused by changes in operating conditions, ensuring that the operating parameters of each injection point always match the current operating conditions, guaranteeing the stability and reliability of the gas lift system, and reducing construction failures caused by parameter deviations. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is a flowchart illustrating the segmented air delivery optimization control method for a dual-wheel milling air lift system according to an embodiment of this application. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0060] Example

[0061] Please refer to Figure 1 This embodiment provides a segmented air delivery optimization control method for a dual-wheel milling air lift system, including the following steps:

[0062] S1: Obtain the geological parameters, mud performance parameters, and ambient temperature parameters of the construction location, and input the above parameters into the pre-constructed construction correction model; the construction correction model is trained based on historical construction data, and the parameters are updated online every 5-10 minutes according to the collected real-time construction data, and the update is triggered immediately when the geological parameters change abruptly.

[0063] In the process, three key basic parameters of the construction site are accurately obtained through professional equipment such as geological survey equipment, mud performance testing instruments, and temperature sensors. Geological parameters include stratum hardness, particle size distribution, and rock layer distribution, which directly affect the difficulty of trenching and the amount of sediment.

[0064] The aforementioned mud performance parameters include mud density, viscosity, and sand content. As the carrier for air-lift slag removal, the mud's performance directly determines the air-lift efficiency and slag removal effect. The aforementioned ambient temperature parameter affects gas density, mud viscosity, and equipment operating accuracy, thus impacting the adaptability of injection parameters. These three types of parameters comprehensively cover the construction environment, slag removal carrier, and external influences, ensuring the integrity of the basic data.

[0065] The aforementioned construction correction model is based on a large amount of historical construction data, covering injection parameters, slag removal effects, and energy consumption data under different geological conditions, trench depths, and mud properties. It is trained and constructed using machine learning algorithms such as regression analysis and neural networks, possessing both initial adaptation and real-time correction capabilities. In the initial stage, three types of basic parameters are input into the model. The model, combined with historical data from similar working conditions, provides an initial benchmark for setting initial injection parameters and calculating optimal total injection parameters. Simultaneously, the model can be updated online based on various parameters collected in real-time during construction, continuously correcting model parameters, improving the accuracy of calculation results, and avoiding the drawbacks of fixed models being unable to adapt to changes in working conditions. This achieves adaptive control based on changes in working conditions, model, and parameters. The construction correction model, trained on historical data, can quickly match the current construction conditions. Simultaneously, through online updates, it responds in real-time to fluctuations in working conditions during construction, such as sudden geological changes and changes in mud properties, ensuring that the subsequently calculated injection parameters always closely match the actual working conditions, avoiding the problem of excessive calculation deviations in fixed models.

[0066] S2: Based on the current milling depth H, the construction process is divided into shallow trench stage, medium trench stage and deep trench stage. According to the air lift slag discharge characteristics of different construction stages, the corresponding air injection points arranged in advance along the slag discharge pipe axis are dynamically opened, and the initial air injection pressure and initial air injection volume are set for each opened air injection point.

[0067] In this step, based on the current milling depth H and combined with the engineering experience of dual-wheel milling air-lift slag removal, the construction process is divided into shallow trench stage, medium trench stage, and deep trench stage. The basis for this division is that there are significant differences in trench pressure, mud flow characteristics, and air-lift slag removal mechanism at different depths: in the shallow trench stage, the trench pressure is low, the mud column height is low, and the power required for air lift is low; in the deep trench stage, the trench pressure is high, the mud column height is high, and a higher air injection pressure is required to achieve effective slag removal; the medium trench stage is between the two, and a balance needs to be struck between slag removal efficiency and energy consumption.

[0068] Multiple air injection points at different heights are preset in the dual-wheel milling air-lift system. These points are specifically positioned at the junction of the two slag discharge pipes based on the overall construction height. Step S2 dynamically activates the corresponding air injection points according to the current construction stage. In the shallow trench stage, the upper air injection point is activated to avoid excessive pressure due to air injection at the lower point; in the deep trench stage, the lower and middle air injection points are activated to ensure sufficient slag discharge power; and in the medium trench stage, the middle air injection point is activated to balance efficiency and energy consumption. Simultaneously, based on the air-lift slag discharge characteristics of each stage, initial air injection pressure and initial air injection volume are set for the activated air injection points. These initial parameters are provided by the construction correction model in step S1 combined with the current basic parameters, ensuring the rationality of the initial parameters and laying the foundation for subsequent precise control.

[0069] S3: During construction, the density parameters of the circulating mud and the slag content parameters at the outlet of the slag discharge pipe are collected in real time. These real-time parameters are input into the construction correction model to calculate the optimal total gas injection pressure of the system under the current working conditions. and optimal total gas injection volume ;

[0070] In this step, during construction, the density parameters and slag content of the circulating mud are collected in real time using mud density sensors and slag content detection equipment. The mud density directly reflects the slag content in the mud; excessive density indicates too much slag, requiring an increase in air injection; insufficient density indicates excessive air injection, requiring a reduction in air injection. The slag content directly reflects the slag removal effect; a low slag content indicates incomplete slag removal, requiring adjustment of air injection parameters; a high slag content may lead to blockage of the slag removal pipe, requiring optimization of the air injection strategy. These two types of parameters reflect the current slag removal effect and changes in operating conditions in real time, and are used to dynamically adjust the air injection pressure and volume.

