Intelligent installation and preload control system and method for anchor bolts

Abstract

The invention discloses an intelligent installation and pre-tightening force control system and method for anchor bolts, and relates to the technical field of pre-tightening force control. By collecting N groups of training sample sets, the installation feature vector of each training sample is used as input, and the tightening state label is used as output to train a tightening state judgment model. During the installation process, the installation feature set is collected in real time. Based on the installation feature set and the tightening state judgment model, the current tightening state of the anchor bolt to be installed is obtained. For the current tightening state, a corresponding parameter control model is constructed, and based on the installation feature set, the parameter control model outputs recommended installation parameters, and controls an automatic tightening mechanism to adjust the installation parameters according to the recommended installation parameters; the entire tightening process is dynamically controlled and optimized throughout the process, so as to achieve high-precision pre-tightening force control.

Description

Technical Field

[0001] This invention relates to the field of preload control technology, specifically to an intelligent installation and preload control system and method for anchor bolts. Background Art

[0002] Anchor bolts are widely used in infrastructure construction and machinery installation to reliably connect structural components to the foundation. The installation quality of anchor bolts directly affects the stability and safety of the entire structure; therefore, the tightening process of anchor bolts has strict requirements. Traditionally, the installation of anchor bolts mainly relies on manual operation, which suffers from low efficiency and difficulty in quality control.

[0003] Furthermore, the tightening process of anchor bolts is not a simple, single state; it can actually be divided into several key states, such as initial contact, pre-tightening, and critical tightening. The key factors affecting tightening quality differ under each state, requiring targeted parameter control strategies. Relying solely on empirical values ​​for uniform control is insufficient to meet the varying needs of each state, easily leading to defects such as over-tightening or loosening. Therefore, it is necessary to divide the tightening process into states and develop refined parameter control models for each state to ensure quality control throughout the entire process.

[0004] Therefore, this invention proposes an intelligent installation and pre-tightening force control system and method for anchor bolts. Summary of the Invention

[0005] This invention aims to solve at least one of the technical problems existing in the prior art. To this end, this invention proposes an intelligent installation and preload control system and method for anchor bolts, which dynamically controls and optimizes the entire tightening process to achieve high-precision preload control.

[0006] To achieve the above objectives, a method for intelligent installation and preload control of anchor bolts is proposed, including the following steps:

[0007] Step 1: Collect installation history data in advance, and based on the installation history data, divide the data into N training sample sets. Each training sample includes an installation feature vector and a tightening status label; N is the number of training sample sets.

[0008] Step 2: Train the tightening status judgment model by taking the installation feature vector of each training sample as input and the tightening status label as output.

[0009] Step 3: Use an automatic tightening mechanism to install the anchor bolts to be installed. During the installation process, collect the installation feature set in real time.

[0010] Step 4: Based on the installation feature set and tightening status judgment model, obtain the current tightening status of the anchor bolts to be installed;

[0011] Step 5: For the current tightening state, construct the corresponding parameter control model, and based on the installation feature set, output the suggested installation parameters from the parameter control model. Control the automatic tightening mechanism to adjust the installation parameters according to the suggested installation parameters, and use the adjusted installation feature set to fine-tune the parameter control model.

[0012] Specifically, the method for pre-collecting installation history data is as follows:

[0013] N1 installation tests were conducted in advance, with one anchor bolt used in each test. The installation was carried out on different geological conditions in each test, and an automatic tightening mechanism was used in each test. N1 represents the number of installation tests selected.

[0014] During each installation experiment, a set of geological parameters and a set of mechanical parameters are collected using parameter sensors, and the real-time preload of the anchor bolts is also collected; each parameter sensor collects one geological parameter or mechanical parameter.

[0015] The real-time geological parameter set, mechanical parameter set, and real-time preload collected during the installation experiment are used as installation history data;

[0016] The mechanical parameters in the set of mechanical parameters include, but are not limited to, real-time torque value, real-time rotation angle value, and real-time axial displacement value.

[0017] Specifically, the method for dividing the training sample set into N groups based on installation history data is as follows:

[0018] The real-time preload of each installation experiment is connected sequentially in chronological order to form a preload curve;

[0019] According to the preload increase rule, the preload curve is divided into tightening states, and the maximum and minimum values ​​of preload in each tightening state are collected to form the preload range; the tightening states include the initial loosening stage, the transition contact stage, the linear tightening stage, the nonlinear tightening stage, and the stable convergence stage.

