Intelligent control method and system for serialized screw tightening process in ultra-high vacuum environment

Abstract

The application relates to the technical field of industrial control, and discloses a serialized screw tightening process intelligent control method and system under an ultrahigh vacuum environment. The method comprises the following steps: acquiring screw coordinates, material and environmental vacuum degree; correcting the interval according to the screw coordinates to obtain corrected coordinates; matching the initial friction coefficient according to the screw material, and correcting the initial friction coefficient by using the environmental vacuum degree to obtain a verified friction coefficient; calculating a distribution proportion coefficient, generating an initial tightening instruction, and executing a calculated torque deviation rate, and performing hierarchical processing according to the torque deviation rate to obtain a depth compensation list; performing torque depth compensation on the depth compensation list to obtain a compensation tightening instruction; executing the compensation tightening instruction and identifying a stress safety state, performing stress deviation quantitative calculation on a stress anomaly list to calculate an initial stress deviation rate, and performing stress compensation tightening iterative optimization to evaluate stress uniformity and obtain an optimization report. The method improves the stability of screw tightening under a vacuum environment.

Description

Technical Field

[0001] This invention relates to the field of industrial control technology, and in particular to an intelligent control method and system for a serialized screw tightening process under ultra-high vacuum conditions. Background Technology

[0002] Currently, in ultra-high vacuum environments, when connecting multiple screws, changes in vacuum level alter the material's frictional properties and stress behavior, making it difficult to precisely apply the preset torque value. Improper tightening sequence can easily lead to localized stress concentration or sealing failure, directly affecting the airtightness of the cavity and the stability of the equipment. Therefore, an industrial control system capable of real-time sensing of operating condition fluctuations and dynamically adjusting the force application strategy is needed to collaboratively plan the torque increment and force application sequence for each screw, thereby ensuring the consistency and reliability of the connection process under extreme conditions.

[0003] In a prior art, during the assembly of multiple screws for an ultra-high vacuum chamber cover, a wrench with a preset fixed torque value is used to tighten the screws one by one in a diagonal circumferential sequence. During operation, all screws are initially pre-tightened with a low torque, and then the torque is gradually increased in two to three turns until the set value is reached, with each turn following the same tightening sequence. This prior art, through an industrial control system, tightens the screws one turn at a preset fixed torque value and in a fixed diagonal sequence, without adjusting the applied force parameters online according to changes in vacuum level. This causes the actual pre-tightening force to deviate from the target and results in uneven force distribution.

[0004] Therefore, existing technologies suffer from low screw tightening stability in a vacuum environment. Summary of the Invention

[0005] This invention provides an intelligent control method and system for a serialized screw tightening process under ultra-high vacuum conditions, so as to improve the stability of screw tightening under vacuum conditions.

[0006] Firstly, to address the aforementioned technical problems, this invention provides an intelligent control method for a serialized screw tightening process under ultra-high vacuum conditions, comprising: Obtain the screw's location coordinates, screw material, and ambient vacuum level; Based on the screw position coordinates, perform spacing constraint position correction to obtain the corrected position coordinates; The initial friction coefficient is matched according to the screw material, and the initial friction coefficient is corrected for vacuum influence using the ambient vacuum level to obtain the verified friction coefficient. Based on the corrected position coordinates and the verified friction coefficient, priority weight allocation calculation is performed to obtain the allocation ratio coefficient; An initial tightening command is generated based on the verified friction coefficient and the distribution ratio coefficient, and feedback is performed to calculate the torque deviation rate. Then, a graded torque compensation process is performed based on the torque deviation rate to obtain a depth compensation list. Based on the depth compensation list and the torque deviation rate, torque depth compensation is performed to obtain a compensated tightening command; The compensation tightening command is executed and the stress safety status is identified to obtain a stress safety list and a stress anomaly list. The stress deviation of the stress anomaly list is quantified to calculate the initial stress deviation rate. The initial stress deviation rate is used to perform stress compensation tightening iterative optimization to obtain the final stress value. Based on the final stress value and the initial stress deviation rate, a quantitative assessment of stress uniformity is performed to obtain an optimization report.

