Vertical control method for ultra-deep foundation pit slot milling machine

Abstract

The invention provides a vertical control method for an ultra-deep foundation pit slot milling machine, and belongs to the technical field of vertical control of slot milling machines. During the working period of the ultra-deep foundation pit slot milling machine, a control system of the ultra-deep foundation pit slot milling machine collects basic parameters and data; the control system of the ultra-deep foundation pit slot milling machine constructs a deviation risk prediction model; the control system of the ultra-deep foundation pit slot milling machine implements a real-time self-adaptive adjustment mechanism; and the control system of the ultra-deep foundation pit slot milling machine performs prediction model self-updating and feedback optimization. By means of the method, the detection frequency can be dynamically adjusted according to the soft soil characteristics, the construction state and the deviation trend, automatic encryption is achieved in deviation sensitive intervals such as 20-30 m, the number of invalid detection times is reduced by 30% compared with fixed frequency detection, and meanwhile the deviation omission ratio is reduced to 5% or below.

Description

TECHNICAL FIELD

[0001] The present application belongs to the vertical control technical field of slot milling machine, and particularly relates to a vertical control method for a slot milling machine for super deep foundation pit. BACKGROUND

[0002] The slot milling machine for super deep foundation pit is an advanced device specially designed for underground continuous wall construction, mainly used for excavating support structure slots of deep foundation pit, providing reliable waterproof, soil retaining and bearing functions. As mentioned in the prior art with patent publication number CN116145645A, the verticality of the super deep continuous wall is often controlled.

[0003] However, the existing method does not clearly specify the specific frequency of stage depth detection (such as detecting once every 5m or once every 10m) during the control of the verticality of the super deep continuous wall (the super deep continuous wall is the super deep foundation pit). In the super deep soft soil layer of the super deep foundation pit, if the detection interval is too large (such as more than 10m), the critical interval of rapid deviation accumulation (such as the soft plastic layer at a depth of 20-30m) may be missed. SUMMARY

[0004] To solve the defects in the prior art, the present application provides a vertical control method for a slot milling machine for super deep foundation pit, which aims to overcome the problem of verticality control caused by unclear detection frequency during the slotting stage of the super deep continuous wall in soft soil layer.

[0005] The present application uses the following technical solutions.

[0006] A vertical control method for a slot milling machine for super deep foundation pit, comprising:

[0007] Setting a dynamic adaptive detection frequency for the slot milling machine for super deep foundation pit, that is, the control system of the slot milling machine for super deep foundation pit uses a machine learning algorithm to establish a prediction model based on the previous geological survey data and the detected data during the slotting process of the slot milling machine for super deep foundation pit; during the slotting construction of the slot milling machine for super deep foundation pit, the prediction model analyzes the data in real time, and if it is judged that the verticality deviation may change rapidly at a certain depth interval, the detection frequency is automatically encrypted to one detection per 2m; otherwise, if the slotting state is stable, it is appropriately relaxed to one detection per 10m.

[0008] Further, the method of setting a dynamic adaptive detection frequency for the slot milling machine for super deep foundation pit specifically comprises:

[0009] Step 1: During the operation of the slot milling machine for super deep foundation pit, the control system of the slot milling machine for super deep foundation pit collects basic parameters and data;

[0010] Step 2: The control system of the slot milling machine for super deep foundation pit constructs a deviation risk prediction model;

[0011] Step 3: The control system of the super-deep foundation trenching machine implements a real-time adaptive adjustment mechanism.

[0012] Step 4: The control system of the super-deep foundation trenching machine performs predictive model self-updating and feedback optimization.

[0013] Further, Step 1 specifically includes: the control system of the super-deep foundation trenching machine establishes a database containing three types of key parameters, including:

[0014] Geological parameters as basic parameters: plasticity index I p , water content w, cohesion c, and internal friction angle of the soft soil of the super-deep foundation, and geological stratification data divided by every 5m depth, plasticity index I p , water content w, cohesion c, and internal friction angle are respectively collected in real time by HJ03-STDS-1 disc liquid limit instrument, water content rapid tester, cohesion tester, and friction angle tester connected with the control system and transmitted to the control system, with a real-time collection frequency of 1Hz for HJ03-STDS-1 disc liquid limit instrument, water content rapid tester, cohesion tester, and friction angle tester.