[0020] In the installation experiment, the overlapping range of the preload force range for each tightening state is taken as the secondary standard preload force range for that tightening state; the overlapping range refers to the range that is included in the preload force range of each installation experiment.

[0021] Number the tightening states sequentially by time as 1, 2, 3, ..., I; I represents the number of tightening states; label the tightening state number as i;

[0022] For any i-th tightening state, the sub-standard preload range of the i-th tightening state is marked as [di, ui), then ui is updated to... ,di updated to Therefore, the sub-standard preload range for the i-th tightening state is updated to [ , The updated substandard preload range will be expressed as the standard preload range.

[0023] The set of geological parameters and the set of mechanical parameters for each unit time of each installation experiment are combined to form the installation feature vector;

[0024] The real-time preload at each unit moment is used to determine the tightening state according to the standard preload range of each tightening state, and the number of the determined tightening state is used as the tightening state label for that unit moment.

[0025] The installation feature vector and tightening status label at each unit time of each installation experiment are used as a set of training samples.

[0026] Specifically, the method for training the tightening state judgment model is as follows:

[0027] The installation feature vector of each training sample is used as the input to the tightening state judgment model, which outputs the predicted value of the tightening state number at each time step. The tightening state judgment model uses the tightening state label corresponding to each time step as the prediction target, the difference between the predicted value of the tightening state number and the tightening state label as the prediction error, and minimizes the sum of squares of the prediction errors at all times as the training objective. The tightening state judgment model is trained until the sum of squares of the prediction errors converges, at which point training stops. The tightening state judgment model can be any one of the regression models.

[0028] Specifically, the method for collecting the installation feature set in real time is as follows:

[0029] The installation feature set is composed of real-time collection of geological and mechanical parameters during the installation process of the anchor bolts to be installed.

[0030] Specifically, the method for obtaining the current tightening status of the anchor bolts to be installed is as follows:

[0031] The installation feature set is input into the tightening state judgment model, and the predicted value of the tightening state number output by the tightening state judgment model is used as the current tightening state.

[0032] Specifically, the method for constructing the corresponding parameter control model for the current tightening state is as follows:

[0033] The parameter control model was chosen as the Actor-Critic network model.

[0034] Define the state space of the parameter control model; the state space includes the current set of mechanical parameters of the anchor bolts.

[0035] Define the action space of the parameter control model; the action space includes the adjustment values ​​of various installation parameters of the automatic tightening mechanism;

[0036] Construct an Actor network model; the Actor network model takes the current state as input and outputs the adjustment value of each installation parameter in the action space; train it using the policy gradient method, with the goal of maximizing the expected reward value Q;

[0037] Define different reward value Q functions for different tightening states;

[0038] Construct a Critic network model; the input of the Critic network model is the current state and the action selected in the previous time step, and the output is the predicted reward value corresponding to the current state-action; train it using the temporal difference method to minimize the mean square error between the predicted reward value and the actual reward value;

[0039] For the current tightening state, the mechanical parameters of the anchor bolts to be installed are collected in real time as the current state. This current state is input into the Actor network model. The Actor network model outputs the adjustment values ​​of the installation parameters of the automatic tightening mechanism that maximize the reward value Q function, forming the suggested installation parameters. The Actor-Critic network model calculates the actual reward value based on the adjusted set of mechanical parameters of the anchor bolts to be installed. Based on the actual reward value, the parameters of the Actor network model and the Critic network model are adjusted.

[0040] An intelligent installation and preload control system for anchor bolts is proposed, comprising a historical data collection module, a model training module, a tightening status judgment module, and an installation control module; wherein, the modules are connected electrically.

[0041] The historical data collection module collects installation history data in advance and divides it into N training sample sets based on the installation history data. Each training sample includes an installation feature vector and a tightening status label. The training sample set is then sent to the model training module.

[0042] The model training module takes the installation feature vector of each training sample as input and the tightening status label as output to train the tightening status judgment model, and then sends the tightening status judgment model to the tightening status judgment module.