[0007] Secondly, the present invention provides an intelligent control system for a serialized screw tightening process under ultra-high vacuum conditions, comprising: The data acquisition module is used to acquire the screw position coordinates, screw material, and ambient vacuum level; The position correction module is used to perform spacing constraint position correction based on the screw position coordinates to obtain the corrected position coordinates; The vacuum influence module is used to match the initial friction coefficient according to the screw material, and to perform vacuum influence correction on the initial friction coefficient using the ambient vacuum level to obtain the verification friction coefficient. The allocation calculation module is used to perform priority weight allocation calculation based on the corrected position coordinates and the verified friction coefficient to obtain the allocation ratio coefficient; The graded processing module is used to generate an initial tightening command based on the verified friction coefficient and the distribution ratio coefficient, perform feedback calculation to determine the torque deviation rate, and perform graded torque compensation processing based on the torque deviation rate to obtain a depth compensation list. The torque compensation module is used to perform torque depth compensation based on the depth compensation list and the torque deviation rate to obtain a compensation tightening command; The stress optimization module is used to execute the compensation tightening command and perform stress safety status identification to obtain a stress safety list and a stress anomaly list. It then performs stress deviation quantification on the stress anomaly list to calculate the initial stress deviation rate and uses the initial stress deviation rate to perform stress compensation tightening iterative optimization to obtain the final stress value. The output module is used to perform a quantitative evaluation of stress uniformity based on the final stress value and the initial stress deviation rate, and to obtain an optimization report.

[0008] Compared with the prior art, the present invention has the following beneficial effects: (1) This invention obtains the initial friction coefficient matching the screw material and uses the environmental vacuum degree to correct the vacuum effect through a pre-constructed friction adaptation model to obtain the verification friction coefficient. Since the model is trained based on historical data, it can quantify the nonlinear effect of vacuum degree on friction coefficient. The corrected verification friction coefficient is closer to the real friction behavior under ultra-high vacuum environment. At the same time, it combines the correction position coordinates to allocate priority weights and dynamically adjusts the tightening sequence and allocation ratio coefficient according to the distance between adjacent screws and vibration frequency. This solves the problem that the existing technology cannot adapt to changes in vacuum degree and the risk of local stress concentration by using fixed torque value and fixed sequence. It realizes the optimized allocation of torque load in spatial sequence, suppresses the tendency of uneven force from the root, and improves the adaptability of tightening process to extreme environment.

[0009] (2) This invention generates an initial tightening command and performs feedback to calculate the torque deviation rate. Based on the deviation rate, it performs graded torque compensation processing: when the deviation rate is between the first and second deviation thresholds, a single tightening operation is performed; when the deviation rate is higher than the second deviation threshold, a depth compensation list is output, and the compensation torque value is further calculated through the torque compensation function to generate a compensation tightening command. Since the graded mechanism distinguishes between slight deviation and severe deviation, the depth compensation reorders and incrementally compensates for screws with severe deviations. This solves the problem of torque deviation accumulation caused by single-level compensation or lack of compensation strategy in the prior art, realizes multi-level torque closed-loop control from coarse adjustment to fine adjustment, reduces the torque deviation rate, and ensures the consistency of preload for each screw.

[0010] (3) This invention obtains the stress value around the screw after executing the compensation tightening command, identifies the stress safety status to obtain a stress safety list and a stress anomaly list, calculates the initial stress deviation rate for the anomaly list, and uses the deviation rate to perform stress compensation tightening iterative optimization. The initial stress deviation rate is ordered from large to small, and the stress optimization torque value is incremented for single tightening. The iteration is repeated until the stress value falls into the safety range, and the final stress value is obtained. Since this method uses the stress deviation rate as a feedback quantity for closed-loop iterative control, and the iteration order prioritizes the screw with the largest deviation, it solves the problem that the existing technology only monitors the torque and ignores the actual stress distribution, and cannot eliminate the problem of excessively high or low local stress. It realizes the active balance adjustment of stress state. Finally, through the quantitative evaluation of stress uniformity, an optimization report containing the change in non-uniformity and the percentage increase is generated, which provides a quantifiable evaluation basis for process quality. Attached Figure Description

[0011] Figure 1 This is a schematic diagram of the intelligent control method for serialized screw tightening process under ultra-high vacuum environment provided in the first embodiment of the present invention; Figure 2This is a schematic diagram of the intelligent control system for serialized screw tightening process under ultra-high vacuum environment provided in the second embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] Reference Figure 1 The first embodiment of the present invention provides an intelligent control method for a serialized screw tightening process under ultra-high vacuum conditions, comprising the following steps: S11, obtain the screw position coordinates, screw material and ambient vacuum level; S12, perform spacing constraint position correction based on the screw position coordinates to obtain the corrected position coordinates; S13, Match the initial friction coefficient according to the screw material, and use the ambient vacuum degree to correct the vacuum effect of the initial friction coefficient to obtain the verification friction coefficient; S14, Based on the corrected position coordinates and the verified friction coefficient, perform priority weight allocation calculation to obtain the allocation ratio coefficient; S15, an initial tightening command is generated based on the verified friction coefficient and the distribution ratio coefficient, and feedback is performed to calculate the torque deviation rate. Then, a graded torque compensation process is performed based on the torque deviation rate to obtain a depth compensation list. S16, Perform torque depth compensation according to the depth compensation list and the torque deviation rate to obtain a compensation tightening command; S17, execute the compensation tightening command and perform stress safety status identification to obtain a stress safety list and a stress anomaly list, perform stress deviation quantification on the stress anomaly list to calculate the initial stress deviation rate, and use the initial stress deviation rate to perform stress compensation tightening iterative optimization to obtain the final stress value. S18. Based on the final stress value and the initial stress deviation rate, perform a quantitative assessment of stress uniformity to obtain an optimization report.