[0071] When calculating the optimal total gas injection parameters, the collected real-time parameters are input into the construction correction model in step S1. The model combines the initial input basic parameters, the current construction stage, and the online updated model parameters to calculate the optimal total gas injection pressure of the system under the current working conditions through an algorithm. and optimal total gas injection volume The core logic of the calculation is to balance the slag removal efficiency with energy consumption. The mud column pressure at the current trench depth must be met to ensure sufficient air lift power; It is necessary to match the current amount of sludge to ensure thorough sludge removal without wasting energy, while avoiding mud splashing due to excessive air injection or sludge accumulation due to insufficient air injection.

[0072] S4: Based on the height difference of each injection point and the current air lift efficiency coefficient (calculated by the construction correction model based on real-time mud viscosity and gas density), the optimal total injection pressure is... and optimal total gas injection volume The gas is allocated to each injection point, and the target pressure and target flow rate of each injection point are obtained.

[0073] In this step, the allocation is based on the height difference of each injection point. The pressure inside the tank is different at injection points at different heights. The pressure at the lower points is greater than that at the upper points, so a higher injection pressure needs to be allocated to achieve effective air lift. The current air lift efficiency coefficient is calculated by the construction correction model based on real-time working condition data such as mud viscosity and gas density, and reflects the air lift efficiency of each injection point.

[0074] A differentiated allocation logic is employed, combining height difference and gas lift efficiency coefficient, to establish an allocation algorithm that optimizes the total injection pressure. and optimal total gas injection volume The gas is allocated to each activated injection point to obtain the target pressure and target flow rate for each point.

[0075] For example, the lower gas injection points, which are opened during the deep trench stage, have higher target pressures than the upper points due to their higher pressures. Points with high gas lift efficiency coefficients can be allocated more gas injection volume to improve overall gas lift efficiency. During the allocation process, it is ensured that the sum of the target parameters for each point is consistent with the optimal total gas injection parameters to avoid wasting or underestimating the total parameters.

[0076] S5: Real-time acquisition of actual pressure and flow values ​​at each injection point, comparison with the corresponding target values, and dynamic adjustment of the valve opening at each injection point through a PID controller to ensure that the actual parameters track the target parameters;

[0077] In this step, the actual pressure and flow rates at each injection point are collected in real time using pressure and flow sensors. The collection frequency is matched with the construction conditions to ensure real-time performance and timely capture of parameter fluctuations.

[0078] The collected actual parameters are compared in real time with the target pressure and target flow values ​​obtained in step S4, and the deviation between the two is calculated. If the actual pressure is lower than the target pressure or the actual flow is higher than the target flow, the deviation signal is input to the PID controller, proportional, integral and derivative controller. The PID controller automatically calculates the adjustment amount according to the magnitude and rate of change of the deviation, and dynamically adjusts the opening of the regulating valve at each gas injection point. The larger the deviation, the larger the adjustment range, ensuring that the actual parameters quickly approach the target parameters. At the same time, the adjusted actual parameters are collected and compared again to form a closed-loop feedback, continuously optimizing the adjustment effect until the deviation between the actual parameters and the target parameters is within the preset allowable range.

[0079] S6: Pre-establish a standard characteristic curve library of gas injection pressure-slag discharge flow rate under different working conditions, collect the current gas injection pressure and slag discharge flow rate data in real time and generate real-time characteristic curves, compare the real-time characteristic curves with the standard characteristic curves under the corresponding working conditions; when the deviation between the two exceeds the preset threshold, it is determined to be a precursor to slag discharge pipe blockage and automatically triggers graded anti-blockage measures.

[0080] In this step, a library of standard characteristic curves for air injection pressure and slag discharge flow rate is established beforehand, based on extensive experiments and historical construction data, for different working conditions including varying geological conditions, milling depths, and mud properties. The logic of the curves is: under normal working conditions, air injection pressure and slag discharge flow rate show a stable positive correlation; as air injection pressure increases, slag discharge flow rate increases accordingly; as air injection pressure decreases, slag discharge flow rate decreases accordingly, with a stable curve trend and minimal fluctuations. Different standard curves correspond to different working conditions to ensure the accuracy of the comparison.