[0043] The tightening status judgment module uses an automatic tightening mechanism to install the anchor bolts to be installed. During the installation process, it collects the installation feature set in real time. Based on the installation feature set and the tightening status judgment model, it obtains the current tightening status of the anchor bolts to be installed and sends the installation feature set and the current tightening status to the installation control module.

[0044] The installation control module constructs a corresponding parameter control model for the current tightening state. Based on the installation feature set, the parameter control model outputs suggested installation parameters, controls the automatic tightening mechanism to adjust the installation parameters according to the suggested installation parameters, and uses the adjusted installation feature set to fine-tune the parameter control model.

[0045] An electronic device is proposed, comprising: a processor and a memory, wherein the memory stores a computer program that can be called by the processor;

[0046] The processor executes the above-described intelligent installation and preload control method for anchor bolts by calling the computer program stored in the memory.

[0047] A computer-readable storage medium is proposed, on which an erasable and rewritable computer program is stored;

[0048] When the computer program is run on a computer device, the computer device performs the above-described intelligent installation and preload control method for anchor bolts.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] This invention pre-collects historical installation data and, based on this data, divides it into N training sample sets. Each training sample includes an installation feature vector and a tightening status label. Using the installation feature vector as input and the tightening status label as output, a tightening status judgment model is trained. An automatic tightening mechanism is used to install the anchor bolts to be installed. During the installation process, the installation feature set is collected in real time. Based on the installation feature set and the tightening status judgment model, the current tightening status of the anchor bolts to be installed is obtained. For the current tightening status, a corresponding parameter control model is constructed, and based on the installation feature set, the parameter control model... The system outputs suggested installation parameters, controls the automatic tightening mechanism to adjust the installation parameters according to these parameters, and uses the adjusted installation feature set to fine-tune the parameter control model. First, a tightening state judgment model is trained using historical data, enabling real-time identification of the current tightening state. Then, for each tightening state, a separate parameter control model is constructed, outputting appropriate installation parameters. Finally, the automatic tightening mechanism performs parameter adjustments to achieve closed-loop optimized control. Based on the specific control requirements of different states, the system can autonomously learn corresponding parameter adjustment strategies, thereby dynamically controlling and optimizing the entire tightening process and achieving high-precision preload control. Attached Figure Description

[0051] Figure 1 This is a flowchart of the intelligent installation and preload control method for anchor bolts in Embodiment 1 of the present invention;

[0052] Figure 2 This is a module connection diagram of the intelligent installation and preload control system for anchor bolts in Embodiment 2 of the present invention;

[0053] Figure 3 This is a schematic diagram of the electronic device structure in Embodiment 3 of the present invention;

[0054] Figure 4 This is a schematic diagram of the computer-readable storage medium structure in Embodiment 4 of the present invention. Detailed Implementation

[0055] The technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0056] Example 1

[0057] like Figure 1 As shown, the intelligent installation and preload control method for anchor bolts includes the following steps:

[0058] Step 1: Collect installation history data in advance, and based on the installation history data, divide the data into N training sample sets. Each training sample includes an installation feature vector and a tightening status label; N is the number of training sample sets.

[0059] Step 2: Train the tightening status judgment model by taking the installation feature vector of each training sample as input and the tightening status label as output.

[0060] Step 3: Use an automatic tightening mechanism to install the anchor bolts to be installed. During the installation process, collect the installation feature set in real time.

[0061] Step 4: Based on the installation feature set and tightening status judgment model, obtain the current tightening status of the anchor bolts to be installed;

[0062] Step 5: For the current tightening state, construct the corresponding parameter control model, and based on the installation feature set, output the suggested installation parameters from the parameter control model. Control the automatic tightening mechanism to adjust the installation parameters according to the suggested installation parameters, and use the adjusted installation feature set to fine-tune the parameter control model.

[0063] The method for pre-collecting installation history data is as follows:

[0064] N1 installation tests were conducted in advance, with one anchor bolt used in each test. Different geological conditions were selected for each test, and an automatic tightening mechanism was used in each test. N1 represents the number of installation tests. The automatic tightening mechanism uses a servo motor and a remote control system to drive a wrench head to tighten the bolt with a set torque and angle. The remote control system controls the installation parameters such as torque, speed, angular displacement, and axial displacement of the wrench head through remote commands.