[0015] Construction parameters: bucket volume V l , lifting speed V d , and opening and closing angle α of the grab bucket of the trenching machine, bucket volume V l , lifting speed V d , and opening and closing angle α of the grab bucket are respectively collected in real time by a material disc instrument, a speed sensor one, a speed sensor two, and an angle instrument connected with the control system and transmitted to the control system, with a real-time collection frequency of 1Hz for the material disc instrument, the speed sensor one, the speed sensor two, and the angle instrument.

[0016] Deviation data: verticality deviation growth rate under the same geological conditions in historical projects, and the i-th verticality value of the super-deep foundation detected in the trenching process of the current trenching machine and the corresponding i-th depth value h i of the super-deep foundation, the verticality value and the depth value are synchronously collected by a verticality measuring instrument and a depth measuring instrument connected with the control system and transmitted to the control system.

[0017] Further, Step 1 specifically further includes: setting an initial detection frequency.

[0018] Further, the method of setting the initial detection frequency comprises: adopting a reference frequency in the initial stage of the trenching machine construction; in the shallow soft soil layer of the super deep foundation pit with a depth of 0-10 m, detecting once every 3 m, and 3 m is the detection interval corresponding to the depth; in the medium deep soft soil layer of the super deep foundation pit with a depth of 10-30 m, detecting once every 5 m, and 5 m is the detection interval corresponding to the depth; in the super deep foundation pit with a depth of more than 30 m, detecting once every 5 m temporarily, and 5 m is the detection interval corresponding to the depth; synchronously recording the geological parameters and the construction parameters corresponding to the depth each time to form an initial sample set and display on the display screen connected to the control system.

[0019] Further, step 2 specifically comprises:

[0020] Step 2-1: the control system of the super deep foundation pit trenching machine constructs a deviation growth rate calculation model;

[0021] Step 2-2: the control system of the super deep foundation pit trenching machine trains a prediction model based on machine learning.

[0022] Further, in step 2-1, the method of constructing the deviation growth rate calculation model comprises:

[0023] Define the verticality deviation growth rate As a core index for judging whether the detection frequency needs to be encrypted, the calculation formula of the verticality deviation growth rate is as follows:

[0024]

[0025] The parameters of the calculation formula are as follows: represents the Kth verticality value of the collected super deep foundation pit, represents the K-1th verticality value of the collected super deep foundation pit, h k represents the Kth depth value of the collected super deep foundation pit, h k-1 represents the K-1th depth value of the collected super deep foundation pit, The unit of is % / m.

[0026] Further, step 2-2 specifically comprises:

[0027] The prediction model of is constructed by using a random forest regression algorithm, and the steps are as follows:

[0028] Step 2-2-1, sample set construction: collecting historical data of at least 3 similar super deep foundation pit trenching machine construction projects, extracting not less than 500 groups of samples from each construction project, and each group of samples containing geological parameters, construction parameters and corresponding values;

[0029] ​Step 2-2-2, feature importance ranking: screening by the prediction model training through the random forest regression algorithm The top three feature parameters that have the greatest impact are the core variables for the prediction model input.

[0030] Step 2-2-3, prediction model verification: 10-fold cross-validation is used to ensure that the prediction error of the prediction model for the deviation growth rate is ≤10%.

[0031] Further, step 3 specifically includes:

[0032] Step 3-1: During the construction process of the slot milling machine, the detection data is sent to the prediction model in real time, and the prediction model obtains new detection data every time With , the following steps are executed in real time.

[0033] Step 3-1-1: Calculate the current deviation growth rate , and use the prediction model to predict the deviation growth rate of the next depth interval , which is h k to , where is the detection interval corresponding to the reference frequency of h k .

[0034] Step 3-1-2: According to the deviation growth rate of the next depth interval , determine the actual detection interval of the next depth interval , the determination method is as follows:

[0035] If , it is determined as low risk, and the actual detection interval of the next depth interval is adjusted to 8~10m;

[0036] If , it is determined as low risk, and the actual detection interval of the next depth interval is adjusted to 5~6m;

[0037] If , it is determined as high risk, and the actual detection interval of the next depth interval is encrypted to 2~3m;

[0038] Step 3-1-3: Modify the actual detection interval of the next depth interval based on the geological stratification data, that is, if the prediction model predicts that the next depth interval contains a known deviation sensitive layer, automatically shorten the actual detection interval of the next depth interval by 20%, and the deviation sensitive layer is 20~25m flow plastic layer.