[0081] During construction, real-time data on current gas injection pressure and slag discharge flow rate are collected, and a real-time characteristic curve is generated based on the collected data. The real-time characteristic curve is compared with the standard characteristic curve corresponding to the current working condition, and the deviation between the two is calculated. For example, in the real-time curve, under the same gas injection pressure, the slag discharge flow rate is significantly lower than that of the standard curve, or the curve fluctuation amplitude is much greater than that of the standard curve. When the deviation exceeds a preset threshold, it is judged as a precursor to slag discharge pipe blockage. At this time, a small amount of sediment has accumulated in the slag discharge pipe, but it has not yet formed a complete blockage. Subsequently, graded anti-blockage measures are automatically triggered, such as: Level 1 anti-blockage: increase the gas injection volume at the corresponding gas injection point to flush the sediment in the pipe; Level 2 anti-blockage: adjust the gas injection pressure distribution to enhance the flushing power; Level 3 anti-blockage: suspend milling and start high-pressure flushing to thoroughly remove sediment. The anti-blockage measures are gradually upgraded according to the severity of the precursor to blockage to prevent the blockage from expanding.

[0082] Furthermore, in step S2, the criteria for dividing the construction stages are as follows:

[0083] When the milling depth H≤H1, it is the shallow trench stage, and only the bottom first air injection point is opened. In this stage, the mud column height in the trench is low, the pressure in the trench is low, and the power required for air lift slag removal is relatively small. Only the bottom first air injection point needs to be opened to meet the slag removal requirements. The bottom first air injection point can directly act on the slag area at the bottom of the trench to achieve efficient disturbance and transportation of slag. There is no need to open the upper air injection point to avoid power redundancy.

[0084] When the milling depth H1 < H ≤ H2, it is the middle tank stage, and both the first and second air injection points are activated simultaneously. At this stage, the mud column height within the tank is moderate, and the pressure inside the tank is higher than in the shallow tank stage. Activating only the first air injection point is insufficient to meet the requirements for efficient slag removal throughout the process; therefore, both the first and second air injection points must be activated simultaneously, with the second air injection point higher than the first. The two air injection points work synergistically: the first air injection point provides bottom power, disturbing the sediment at the bottom of the tank, while the second air injection point supplements the power in the middle, accelerating the upward speed of the mud and sediment, thus balancing slag removal efficiency and energy consumption.

[0085] When the milling depth H > H2, it is the deep trench stage, and the first, second, and third air injection points are activated simultaneously. During this stage, the mud column height in the trench is the highest, the pressure is greatest, and the power required for air-lift slag removal is strongest. Therefore, the first, second, and third air injection points must be activated simultaneously, with the third injection point being higher than the second. The three injection points provide power in a tiered manner from top to bottom, forming a stepped air-lift effect. The first injection point is responsible for disturbing the sediment at the bottom of the trench, the second injection point assists in lifting, and the third injection point further accelerates the rise of the mud and sediment, ensuring sufficient power and thorough slag removal in the deep trench environment.

[0086] H1 and H2 are depth thresholds preset based on geological conditions and mud properties.

[0087] The depth thresholds H1 and H2 are not fixed values. They need to be calibrated in advance based on the geological conditions and mud properties of the construction site through experiments and historical construction data to ensure that the classification criteria are accurately matched with the actual construction conditions.

[0088] The aforementioned dual-wheel milling air-lift system has three pre-set air injection points at different heights, sequentially designated as the first, second, and third air injection points along the slag discharge pipe axis from top to bottom. Step S2 strictly follows the above-mentioned division criteria, dynamically activating the corresponding air injection points based on the current construction stage to ensure precise matching between the number and location of the activated air injection points and the milling depth. Simultaneously, considering the air-lift slag discharge characteristics at each stage, initial air injection pressure and initial air injection volume are set for the activated air injection points. These initial parameters are provided by the construction correction model in step S1, combined with current basic parameters and depth threshold setting logic. Within the same stage, the initial air injection pressure at the lower air injection point (first air injection point) is higher than that at the upper air injection points (second and third air injection points), adapting to the pressure differences within the trench at different heights and ensuring the rationality of the initial parameters, laying the foundation for subsequent precise control.

[0089] Furthermore, in step S3, the construction correction model adopts an improved BP neural network model, whose input layer nodes include geological parameters, mud density, slag content, milling depth and ambient temperature, and output layer nodes include optimal total gas injection pressure and optimal total gas injection volume.

[0090] Furthermore, in step S4, the formula for allocating the gas injection parameters is:

[0091] ;

[0092] ;

[0093] in, Let i be the target pressure value for the i-th injection point. For the target flow value, H represents the height of the i-th air injection point from the ground, and H represents the current milling groove depth. The set minimum safe immersion depth correction value; and These are the pressure distribution coefficient and flow distribution coefficient for the i-th gas injection point, respectively.

[0094] It should be noted that, Let be the target pressure value for the i-th injection point, which is the final injection pressure standard that this point needs to achieve. The optimal total gas injection pressure of the system calculated in step S3 is the core benchmark for parameter allocation; H is the current milling depth, which is the current construction depth obtained in real time in step S2. The height of the i-th injection point from the ground is a preset fixed parameter, determined according to the installation location of the injection point. From top to bottom, these are the first, second, and third injection points. The values ​​increase sequentially; The minimum safe immersion depth correction value is set to correct the distribution deviation in shallow trench stages or special working conditions, and to avoid air lift failure due to insufficient immersion depth. Its value is preset according to geological conditions and mud properties to ensure the safety and rationality of distribution. The pressure distribution coefficient for the i-th injection point is calculated in real time by the construction correction model based on the current working conditions, reflecting the pressure adaptation requirements at that point. This coefficient is also applied to the lower injection points. The values ​​are usually higher than those at the upper points, which is to match the characteristic that the pressure in the tank increases from top to bottom.