[0065] During each installation experiment, various parameter sensors are used to collect geological and mechanical parameter sets, and to collect the real-time preload of the anchor bolts; each parameter sensor collects a corresponding geological or mechanical parameter.

[0066] All real-time geological parameters, mechanical parameters, and real-time preload collected during the installation experiment process are used as historical installation data.

[0067] Specifically, the real-time preload collection method can be calculated using a fitting formula method, or it can be obtained by simulating the installation process using simulation software.

[0068] Specifically, a common fitting formula is as follows: In this formula, T represents the tightening torque, K is the tightening torque coefficient, F is the preload, and d is the nominal diameter of the thread. The torque can be obtained using a torque sensor, and K and d are inherent constants of the anchor bolt. However, this formula is a simplified approximation and does not fully consider all factors affecting the preload, thus exhibiting some inaccuracy. For example, unconsidered factors include dynamic factors such as installation angle and speed, as well as surface temperature and humidity (high temperature or humid environments affect friction conditions, thereby affecting the maintenance of the preload).

[0069] In another embodiment, the simulation software can be ABAQUS. The installation process of anchor bolts is simulated using ABAQUS, followed by nonlinear static analysis using the ABAQUS solver. Relevant data such as bolt axial stress and contact surface pressure are then extracted, and the equivalent preload is calculated based on theoretical formulas or experimental data. However, this method has a complex modeling process and makes it difficult to obtain the preload in real time. Therefore, this method can be used in installation experiments to collect accurate preload data, but it is difficult to implement on a large scale.

[0070] The geological parameters included in the geological parameter set include, but are not limited to, ground hardness parameters (such as stratum type, water content, etc.), ground flatness parameters (such as degree of undulation, etc.), ground bearing capacity parameters (such as stratum strength, etc.), and environmental temperature and humidity values.

[0071] The mechanical parameters in the set of mechanical parameters include, but are not limited to, real-time torque value, real-time rotation angle value, and real-time axial displacement value.

[0072] Furthermore, the method for dividing the training sample set into N groups based on installation history data is as follows:

[0073] The real-time preload of each installation experiment is connected sequentially in chronological order to form a preload curve;

[0074] According to the preload increase rule, the preload curve is divided into tightening states, and the maximum and minimum values ​​of preload in each tightening state are collected to form the preload range; specifically, the tightening state changes with time in the following order: initial loosening stage, transitional contact stage, linear tightening stage, nonlinear tightening stage, and stable convergence stage.

[0075] The method for determining the preload increase rule can be as follows:

[0076] Initial loosening stage: The preload is close to 0, and there is no obvious contact between the bolt and the connected parts;

[0077] Transitional contact stage: The preload increases slowly, and contact begins to form between the bolt and the connected parts, but the contact surface is unstable;

[0078] Linear tightening stage: The preload increases rapidly and exhibits an approximately linear relationship. The bolt and the connected parts are in full contact, and the deformation is mainly concentrated in the bolt and washer.

[0079] Nonlinear fastening stage: The preload increases slowly, the curve begins to bend, and the connected parts begin to deform significantly.

[0080] Stable convergence phase: The preload increases very slowly, approaching the convergence value, and the entire connection system enters a relatively stable state.

[0081] Obviously, the installation parameters of the wrench are different for each tightening state;

[0082] Collect the overlapping range of the preload force range for each tightening state in all installation tests as the secondary standard preload force range for that tightening state; the overlapping range refers to the range that is included in the preload force range of each installation test.

[0083] Number the tightening states sequentially by time as 1, 2, 3, ..., I; I represents the number of tightening states; label the tightening state number as i;

[0084] For any i-th tightening state, the sub-standard preload range of the i-th tightening state is marked as [di, ui), then ui is updated to... ,di updated to Therefore, the sub-standard preload range for the i-th tightening state is updated to [ , The updated sub-standard preload range is represented as the standard preload range. It is understood that this update method ensures that there is no gap between the standard preload ranges of any two adjacent tightening states, thus avoiding the situation where the preload cannot fall into any tightening state.

[0085] A typical example of determining the standard preload range for each stage based on multiple installation experiments is as follows: Initial loosening stage: preload range [0, 500 N); Transition contact stage: preload range [500, 2000 N); Linear tightening stage: preload range [2000, 8000 N); Nonlinear tightening stage: preload range [8000, 12000 N); Stable convergence stage: preload range [12000 N, + ).