[0039] Step 3-2: When one of the following situations occurs, start immediate detection without the actual detection interval of the current next depth interval Limitations:

[0040] The sudden increase of the lowering speed of the grab by more than 20%, i.e. V d >1.2V d,zvg , V d,zvg is the arithmetic mean of the lowering speed of the grab in the previous 5 acquisitions;

[0041] The pressure fluctuation of the slurry wall of the super-deep foundation pit exceeds 5kPa, and the pressure of the slurry wall of the super-deep foundation pit is collected by the connected pressure sensor of the control system and transmitted to the control system.

[0042] The historical data shows that the depth interval has a sudden mutation of deviation.

[0043] Further, step 4 specifically includes:

[0044] Step 4-1, the control system of the super-deep foundation pit milling machine real-time corrects the prediction model, i.e. after completing 10 times of depth detection, the new deviation growth rate , geological parameters, construction parameters are input into the prediction model, and the online learning algorithm is used to update the prediction model parameters;

[0045] Step 4-2, the control system of the super-deep foundation pit milling machine controls the upper and lower limits of the detection frequency, i.e. to avoid too frequent detection affecting the construction efficiency or too large interval missing the deviation, the constraint conditions of the prediction model are set:

[0046] Minimum detection interval: not less than 2m;

[0047] Maximum detection interval: not more than 10m;

[0048] Daily detection times: not more than 1 / 3 of the total slot depth.

[0049] The beneficial effects of the present application are that, compared with the prior art, the technical effects of the present application are as follows:

[0050] During the operation of the super-deep foundation pit milling machine, the control system of the super-deep foundation pit milling machine collects basic parameters and data; the control system of the super-deep foundation pit milling machine constructs a deviation risk prediction model; the control system of the super-deep foundation pit milling machine implements a real-time self-adaptive adjustment mechanism; and the control system of the super-deep foundation pit milling machine performs prediction model self-updating and feedback optimization. Through the method of the present application, the detection frequency can be dynamically adjusted according to the characteristics of soft soil, the construction state and the deviation trend, and the deviation sensitive interval of 20-30m is automatically encrypted, which reduces the invalid detection times by 30% compared with fixed frequency detection, and reduces the deviation missing rate to below 5%. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 is a flow chart of a vertical control method for a super-deep foundation trenching machine in the present application. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described below in combination with the drawings in the embodiments of the present application. The embodiments expressed in the present application are only a part of the embodiments of the present application, but not all the embodiments. According to the spirit of the present application, other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0053] As shown in Figure 1 , a vertical control method for a super-deep foundation trenching machine, comprising:

[0054] The dynamic adaptive detection frequency of the super-deep foundation trenching machine is set, that is, the control system of the super-deep foundation trenching machine uses a machine learning algorithm to establish a prediction model according to the early geological survey data (such as the plasticity index, water content, etc. of soft soil) and the detected data (verticality, slot width change, etc. obtained by the ultrasonic slot quality detector connected with the control system every 5m as the initial detection interval) in the trenching process of the super-deep foundation trenching machine; during the trenching construction of the super-deep foundation trenching machine, the prediction model analyzes the data in real time, and if it is judged that the verticality deviation may change rapidly in a certain depth interval (such as the soft plastic layer at a depth of 20-30m is found to be prone to rapid deviation accumulation through historical data), the detection frequency is automatically encrypted to 2m per detection; otherwise, if the trenching state is stable, it can be appropriately relaxed to 10m per detection, realizing dynamic and intelligent adjustment of the detection frequency and accurately capturing the deviation accumulation critical interval. The trenching machine is provided with a grab bucket. The control system of the super-deep foundation trenching machine controls the connection of the grab bucket.

[0055] In the preferred but non-limiting embodiments of the present application, the method for setting the dynamic adaptive detection frequency of the super-deep foundation trenching machine specifically comprises:

[0056] Step 1: During the operation of the super-deep foundation trenching machine, the control system of the super-deep foundation trenching machine collects basic parameters and data;

[0057] In the preferred but non-limiting embodiments of the present application, step 1 specifically comprises: the control system of the super-deep foundation trenching machine establishes a database containing 3 types of key parameters to provide a basis for detection frequency adjustment, and the 3 types of key parameters include:

[0058] The geological parameters as the basic parameters: the plasticity index I p , water content w, cohesion c and internal friction angle of the soft soil of the super-deep foundation geological stratification data (e.g., 20-25 m is flow plastic soft soil, 25-30 m is soft plastic soft soil) divided by every 5 m depth, plasticity index I p , water content w, cohesion c, internal friction angle are respectively collected in real time by HJ03-STDS-1 disc liquid limit instrument, water content rapid tester, cohesion tester and friction angle tester connected with the control system and transmitted to the control system, and the real-time collection frequency of HJ03-STDS-1 disc liquid limit instrument, water content rapid tester, cohesion tester and friction angle tester is 1 Hz.