[0095] Specifically, the above This is a depth correction coefficient, which is positively correlated with the actual working depth of the injection point. The closer the injection point is to the bottom of the tank, the larger the coefficient, and the higher the pressure percentage allocated to it, which meets the needs of lower injection points where the tank pressure is high and requires higher injection pressure; then, through the pressure distribution coefficient... Further refinements were made, and pressure distribution was optimized based on real-time operating conditions to ensure... It satisfies the airlift power requirements at this location while avoiding pressure redundancy.

[0096] In the target flow value allocation formula The target flow rate value for the i-th gas injection point is the gas injection flow rate standard that the point ultimately needs to achieve. The optimal total gas injection rate calculated in step S3 is the core benchmark for flow distribution; H, , The meaning is consistent with the pressure distribution formula, ensuring the consistency of the deep correction logic; The flow distribution coefficient for the i-th injection point is also calculated in real time by the construction correction model based on real-time operating conditions, and is consistent with... Synergistic effect, reflecting the flow adaptation requirements at this location, is usually positively correlated with the air lift efficiency coefficient; the higher the air lift efficiency coefficient, the better. The higher the value, the higher the proportion of flow allocated, thus improving the overall airlift efficiency.

[0097] The flow rate allocation formula and the pressure allocation formula use the same depth correction coefficient to ensure the synergy between flow rate and pressure allocation, and to avoid a decrease in air lift efficiency due to pressure and flow rate mismatch; through the flow rate allocation coefficient... It adapts to the differences in air lift efficiency at various locations, making flow distribution more targeted, while ensuring that all air injection points are activated. The sum equals , The sum equals This avoids wasting or underutilizing total parameters and enables precise allocation and efficient utilization of total injection parameters.

[0098] Furthermore, in step S6, the graded prevention and blocking measures include:

[0099] Based on the deviation between the real-time characteristic curve and the standard characteristic curve under the corresponding working condition, three warning levels are clearly defined. The deviation range is set based on extensive testing and calibration using historical construction data to ensure the accuracy and rationality of the warnings. The specific classification criteria are as follows:

[0100] Level 1 warning: When the deviation is between 10% and 20%, the current gas injection pressure will be automatically increased by 5% to 10%, and the gas injection volume will be increased by 10% to 15%. The system will be observed for 30 seconds. This corresponds to the early signs of a minor blockage. At this time, there is a small amount of sediment accumulation in the slag discharge pipe. It does not affect the normal slag discharge process, but there is an abnormal fluctuation in the flow rate. There is no need to stop construction. The potential danger can be resolved by slightly adjusting the gas injection parameters.

[0101] The system automatically increases the current injection pressure by 5%-10% and the injection volume by 10%-15%, observing for 30 seconds. By slightly increasing the injection pressure and volume, the airflow velocity and impact force inside the pipe are increased to flush away a small amount of accumulated sediment. The system is observed for 30 seconds to monitor deviation changes in real time. If the deviation drops below 10% after 30 seconds, the system automatically restores the original injection parameters. If the deviation does not decrease or continues to increase, the system automatically escalates to a level two warning and executes corresponding anti-blocking measures.

[0102] Level 2 warning: When the deviation is between 20% and 30%, turn off the milling head rotation, maintain the air-lift slag discharge state, and alternately open and close each air injection point to perform pulse clearing; this corresponds to the precursor of moderate blockage. At this time, the amount of slag accumulation in the slag discharge pipe increases, which has affected the slag discharge efficiency. Simply adjusting the air injection parameters is not enough to completely remove the slag. It is necessary to disturb the slag in the pipe by pulse clearing to avoid further accumulation.

[0103] With the milling head rotating off, maintain the air-lift slag removal state while alternately opening and closing each air injection point for pulsed clearing. Closing the milling head rotation prevents further cutting of the formation and the generation of new sediment, reducing the burden on the slag removal pipe; maintaining the air-lift slag removal state ensures continuous mud flow inside the pipe, preventing further sediment deposition; alternately opening and closing each air injection point creates a pulsed airflow, utilizing the instantaneous impact of the airflow to disrupt the sediment accumulation balance inside the pipe, disturbing the sediment and discharging it with the mud, achieving precise clearing and avoiding excessive intervention that could affect the construction progress.