[0086] The set of geological parameters and the set of mechanical parameters for each unit time of each installation experiment are combined to form the installation feature vector;

[0087] The real-time preload at each unit moment is used to determine the tightening state according to the standard preload range of each tightening state, and the number of the determined tightening state is used as the tightening state label for that unit moment.

[0088] The installation feature vector and tightening status label at each unit time of each installation experiment are used as a set of training samples.

[0089] Furthermore, the method for training the tightening state judgment model, using the installation feature vector of each training sample as input and the tightening state label as output, is as follows:

[0090] The installation feature vector of each training sample is used as the input to the tightening state judgment model, which outputs the predicted value of the tightening state number at each time step. The tightening state judgment model uses the tightening state label corresponding to each time step as the prediction target, the difference between the predicted value of the tightening state number and the tightening state label as the prediction error, and minimizing the sum of squares of the prediction errors at all times as the training objective. The tightening state judgment model is trained until the sum of squares of the prediction errors converges, at which point training stops. The tightening state judgment model can be any regression model, including but not limited to the SVR model or the multinomial regression model.

[0091] Furthermore, the method for collecting the installation feature set in real time is as follows:

[0092] The installation feature set is composed of real-time collection of all geological parameters and all mechanical parameters during the installation process of the anchor bolts to be installed.

[0093] Furthermore, the method for obtaining the current tightening status of the anchor bolts to be installed based on the installation feature set and tightening status judgment model is as follows:

[0094] The installation feature set is input into the tightening state judgment model, and the predicted value of the tightening state number output by the tightening state judgment model is used as the current tightening state.

[0095] Furthermore, the method for constructing the corresponding parameter control model for the current tightening state is as follows:

[0096] The parameter control model was chosen as the Actor-Critic network model.

[0097] It should be noted that the Actor model is used to output suggested installation parameters, the Critic model is used to evaluate and score the suggested installation parameters output by the Actor model, and the Actor model then updates its model parameters based on the score.

[0098] Define the state space of the parameter control model; the state space includes the current set of mechanical parameters of the anchor bolts; then, for each moment, the current state of the model is the current set of mechanical parameters of the anchor bolts.

[0099] Define the action space of the parameter control model; the action space includes the adjustment values ​​of various installation parameters of the automatic tightening mechanism; for example, the adjustment ratio of torque, the adjustment ratio of speed, etc.

[0100] Construct an Actor network model; the Actor network model takes the current state as input and outputs the adjustment value of each installation parameter in the action space; train it using the policy gradient method, with the goal of maximizing the expected reward value Q;

[0101] Define different reward value Q functions for different tightening states;

[0102] Specifically, for the initial loosening stage, a positive reward is set for rapidly approaching the target preload force;

[0103] For the linear fastening stage, a positive reward is set for gradually approaching the target preload force;

[0104] For the nonlinear convergence phase, a positive reward is set for slowly approaching and stabilizing within the target interval.

[0105] The goal for the transitional contact phase is to pass through this phase as quickly and smoothly as possible, avoiding violent impacts.

[0106] For the stable convergence phase, the goal is to ensure that the preload value can converge stably and remain within the allowable range for a long period of time.

[0107] Therefore, in the preferred embodiment, the specific definition of the reward value Q function in the initial loosening state can be:

[0108] ;

[0109] Here, α and β are positive weighting coefficients used to balance the relative importance of penalty and reward terms.

[0110] F_target is the target preload, and F_curr is the current preload. The target preload is the lower limit of the standard preload range for the next tightening state. The current preload is roughly obtained by calculating using a preload fitting formula.

[0111] For a linearly fixed state, the specific definition of the reward function Q can be:

[0112] ;

[0113] Where δ is the preset maximum allowable deviation range; it can be understood that the reward value Q function aims to make the preload value linearly approach and stabilize within the target value δ range.

[0114] In a nonlinear convergent state, the reward function Q is defined as:

[0115] Let [F_min, F_max] be the target preload range;

[0116] ;

[0117] Here, γ and λ are positive weighting coefficients used to control the severity of the penalty. This function ensures that the preload converges smoothly to the target range, avoiding excessive tightness or looseness.