[0059] Construction parameters: bucket volume V of the grab bucket of the trenching machine, lifting speed V of the grab bucket l , lowering speed V of the grab bucket d , opening and closing angle a of the grab bucket, bucket volume V of the grab bucket, lifting speed V of the grab bucket l , lowering speed V of the grab bucket d of the grab bucket are respectively collected in real time by the disc meter, speed sensor one, speed sensor two and angle meter connected with the control system and transmitted to the control system, and the real-time collection frequency of the disc meter, speed sensor one, speed sensor two and angle meter is 1 Hz.

[0060] Deviation data: verticality deviation growth rate under the same geological condition in the history project, the i-th verticality value of the super-deep foundation pit detected in the trenching process of the current trenching machine and the i-th depth value h of the corresponding super-deep foundation pit i The verticality value and the depth value are synchronously collected by the verticality measuring instrument and the depth measuring instrument connected with the control system and transmitted to the control system.

[0061] In the preferred but non-limiting embodiment of the present application, step 1 further specifically comprises: setting an initial detection frequency.

[0062] In the preferred but non-limiting embodiment of the present application, the method of setting the initial detection frequency comprises: using a reference frequency at the initial stage of the trenching machine construction: detecting once every 3 m in the shallow soft soil layer of the super-deep foundation pit with a depth of 0-10 m, 3 m being the detection interval corresponding to the depth; detecting once every 5 m in the medium-deep soft soil layer of the super-deep foundation pit with a depth of 10-30 m, 5 m being the detection interval corresponding to the depth; detecting once every 5 m in the deep soft soil layer of the super-deep foundation pit with a depth of 30 m or more (to be dynamically adjusted by the prediction model subsequently), 5 m being the detection interval corresponding to the depth, and the corresponding depth geological parameters and construction parameters can be recorded synchronously each time to form an initial sample set and be displayed on the display screen connected with the control system.

[0063] Step 2: constructing a deviation risk prediction model for the control system of the super-deep foundation pit trenching machine;

[0064] In the preferred but non-limiting embodiments of the present application, step 2 specifically comprises:

[0065] Step 2-1: Constructing a deviation growth rate calculation model for the control system of the super-deep foundation trenching machine;

[0066] In the preferred but non-limiting embodiments of the present application, in step 2-1, the method for constructing the deviation growth rate calculation model comprises:

[0067] Define the verticality deviation growth rate As the core indicator for determining whether the detection frequency needs to be encrypted, the verticality deviation growth rate The calculation formula is:

[0068]

[0069] Wherein the parameter meanings of the calculation formula are: represents the kth verticality value of the collected super-deep foundation (unit: %), represents the K-1th verticality value of the collected super-deep foundation (unit: %), h k represents the Kth depth value of the collected super-deep foundation (unit: m), h k-1 represents the K-1th depth value of the collected super-deep foundation (unit: m), The unit of is % / m. When , it is determined as a deviation rapid accumulation interval, and the detection frequency needs to be encrypted.

[0070] Step 2-2: The control system of the super-deep foundation trenching machine is trained based on machine learning to predict the model.

[0071] In the preferred but non-limiting embodiments of the present application, step 2-2 specifically comprises:

[0072] A random forest regression algorithm is used to construct the prediction model of The steps are as follows:

[0073] Step 2-2-1, sample set construction: collect historical data of at least 3 similar super-deep foundation trenching machines of construction projects, extract not less than 500 groups of samples from each construction project, and each group of samples contains geological parameters, construction parameters and corresponding values;

[0074] Step 2-2-2, feature importance sorting: the prediction model is trained by a random forest regression algorithm to filter the top three feature parameters that have the greatest impact on , such as plasticity index I p (weight 0.32), and the lowering speed of the grab V d(wt 0.25), water content w (wt 0.18), as core variables inputted into the prediction model;

[0075] Step 2-2-3, prediction model verification: 10-fold cross-validation is adopted to ensure that the prediction error of the prediction model on the deviation growth rate is ≤10%, meeting the engineering precision requirement.