[0104] Level 3 Warning: When the deviation exceeds 30%, construction should be stopped, the milling head raised to a safe height, and an audible and visual alarm triggered to prompt manual intervention. This corresponds to a severe blockage precursor. At this point, the slag accumulation in the slag discharge pipe is close to a blockage. Continuing construction may lead to complete blockage of the slag discharge pipe, causing equipment failure or tank wall collapse. Construction must be stopped immediately, and the manual intervention procedure initiated. Stopping construction completely prevents the continuous generation and accumulation of slag, preventing complete blockage of the slag discharge pipe; raising the milling head to a safe height prevents the milling head from being jammed by the slag and also prevents tank wall collapse from damaging the milling head equipment; the audible and visual alarm quickly alerts on-site operators to conduct timely manual inspection and clearing of the blockage, preventing the fault from escalating.

[0105] Furthermore, it also includes step S7: real-time monitoring of the working status of each sensor, and when a sensor fault is detected, automatically switching to the experience control mode based on historical data and issuing a fault alarm.

[0106] In this step, the monitoring scope covers all sensors in the entire gas lift system, including geological parameter acquisition sensors, mud performance sensors, temperature sensors, pressure sensors at each gas injection point, flow sensors at each gas injection point, slag content detection sensors, and slag discharge flow sensors.

[0107] The monitoring logic is as follows: real-time acquisition of the output signals of each sensor, and determination of whether the sensor is faulty through signal integrity detection, signal stability analysis, threshold judgment, etc. The monitoring frequency is synchronized with the acquisition frequency of each sensor to ensure timely detection of faults.

[0108] When any one or more sensors malfunction, the system automatically triggers an emergency response mechanism, performing two core operations:

[0109] The system calls upon historical construction data stored in the construction correction model, filters out the historical data most similar to the current construction conditions, extracts the gas injection parameters under those historical conditions, and uses them as the parameter benchmark for the current emergency control. Following the original segmented gas delivery logic, parameter allocation formula, and closed-loop control logic, the system continues to achieve stable operation of the gas lift system, avoiding system shutdown due to sensor failure.

[0110] The system issues a fault warning via an audible and visual alarm device, and simultaneously displays the specific location of the faulty sensor on the control system interface, reminding operators to promptly check, repair, or replace the faulty sensor to ensure that sensor faults are dealt with quickly and the system's normal optimized control mode is restored as soon as possible.

[0111] It should be noted that the experience control mode is only temporarily activated during sensor malfunctions. Once the sensor malfunction is resolved, the system automatically stops the experience control mode and reverts to the original real-time data acquisition, model calculation, and precise control mode to ensure that construction quality and efficiency are not affected in the long term.

[0112] Furthermore, it also includes gas source pressure monitoring and emergency response procedures:

[0113] The outlet pressure of the compressed air source is collected in real time. When the air source pressure is lower than the preset lower limit, the uppermost air injection point is automatically closed and the remaining air volume is concentrated and distributed to the lower air injection points.

[0114] At the same time, reduce the milling head feed speed and issue an alarm for insufficient air supply pressure.

[0115] A dedicated pressure sensor is installed at the outlet of the compressed air source. This sensor is linked with the system control system in real time and collects the outlet pressure data of the air source at a preset frequency. At the same time, the system presets a lower limit value for the air source pressure. This lower limit value is based on the minimum power requirements for air-lift slag removal, the current milling depth, the number of air injection points opened, and other working conditions. It is calibrated and set in combination with historical construction data to ensure that the lower limit value meets the minimum power requirements for slag removal while avoiding setting it too low, which would cause air-lift failure.

[0116] When the air source outlet pressure collected by the sensor is lower than the preset lower limit, the system automatically triggers the emergency response mechanism. Combined with the working condition of dual-wheel milling air lift slag removal, the main function of the uppermost air injection point (such as the third air injection point) is to assist in lifting mud and sludge, and its contribution to the disturbance of bottom sludge and slag removal power is small. Closing this point can reduce air injection loss and concentrate the limited compressed air to the lower air injection points (the first and second air injection points), ensuring the air injection pressure and flow rate of the core air injection points below, ensuring the normal disturbance and discharge of sludge at the bottom of the tank, and avoiding slag removal interruption due to insufficient air source.

[0117] Then, the milling head feed rate directly determines the amount of formation cutting and the amount of slag generated. Reducing the feed rate can reduce the amount of slag generated per unit time, thereby reducing the load of air lift slag removal, so that the limited air source pressure can meet the current slag removal needs, and avoid further aggravating the problem of insufficient air source pressure due to excessive slag and excessive slag removal load, thus achieving a balance between load reduction and slag removal.

[0118] Finally, a clear alarm is issued through the audible and visual alarm device, and the insufficient air supply pressure and the current actual pressure value are displayed on the control system interface. This reminds on-site operators to check the air supply equipment in a timely manner, quickly troubleshoot and resolve the problem of insufficient air supply pressure, restore the normal air supply pressure as soon as possible, and ensure that the air lift system returns to the normal optimized control mode.

[0119] Furthermore, it also includes a leak detection procedure for the gas injection line:

[0120] The flow deviation rate of each gas injection point is calculated in real time. When the deviation rate between the actual flow rate and the target flow rate of a certain gas injection point exceeds 30% and lasts for more than 10 seconds, it is determined that the gas injection pipeline has leaked.