[0118] During the transitional contact phase, the reward value Q function is defined as:

[0119] ;

[0120] Wherein, F_contact is the lower limit of the standard preload range for the transition contact phase, μ and ρ are positive weighting coefficients, and σ controls the variance of the normal distribution to make the curve transition smoothly. When the preload is lower than F_contact, a larger penalty is given to promote rapid lift. Once it exceeds F_contact, an exponential positive reward is given to promote smooth passage through the contact interval.

[0121] During the stable convergence phase, the reward function Q is defined as:

[0122] ;

[0123] Where F_target is the target preload value, ε is the allowable deviation range, κ is the positive weighting coefficient, var(F_window) is the unbiased variance of the preload value within the window, reflecting the degree of fluctuation, and ζ is the positive weighting coefficient, which controls the weight of the variance penalty.

[0124] Understandably, by setting different reward value functions for different tightening states, the system can autonomously learn corresponding parameter adjustment strategies based on the specific control requirements of different states, thereby dynamically controlling and optimizing the entire tightening process and achieving high-precision preload control.

[0125] Construct a Critic network model; the input of the Critic network model is the current state and the action selected in the previous time step, and the output is the predicted reward value corresponding to the current state-action; train it using the temporal difference method to minimize the mean square error between the predicted reward value and the actual reward value;

[0126] For the current tightening state, the mechanical parameters of the anchor bolts to be installed are collected in real time as the current state. This current state is input into the Actor network model. The Actor network model outputs the adjustment values ​​of the installation parameters of the automatic tightening mechanism that maximize the reward value Q function, forming the suggested installation parameters. The Actor-Critic network model calculates the actual reward value based on the adjusted set of mechanical parameters of the anchor bolts to be installed. Based on the actual reward value, the parameters of the Actor network model and the Critic network model are adjusted.

[0127] Example 2

[0128] like Figure 2 As shown, the intelligent installation and preload control system for anchor bolts includes a historical data collection module, a model training module, a tightening status judgment module, and an installation control module; the modules are connected electrically.

[0129] The historical data collection module collects installation history data in advance and divides it into N training sample sets based on the installation history data. Each training sample includes an installation feature vector and a tightening status label. The training sample set is then sent to the model training module.

[0130] The model training module takes the installation feature vector of each training sample as input and the tightening status label as output to train the tightening status judgment model, and then sends the tightening status judgment model to the tightening status judgment module.

[0131] The tightening status judgment module uses an automatic tightening mechanism to install the anchor bolts to be installed. During the installation process, it collects the installation feature set in real time. Based on the installation feature set and the tightening status judgment model, it obtains the current tightening status of the anchor bolts to be installed and sends the installation feature set and the current tightening status to the installation control module.

[0132] The installation control module constructs a corresponding parameter control model for the current tightening state. Based on the installation feature set, the parameter control model outputs suggested installation parameters, controls the automatic tightening mechanism to adjust the installation parameters according to the suggested installation parameters, and uses the adjusted installation feature set to fine-tune the parameter control model.

[0133] Example 3

[0134] Figure 3 This is a schematic diagram of an electronic device structure provided in one embodiment of this application. Figure 3As shown, according to another aspect of this application, an electronic device 100 is also provided. The electronic device 100 may include one or more processors and one or more memories. The memories store computer-readable code that, when executed by the one or more processors, can perform the intelligent installation and preload control method for anchor bolts as described above.

[0135] The method or apparatus according to the embodiments of this application can also be used by means of Figure 3 The architecture of the electronic device shown is used to implement this. For example... Figure 3 As shown, the electronic device 100 may include a bus 101, one or more CPUs 102, ROM 103, RAM 104, a communication port 105 connected to a network, an input / output component 106, a hard disk 107, etc. The storage device in the electronic device 100, such as ROM 103 or hard disk 107, may store the intelligent installation and preload control method for anchor bolts provided in this application.

[0136] Furthermore, the electronic device 100 may also include a user interface 108. Of course, Figure 3 The architecture shown is merely exemplary and can be omitted as needed when implementing different devices. Figure 3 One or more components in the illustrated electronic device.