[0076] Step 3: Real-time self-adaptive adjustment mechanism is implemented in the control system of the super-deep foundation pit slot milling machine.

[0077] In the preferred but non-limiting embodiment of the application, step 3 specifically comprises:

[0078] Step 3-1: During the construction of the slot milling machine, the detection data is sent into the prediction model in real time, and the prediction model acquires new detection data every time and h k , the following steps are executed in real time;

[0079] Step 3-1-1: Calculate the current deviation growth rate , and call the prediction model to predict the deviation growth rate of the next depth interval , which is h k to , wherein is the detection interval corresponding to the reference frequency of h k ;

[0080] Step 3-1-2: According to the deviation growth rate of the next depth interval , determine the actual detection interval of the next depth interval , the determination method is specifically as follows:

[0081] If , it is determined as low risk, and the actual detection interval of the next depth interval is adjusted to 8~10m;

[0082] If , it is determined as low risk, and the actual detection interval of the next depth interval is adjusted to 5~6m;

[0083] If , it is determined as high risk, and the actual detection interval of the next depth interval is encrypted to 2~3m;

[0084] Step 3-1-3: The actual detection interval of the next depth interval is corrected in combination with the geological stratification data, that is, if the prediction model predicts that the next depth interval contains a known deviation sensitive layer, the actual detection interval of the next depth interval is automatically adjusted to ​​​​Shorten 20% (e.g. from 5m to 4m), the deviation sensitive layer is 20~25m flow plastic layer;

[0085] Step 3-2: When one of the following situations occurs, start immediate detection without the actual detection interval of the current next depth interval :

[0086] The sudden increase of the drop speed of the grab is more than 20%, i.e. V d >1.2V d,zvg , V d,zvg is the arithmetic mean of the drop speed of the grab in the last 5 acquisitions;

[0087] The pressure fluctuation of the slurry wall of the super deep foundation pit is more than 5kPa (which may cause micro deformation of the slot wall), and the pressure of the slurry wall of the super deep foundation pit is collected by the pressure sensor connected to the control system and transmitted to the control system.

[0088] The historical data shows that the deviation mutation has occurred in this depth interval (e.g. the deviation of 28m depth has reached 0.08% / m).

[0089] Step 4: The control system of the super deep foundation pit milling machine carries out prediction model self-update and feedback optimization.

[0090] In the preferred but non-limiting embodiment of the application, step 4 specifically includes:

[0091] Step 4-1, the control system of the super deep foundation pit milling machine real-time corrects the prediction model, i.e. after completing 10 times of depth detection, the new deviation growth rate , geological parameters, construction parameters are input into the prediction model, and the prediction model parameters are updated using online learning algorithm (such as incremental random forest) to continuously improve the prediction accuracy with the progress of construction, so as to ensure that the deviation trend can still be accurately captured under complex geological conditions (such as soft soil containing sand interlayer).

[0092] Step 4-2, the control system of the super deep foundation pit milling machine carries out upper limit and lower limit control of detection frequency, i.e. to avoid too frequent detection affecting construction efficiency or too large interval missing deviation, the constraint conditions of the prediction model are set:

[0093] Minimum detection interval: not less than 2m (to ensure the spatial resolution of detection data);

[0094] Maximum detection interval: not more than 10m (whether in low risk interval);

[0095] Daily detection times: not more than 1 / 3 of the total slot depth (e.g. 80m depth, daily detection not more than 27 times).

[0096] Through the method of the application, the detection frequency can be dynamically adjusted according to the characteristics of soft soil, the construction state and the deviation trend, and in the 20-30m deviation sensitive interval, the number of invalid detection times is reduced by 30% compared with the fixed frequency detection, and the deviation missing detection rate is reduced to below 5%. The super-deep diaphragm wall is a super-deep foundation pit. The super-deep foundation pit milling machine is a milling machine used for super-deep foundation pit.