[0121] The regulating valve at the leaking gas injection point will be automatically closed, the corresponding gas injection parameters will be redistributed to other open gas injection points, and a leak alarm will be issued.

[0122] The system is based on the target flow rate values ​​allocated at each gas injection point in step S4. And the actual flow rate values ​​of each gas injection point collected in real time in step S5. The flow deviation rate is calculated point by point and at high frequency according to the following formula:

[0123]

[0124] in, Let be the flow rate deviation rate at the i-th injection point; The target flow rate value for the i-th injection point is calculated in step S4; The actual flow rate value of the i-th injection point is collected in real time by the flow sensor in step S5.

[0125] The calculation frequency is synchronized with the flow sensor's acquisition frequency to ensure real-time capture of abnormal flow changes. Using a relative deviation rate rather than an absolute flow deviation as the criterion allows for adaptation to different construction stages and varying target flow rates at different injection points, avoiding misjudgments caused by fluctuations in the target flow rate itself.

[0126] When the flow rate deviation rate at a certain gas injection point If the flow rate exceeds 30% and this abnormal state persists for more than 10 seconds, the air injection pipeline is considered to be leaking. Based on extensive engineering tests and historical data calibration, the flow fluctuation deviation rate caused by PID controller adjustment under normal operating conditions is usually no more than 15%. The 20%-30% range is the abnormal fluctuation warning zone. If it exceeds 30%, normal adjustment interference can be ruled out, and the pipeline leak can be confirmed. A leak will cause the actual flow rate to be significantly higher than the target flow rate, as the leak point consumes additional compressed air.

[0127] When a leak is detected in a gas injection line, the system automatically and simultaneously executes three core operations, forming a complete emergency closed loop:

[0128] Automatically shuts off the regulating valve at the leak point: Immediately sends a shut-off command to the electric regulating valve corresponding to the leak point, cutting off the compressed air supply at that point to prevent continuous compressed air leakage from causing waste of air source and a sudden drop in total injection pressure.

[0129] The original target pressure value assigned to the leak point and target traffic value Following the quantitative allocation formula in step S4, the gas injection points are redistributed to all other currently active injection points. A leak warning is issued via an audible and visual alarm, and the specific location of the leaking injection point and the estimated current leak flow rate are clearly displayed on the control system interface, reminding operators to perform timely repairs.

[0130] Furthermore, when the flow rate deviation rate at a certain injection point... If the abnormal state persists for more than 10 seconds, the gas injection line is considered to be leaking. The 10-second duration is set to filter out interference signals such as instantaneous sensor fluctuations and instantaneous fluctuations in gas source pressure, avoiding false triggering of emergency measures and ensuring the accuracy of the judgment result.

[0131] Furthermore, the pre-training steps of the construction correction model are as follows:

[0132] Collect historical construction data from multiple different construction sites and under different geological conditions to construct a global construction dataset;

[0133] A pre-trained basic model based on a global construction dataset;

[0134] At the new construction site, the basic model is fine-tuned using the construction data from the previous N meters of the site through transfer learning, resulting in a modified construction model adapted to the new location.

[0135] The aforementioned global construction dataset supports the pre-training of the basic model. Its construction goal is to cover as many construction scenarios as possible, enabling the model to learn the general mapping relationship of air-lift slag removal. The specific construction process is as follows:

[0136] We collected historical construction data for twin-wheel milling machines from multiple construction sites across the country and even globally, under different geological conditions, construction depths, and equipment models, ensuring the diversity and representativeness of the dataset. Data sources included construction logs from completed projects, equipment operation records, and sensor data.

[0137] Each data sample contains a complete input-output correspondence:

[0138] Input features: geological parameters such as formation hardness, particle size distribution, and rock layer distribution; mud performance parameters such as density, viscosity, and sand content; ambient temperature parameters; milling depth H; and number of air injection points opened.

[0139] Output label: Optimal total injection pressure under the corresponding operating condition Optimal total gas injection volume Pressure distribution coefficient at each injection point and flow distribution coefficient .

[0140] Finally, the raw data is cleaned, normalized, and labeled to construct a standardized global construction dataset.

[0141] Furthermore, the depth thresholds H1 and H2 are adaptively adjusted according to geological conditions:

[0142] The sediment load is determined by the formation characteristics, which in turn determines the gas injection demand, which in turn determines the threshold setting. By quantifying the differences in sediment characteristics among different formations, depth thresholds are set specifically for each formation. Furthermore, the number of gas injection points and the grading logic are expanded for ultra-deep working conditions to ensure precise matching between gas injection power and sediment discharge load. The specific principle is as follows:

[0143] When the milling depth H≤60m: This is considered a conventional depth for construction operations with a milling depth ≤60m. In conventional depth construction, the efficiency of air-lift slag removal is mainly affected by the characteristics of the sediment generated by the cutting of the strata. The size, density, and suspension difficulty of sediment particles in different strata vary significantly, so differentiated timing of gas injection point activation is required.