[0137] Example 4

[0138] Figure 4 This is a schematic diagram of a computer-readable storage medium structure provided in one embodiment of this application. Figure 4 The diagram illustrates a computer-readable storage medium 200 according to one embodiment of this application. The computer-readable storage medium 200 stores computer-readable instructions. When executed by a processor, the computer-readable instructions can perform the intelligent installation and preload control method for anchor bolts according to an embodiment of this application, as described above with reference to the accompanying drawings. The computer-readable storage medium 200 includes, but is not limited to, volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and cache memory. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0139] Furthermore, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, this application provides a non-transitory machine-readable storage medium storing machine-readable instructions that can be executed by a processor to perform instructions corresponding to the method steps provided in this application. When the computer program is executed by a central processing unit (CPU), it performs the functions defined in the method of this application.

[0140] The methods, apparatus, and devices of this application may be implemented in many ways. For example, they may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order of steps for the method is for illustrative purposes only, and the steps of the method of this application are not limited to the order specifically described above, unless otherwise specifically stated. Furthermore, in some embodiments, this application may also be implemented as a program recorded on a recording medium, the program including machine-readable instructions for implementing the method according to this application. Thus, this application also covers recording media storing programs for performing the method according to this application.

[0141] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0142] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0143] The preset parameters or preset thresholds mentioned above are all set by those skilled in the art based on actual conditions or obtained through large-scale data simulation.

[0144] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. Intelligent installation and preload control method of anchor bolts, characterized in that: The following steps are involved: Step 1: Collect installation history data in advance, and divide N groups of training sample sets based on the installation history data, each training sample includes an installation feature vector and a tightening state label; N is the number of divided training sample sets; Step 2: Take the installation feature vector of each training sample as input and the tightening state label as output to train the tightening state judgment model; Step 3: Use the automatic tightening mechanism to install the anchor bolts to be installed, and collect the installation feature set in real time during the installation process; Step 4: Based on the installation feature set and the tightening state judgment model, obtain the current tightening state of the anchor bolt to be installed; Step 5: For the current tightening state, a corresponding parameter control model is constructed, and based on the installation feature set, the parameter control model outputs recommended installation parameters, controls the automatic tightening mechanism to adjust the installation parameters according to the recommended installation parameters, and uses the adjusted installation feature set to adjust the parameter control model; For the current tightening state, the method of constructing the corresponding parameter control model is as follows: Select the parameter control model as Actor-Critic network model; Defining a state space of a parameter control model; the state space includes a current set of mechanical parameters of the anchor bolt; Defining an action space of a parameter control model; the action space includes adjustment values ​​of various installation parameters of the automatic tightening mechanism; Construct an Actor network model; the Actor network model takes the current state as input and the adjustment value of each installation parameter in the action space as output; and uses the policy gradient method for training, with the goal of maximizing the expected reward value Q; Define different reward value Q functions for different tightening states; Construct a Critic network model; the input of the Critic network model is the current state and the action selected at the last moment, and the output is the predicted reward value corresponding to the current state-action; Use the temporal difference method to train and minimize the mean square error between the predicted reward value and the actual reward value; For the current tightening state, the mechanical parameter set of the anchor bolts to be installed is collected in real time as the current state, and the current state is input into the Actor network model to obtain the output of the Actor network model to maximize the reward value Q function. The adjustment values ​​of the installation parameters of the automatic tightening mechanism constitute the recommended installation parameters. The Actor-Critic network model calculates the real reward value according to the mechanical parameter set of the anchor bolts to be installed after adjustment, and adjusts the parameters of the Actor network model and the Critic network model based on the real reward value.

2. The method for intelligent installation and preload control of anchor bolts according to claim 1, characterized in that: The method of pre-collecting installation history data is: Conduct N1 installation experiments in advance, use one anchor bolt for each installation experiment, and select a surface with different geological conditions for installation in each installation experiment. Use an automatic tightening mechanism for installation in each installation experiment; N1 is the number of selected installation experiments; During each installation experiment, a parameter sensor is used to collect a set of geological parameters and a set of mechanical parameters, and to collect the real-time preload force of the anchor bolts; each of the parameter sensors correspondingly collects a geological parameter or a mechanical parameter; The real-time geological parameter set, mechanical parameter set and real-time preload force collected during the installation experiment are used as installation history data.