[0097] The beneficial effects of the application are as follows compared with the prior art:

[0098] During the working period of the super-deep foundation pit milling machine, the control system of the super-deep foundation pit milling machine collects basic parameters and data; the control system of the super-deep foundation pit milling machine constructs a deviation risk prediction model; the control system of the super-deep foundation pit milling machine implements a real-time self-adaptive adjustment mechanism; and the control system of the super-deep foundation pit milling machine performs prediction model self-updating and feedback optimization. Through the method of the application, the detection frequency can be dynamically adjusted according to the characteristics of soft soil, the construction state and the deviation trend, and in the 20-30m deviation sensitive interval, the number of invalid detection times is reduced by 30% compared with the fixed frequency detection, and the deviation missing detection rate is reduced to below 5%.

[0099] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the application and not to limit it, although the application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the specific embodiments of the application can still be modified or replaced equivalently without departing from the spirit and scope of the application, any modification or equivalent replacement should be covered in the protection scope of the claims of the application.

Claims

1. A method for vertical control of a trenching machine for ultra-deep foundation pits, characterized in that, The application relates to a method for setting a dynamic adaptive detection frequency of a super-deep foundation trench milling machine. The method comprises the following steps:

2. The vertical control method for a trenching machine for an ultra-deep foundation pit according to claim 1, characterized in that, Step 1: during the operation of the super-deep foundation trench milling machine, the control system of the super-deep foundation trench milling machine collects basic parameters and data; Step 2: the control system of the super-deep foundation trench milling machine constructs a deviation risk prediction model; Step 3: the control system of the super-deep foundation trench milling machine implements a real-time adaptive adjustment mechanism; Step 4: the control system of the super-deep foundation trench milling machine updates and optimizes the prediction model. Step 1 specifically comprises that the control system of the super-deep foundation trench milling machine establishes a database containing three types of key parameters, including:

3. The method for vertical control of a trenching machine for ultra-deep foundation pits according to claim 2, characterized in that, Step 1 specifically further comprises setting an initial detection frequency. Geological parameters as basic parameters: plasticity index I of soft soil of super deep foundation pit p , water content w, cohesion c, internal friction angle , geological stratification data divided by every 5m depth, plasticity index I p , water content w, cohesion c, internal friction angle Respectively through HJ03-STDS-1 disc liquid limit instrument, water content rapid tester, cohesion tester, friction angle tester connected with control system, real-time collection and transmission to control system, real-time collection frequency of HJ03-STDS-1 disc liquid limit instrument, water content rapid tester, cohesion tester, friction angle tester is 1Hz; Construction parameters: the bucket V of the grab of the trenching machine, the lifting speed V of the grab l , the lowering speed V of the grab d , the opening and closing angle α of the grab, the bucket capacity V of the grab, the lifting speed V of the grab l , the lowering speed V of the grab d The opening and closing angle α of the grab is respectively collected in real time by the disc meter, the speed sensor one, the speed sensor two and the angle meter connected with the control system and transmitted to the control system, and the real-time collection frequency of the disc meter, the speed sensor one, the speed sensor two and the angle meter is 1 Hz. Deviation data: verticality deviation growth rate under the same geological conditions in the history project, the i-th verticality value of the over-deep foundation pit detected in the trenching process of the current trencher and the i-th depth value h of the corresponding over-deep foundation pit i The verticality value and the depth value are respectively collected synchronously by the verticality measuring instrument and the depth measuring instrument connected with the control system and transmitted to the control system.

4. The method for vertical control of a trenching machine for ultra-deep foundation pits according to claim 3, characterized in that, The method for setting the initial detection frequency comprises the following steps:

5. The method for vertical control of a trenching machine for ultra-deep foundation pits according to claim 4, characterized in that, During the initial construction of the trench milling machine, a benchmark frequency is adopted: in the shallow soft soil layer with a depth of 0-10 m of the super-deep foundation, the trench milling machine is detected once every 3 m, and 3 m is the detection interval corresponding to the depth; in the medium-deep soft soil layer with a depth of 10-30 m of the super-deep foundation, the trench milling machine is detected once every 5 m, and 5 m is the detection interval corresponding to the depth; in the deep layer with a depth of more than 30 m of the super-deep foundation, the trench milling machine is detected once every 5 m, and 5 m is the detection interval corresponding to the depth; the corresponding geological parameters and construction parameters of the depth are recorded synchronously during each detection, an initial sample set is formed, and the initial sample set is displayed on a display screen connected with the control system.