[0144] If the construction formation is a soft soil formation, H1 = 15m and H2 = 30m; the soft soil sediment is easy to suspend, and only the lowest first gas injection point within 15m depth needs to be opened to meet the slag discharge requirement; two points are opened within 15 - 30m to balance efficiency and energy consumption; three points are opened above 30m to ensure slag discharge in the deep groove.

[0145] If the construction formation is a sandy pebble formation, H1 = 10m and H2 = 25m; the sandy pebble sediment has large particles and is difficult to suspend, and two gas injection points need to be opened at 10m depth to provide sufficient lifting force; three points are opened at 25m depth to prevent large particle sediment from depositing in the slag discharge pipe.

[0146] If the construction formation is a hard rock formation, H1 = 8m and H2 = 20m; the hard rock sediment has the highest density and is the easiest to deposit, and two points need to be opened at 8m depth to enhance bottom disturbance; three points are opened at 20m depth to ensure that hard rock slag can be smoothly lifted to the ground.

[0147] The system automatically identifies the current construction formation type based on the geological parameters obtained in step S1, without manual intervention, and automatically loads the depth thresholds H1 and H2 corresponding to the formation; when the formation alternates during construction, the system updates the formation type in real time and dynamically adjusts the thresholds to ensure that the opening time of the gas injection points is accurately matched with the current formation.

[0148] When the milling groove depth is 200m < H ≤ 300m: When the construction depth exceeds 200m, the mud column pressure in the groove increases sharply, and the effective action distance of the air lift of the three gas injection points cannot cover the entire slag discharge pipe, resulting in insufficient slag discharge power and a sharp drop in efficiency. This solution constructs a stepped air lift system by increasing the number of gas injection points and hierarchical thresholds to solve the problem of insufficient power in ultra - deep working conditions. The specific principle is as follows:

[0149] Four gas injection points are arranged axially along the slag discharge pipe, adding one point compared to the conventional depth, forming a four - level air lift effect of bottom disturbance, middle lifting, and upper acceleration.

[0150] It is set to 35 - 40m. This spacing is calibrated based on the effective action distance of air - lift slag discharge. After the gas - liquid mixed flow generated by air lift rises 35 - 40m, the bubbles will gradually break and dissipate, and the air - lift efficiency will decrease significantly; by setting a gas injection point every 35 - 40m, new bubbles can be continuously replenished to maintain stable air - lift power.

[0151] Four gas injection points are arranged axially along the slag discharge pipe, and the spacing between the gas injection points is 35 - 40m;

[0152] The depth thresholds are successively 、 、H3, where is the depth from single - point to two - point, H3 represents the depth conversion from two points to three points, and H4 represents the depth conversion from three points to four points.

[0153] The benchmark value for sand and gravel strata is: =40m, =80m, H3=120m; The gravel strata have a large sediment load, so more gas injection points need to be opened earlier. Two points are opened at 40m, three points at 80m, and four points at 120m to ensure sufficient power for slag removal throughout the entire depth range.

[0154] The benchmark value for hard rock strata is: =30m, =65m, H3=100m. Hard rock strata sediments are more difficult to lift, so the threshold is lower than that of sand and gravel strata. Two points are opened at 30m, three points are opened at 65m, and four points are opened at 100m, so as to build a complete stepped airlift system in advance.

[0155] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A segmented air delivery optimization control method for a dual-wheel milling air lift system, characterized in that, Includes the following steps: S1: Obtain the geological parameters, mud performance parameters, and ambient temperature parameters of the construction location, and input the above parameters into the pre-constructed construction correction model; the construction correction model is trained based on historical construction data, and the parameters are updated online every 5-10 minutes according to the collected real-time construction data, and the update is triggered immediately when the geological parameters change abruptly. S2: Based on the current milling depth H, the construction process is divided into shallow trench stage, medium trench stage and deep trench stage. According to the air lift slag discharge characteristics of different construction stages, the corresponding air injection points arranged in advance along the slag discharge pipe axis are dynamically opened, and the initial air injection pressure and initial air injection volume are set for each opened air injection point. S3: During construction, the density parameters of the circulating mud and the slag content parameters at the outlet of the slag discharge pipe are collected in real time. These real-time parameters are input into the construction correction model to calculate the optimal total gas injection pressure of the system under the current working conditions. and optimal total gas injection volume ; S4: Based on the height difference of each injection point and the current air lift efficiency coefficient (calculated by the construction correction model based on real-time mud viscosity and gas density), the optimal total injection pressure is... and optimal total gas injection volume The gas is allocated to each injection point, and the target pressure and target flow rate of each injection point are obtained. S5: Real-time acquisition of actual pressure and flow values ​​at each injection point, comparison with the corresponding target values, and dynamic adjustment of the valve opening at each injection point through a PID controller to ensure that the actual parameters track the target parameters; S6: Pre-establish a standard characteristic curve library of gas injection pressure-slag discharge flow rate under different working conditions, collect the current gas injection pressure and slag discharge flow rate data in real time and generate real-time characteristic curves, compare the real-time characteristic curves with the standard characteristic curves under the corresponding working conditions; when the deviation between the two exceeds the preset threshold, it is determined to be a precursor to slag discharge pipe blockage and triggers graded anti-blockage measures.