3. The method for intelligent installation and preload control of anchor bolts according to claim 2, characterized in that: The method of dividing N groups of training sample sets based on the installation history data is as follows: The real-time preload force of each installation experiment is connected in chronological order to form a preload force curve; According to the preload increase rule, the preload curve is divided into tightening states, and the maximum and minimum values ​​of the preload under each tightening state are collected to form the preload range; Collect the overlapping range of the pre-tightening force range of each tightening state in the installation experiment as the sub-standard pre-tightening force range of the tightening state; The tightening states are numbered 1, 2, 3, ..., I in chronological order; I is the number of the tightening states; the number of the tightening state is marked as i; For any i-th tightening state, the substandard preload range of the i-th tightening state is marked as [d i ,u i ), then u i Updated to , d i Updated to ; Thus, the substandard preload range of the i-th tightening state is updated to , the updated substandard preload range is expressed as the standard preload range; The geological parameter set and mechanical parameter set of each installation experiment at each unit time are combined into an installation feature vector; The real-time pre-tightening force of each unit time is used to determine the tightening state according to the standard pre-tightening force range of each tightening state, and the number of the determined tightening state is used as the tightening state label of the unit time; The installation feature vector and tightening state label of each unit moment of each installation experiment are used as a set of training samples.

4. The method for intelligent installation and preload control of anchor bolts according to claim 3, characterized in that: The tightening state includes an initial loosening stage, a transition contact stage, a linear tightening stage, a nonlinear tightening stage and a stable convergence stage.

5. The method for intelligent installation and preload control of anchor bolts according to claim 4, characterized in that: The method of training the tightening state judgment model is: The installation feature vector of each training sample is used as the input of the tightening state judgment model, and the tightening state judgment model takes the predicted value of the tightening state number at each moment as the output; the tightening state judgment model takes the tightening state label corresponding to each moment as the prediction target, takes the difference between the predicted value of the tightening state number and the tightening state label as the prediction error, and takes minimizing the sum of squares of the prediction errors at all moments as the training target; The tightening state judgment model is trained until the sum of squares of the prediction errors reaches convergence and the training is stopped; the tightening state judgment model is any one of the regression models.

6. The method for intelligent installation and preload control of anchor bolts according to claim 5, characterized in that: The method of collecting the installation feature set in real time is: The parameter values ​​of geological parameters and mechanical parameters during the installation of the anchor bolts to be installed are collected in real time to form an installation feature set.

7. The method for intelligent installation and preload control of anchor bolts according to claim 6, characterized in that: The method for obtaining the current tightening state of the anchor bolt to be installed is: The installation feature set is input into the tightening state judgment model, and the predicted value of the tightening state number output by the tightening state judgment model is obtained as the current tightening state.

8. An intelligent installation and pre-tightening force control system for anchor bolts, which is used to implement the intelligent installation and pre-tightening force control method for anchor bolts as described in any one of claims 1 to 7, characterized in that: It includes a historical data collection module, a model training module, a tightening state judgment module and an installation control module; wherein each module is electrically connected; A historical data collection module collects installation historical data in advance, and divides N sets of training sample sets based on the installation historical data, each training sample includes an installation feature vector and a tightening state label, and sends the training sample sets to the model training module; The model training module takes the installation feature vector of each training sample as input and the tightening state label as output, trains the tightening state judgment model, and sends the tightening state judgment model to the tightening state judgment module; The tightening state judgment module uses an automatic tightening mechanism to install the anchor bolts to be installed. During the installation process, the installation feature set is collected in real time. Based on the installation feature set and the tightening state judgment model, the current tightening state of the anchor bolts to be installed is obtained, and the installation feature set and the current tightening state are sent to the installation control module; The installation control module constructs a corresponding parameter control model for the current tightening state, and based on the installation feature set, the parameter control model outputs recommended installation parameters, controls the automatic tightening mechanism to adjust the installation parameters according to the recommended installation parameters, and uses the adjusted installation feature set to adjust the parameter control model.

9. An electronic device, characterized in that: include: processor and memory, wherein The memory stores a computer program that can be called by the processor; The processor executes the anchor bolt intelligent installation and preload control method according to any one of claims 1 to 7 in the background by calling the computer program stored in the memory.

10. A computer-readable storage medium, characterized in that: A rewritable computer program is stored thereon; When the computer program is run on a computer device, the computer device executes the anchor bolt intelligent installation and preload control method according to any one of claims 1 to 7 in the background.

Images