6. The method for vertical control of a trenching machine for ultra-deep foundation pits according to claim 5, characterized in that, Step 2 specifically comprises the following steps: Step 2-1: the control system of the super-deep foundation trench milling machine constructs a deviation growth rate calculation model; Step 2-2: the control system of the super-deep foundation trench milling machine trains a prediction model based on machine learning.

7. The method for vertical control of a trenching machine for ultra-deep foundation pits according to claim 6, characterized in that, In step 2-1, the method for constructing the deviation growth rate calculation model comprises the following steps: Definition of verticality deviation growth rate As the core index to determine whether the detection frequency needs to be encrypted, the verticality deviation growth rate The calculation formula is: wherein the parameters of the calculation formula have the following meanings: denotes the Kth value of the verticality of the acquired super-deep foundation, denotes the K-1th value of the verticality of the acquired super-deep foundation, h K denotes the Kth value of the depth of the acquired super-deep foundation, h k-1 denotes the K-1th value of the depth of the acquired super-deep foundation, in % / m.

8. The method for vertical control of a trenching machine for ultra-deep foundation pits according to claim 7, characterized in that, Step 2-2 specifically comprises the following steps: Constructing using the random forest regression algorithm The prediction model, its steps are as follows: Step 2-2-1, sample set construction: collect at least 3 sets of historical data of the construction engineering of the trenching machine of the same type of super deep foundation pit, extract not less than 500 groups of samples from each construction engineering, and each group of samples contains geological parameters, construction parameters and corresponding values; Step 2-2-2, Feature importance ranking: screening of prediction model training by random forest regression algorithm The top three feature parameters with the greatest impact are the core variables for input into the prediction model. Step 2-2-3, prediction model verification: 10-fold cross-validation is adopted to ensure that the prediction error of the prediction model on the deviation growth rate is less than or equal to 10%.

9. The method for vertical control of a trenching machine for ultra-deep foundation pits according to claim 8, characterized in that, Step 3 specifically comprises the following steps: Step 3-1: During the milling process, the detection data is sent to the prediction model in real time, and the prediction model acquires new detection data every time With h k The following steps are executed in real time. Step 3-1-1: Calculate the current bias growth rate and call the prediction model to predict the bias growth rate for the next depth interval , which is h k to where is the detection interval corresponding to the reference frequency of h k ; Step 3-1-2: The deviation growth rate of the next depth interval determining the actual detection interval of the next depth interval The determination method is as follows: If If the risk is determined to be low, the actual detection interval for the next depth interval is adjusted to 8~10m; If If the risk is determined to be low, the actual detection interval for the next depth interval is adjusted to 5-6 m; If If the next depth interval is determined to be high risk, the actual detection interval for the next depth interval is encrypted to 2~3m; Step 3-1-3: Actual detection interval of next depth interval combined with geological stratification data Amendments, that is, if the prediction model predicts that the next depth interval contains a known bias-sensitive layer, automatically shorten the actual detection interval of the next depth interval Shorten by 20% again, and the bias-sensitive layer is 20~25m flow plastic layer; Step 3-2: Start immediate detection when one of the following occurs without the actual detection interval of the current next depth interval of the limitation: the sudden increase in the lowering speed of the grab by more than 20%, i.e. V d > 1.2V d,zvg , V d,zvg is the arithmetic mean of the lowering speeds of the grab for the previous 5 acquisitions; The mud wall pressure of the super-deep foundation fluctuates by more than 5 kPa, the mud wall pressure of the super-deep foundation is collected by a pressure sensor connected with the control system and transmitted to the control system; The historical data show that the deviation mutation has occurred in the depth interval.

10. The method for vertical control of a trenching machine for ultra-deep foundation pits according to claim 9, characterized in that, Step 4 specifically comprises the following steps: Step 4-1, the control system of the super-deep foundation pit slot milling machine corrects the prediction model in real time, that is, after 10 depth detections are completed, a new deviation growth rate formed by the 10 depth detections , geological parameters and construction parameters are input into the prediction model, and an online learning algorithm is used to update the prediction model parameters; Step 4-2, the control system of the super-deep foundation trench milling machine controls the upper limit and the lower limit of the detection frequency, that is, in order to avoid that the detection is too frequent to affect the construction efficiency or the interval is too large to miss the deviation, the constraint conditions of the prediction model are set as follows: The minimum detection interval is not less than 2 m; The maximum detection interval is not greater than 10 m; The number of detections per day is not more than 1 / 3 of the total trench depth.

Images