2. The segmented air supply optimization control method for the dual-wheel milling air lift system according to claim 1, characterized in that, In step S2, the criteria for dividing the construction stages are as follows: When the milling groove depth H≤H1, it is the shallow groove stage, and only the bottom first air injection point is opened; When the milling groove depth H1<H≤H2, it is the middle groove stage, and the first and second air injection points are opened at the same time. When the milling groove depth H > H2, it is the deep groove stage, and the first, second, and third air injection points are opened simultaneously. H1 and H2 are depth thresholds preset based on geological conditions and mud properties.

3. The segmented air delivery optimization control method for the dual-wheel milling air lift system according to claim 1, characterized in that, In step S3, the construction correction model adopts an improved BP neural network model. Its input layer nodes include geological parameters, mud density, slag content, milling depth and ambient temperature, and its output layer nodes include optimal total gas injection pressure and optimal total gas injection volume.

4. The segmented air supply optimization control method for the dual-wheel milling air lift system according to claim 1, characterized in that, In step S4, the formula for allocating the gas injection parameters is: ； ； in, Let i be the target pressure value for the i-th injection point. For the target traffic value, H represents the height of the i-th air injection point from the ground, and H represents the current milling groove depth. The set minimum safe immersion depth correction value; and These are the pressure distribution coefficient and flow distribution coefficient for the i-th gas injection point, respectively.

5. The segmented air supply optimization control method for the dual-wheel milling air lift system according to claim 1, characterized in that, In step S6, the graded anti-blocking measures include: Level 1 warning: When the deviation is between 10% and 20%, automatically increase the current gas injection pressure by 5% to 10% and increase the gas injection volume by 10% to 15%, and continue to observe for 30 seconds; Level 2 warning: When the deviation is between 20% and 30%, turn off the milling head rotation, maintain the air lift slag removal state, and alternately open and close each air injection point to perform pulse clearing. Level 3 warning: When the deviation exceeds 30%, construction is stopped, the milling head is raised to a safe height, and an audible and visual alarm is issued to prompt manual intervention.

6. The segmented air supply optimization control method for the dual-wheel milling air lift system according to claim 1, characterized in that, It also includes step S7: real-time monitoring of the working status of each sensor, and when a sensor fault is detected, automatically switching to the experience control mode based on historical data and issuing a fault alarm.

7. The segmented air supply optimization control method for the dual-wheel milling air lift system according to claim 1, characterized in that, It also includes gas source pressure monitoring and emergency response procedures: The outlet pressure of the compressed air source is collected in real time. When the air source pressure is lower than the preset lower limit, the uppermost air injection point is automatically closed and the remaining air volume is concentrated and distributed to the lower air injection points. Meanwhile, reduce the feeding speed of the milling head and issue an alarm for insufficient air source pressure.

8. The segmented air supply optimization control method for the dual-wheel milling air lift system according to claim 1, characterized in that, It also includes the detection step of the injection pipeline leakage: Calculate the flow deviation rate of each injection point in real time. When the deviation rate between the actual flow and the target flow of a certain injection point exceeds 30% and lasts for more than 10 seconds, it is determined that the injection pipeline leaks; Automatically close the regulating valve of the leaking injection point, redistribute its corresponding injection parameters to other open injection points, and issue a leakage alarm.

9. The segmented air delivery optimization control method for the dual-wheel milling air lift system according to claim 1, characterized in that, The pre-training step of the construction correction model is as follows: Collect historical construction data from multiple different construction sites and different geological conditions to construct a global construction data set; Pre-train the basic model based on the global construction data set; At a new construction site, use the construction data of the first N meters at this site to fine-tune the basic model through transfer learning to obtain a construction correction model adapted to this site.

10. The segmented air delivery optimization control method for the dual-wheel milling air lift system according to claim 1, characterized in that, The depth thresholds H1 and H2 are adaptively adjusted according to geological conditions: When the milling groove depth H ≤ 60m: If the construction formation is a soft soil formation, H1 = 15m, H2 = 30m; If the construction formation is a sandy cobble formation, H1 = 10m, H2 = 25m; If the construction formation is a hard rock formation, H1 = 8m, H2 = 20m; When 200m < H ≤ 300m of the milling groove depth: Arrange 4 injection points along the axial direction of the slag discharge pipe, and the spacing between the injection points is 35 - 40m; The depth thresholds are as follows: , H3, among which For single-point to two-point depth conversion, H3 represents the depth conversion from two points to three points, and H4 represents the depth conversion from three points to four points. The benchmark value for sand and gravel strata is: =40m, =80m, H3=120m; The benchmark value for hard rock strata is: =30m, =65m, H3=100m.

Images