[0014] In step S11, the screw position coordinates, screw material, and ambient vacuum level are obtained.

[0015] Specifically, the screw position coordinates are acquired by an industrial camera mounted above the ultra-high vacuum chamber in conjunction with a structured light scanning device. The industrial camera captures grayscale images of the tightening area, extracts the edge contour of the screw head using an edge detection operator, and determines the center pixel coordinates of each screw using a least-squares ellipse fitting algorithm. These coordinates are then converted into three-dimensional spatial coordinates with the chamber flange reference point as the origin using a pre-calibrated camera intrinsic and extrinsic parameter matrix, yielding the screw position coordinate data. The horizontal and vertical coordinates represent the distribution position of the screw on the sealing plane, and the vertical coordinate records the height of the screw end face. The screw material information comes from the screw model code preset in the assembly task file. The system looks up the corresponding material grade in a pre-stored material mapping table based on this code. This mapping table contains the correspondence between material codes and commonly used ultra-high vacuum applicable metal materials such as stainless steel, titanium alloy, and copper alloy. The mapping table is imported and stored in a standard material database in one go. The ambient vacuum level is collected in real time by a vacuum gauge installed on the outer wall of the cavity. The gauge converts the pressure signal into an analog voltage signal, which is then converted from analog to digital to obtain a value in Pascals. The arithmetic mean of the five consecutive sampling periods before the acquisition time is taken as the current ambient vacuum level input in the main controller.

[0016] This step provides three types of fundamental data for the entire tightening control process: spatial positioning reference, material constitutive parameter sources, and operating environment conditions. Screw position coordinates provide a clear spatial reference for subsequent spacing constraint correction and tightening queue sequencing; screw material information provides a material basis for the initial matching of the friction coefficient; and real-time acquisition of the ambient vacuum level allows vacuum influence correction to reflect the dynamic changes in friction behavior under current operating conditions. These three variables together constitute a complete set of input elements, from geometric constraints and material properties to environmental influences, ensuring that subsequent processing stages can perform quantitative calculations and closed-loop control within a unified data framework.

[0017] In step S12, spacing constraint position correction is performed based on the screw position coordinates to obtain the corrected position coordinates, including: Based on the screw position coordinates, the Euclidean distance between adjacent screws is calculated as the spacing between adjacent screws; When the distance between adjacent screws is not less than the preset minimum distance, the screw position coordinates are output as correction position coordinates; When the distance between adjacent screws is lower than the preset minimum distance, the scanning angle is adjusted to reacquire the screw position coordinates and the distance between adjacent screws is recalculated until the distance between adjacent screws is not lower than the minimum distance. Then, the screw position coordinates are output as the corrected position coordinates.

[0018] Specifically, first, all screw position coordinates are traversed. For each screw, the Euclidean distance between it and its geometric nearest neighbor is calculated. This is done by summing the squares of the differences between the coordinates of two screws in the x, y, and y axes, and then taking the square root of this sum. The resulting value is the distance between adjacent screws. When there are multiple rings on the flange plane, the nearest neighbor is determined based on the smallest straight-line distance in the three-dimensional coordinates. After the traversal and calculation are completed, a set of adjacent screw distances corresponding to the number of screws is obtained.

[0019] The preset minimum spacing threshold is derived from historical assembly data statistics of the same type of ultra-high vacuum cavity flange. During the process calibration phase before system deployment, assembly cases from previous batches with acceptable stress uniformity on the sealing surface after screw tightening are collected. The actual measured values ​​of the spacing between all adjacent screws in each batch are extracted. All historical spacing data are sorted from smallest to largest, and the value at the 95th percentile after sorting is taken as the minimum spacing threshold. This threshold is written into the system configuration file and called as a fixed constant during operation. Using the 95th percentile to determine the threshold can eliminate extremely small spacing samples caused by measurement noise or occasional installation misalignment, while retaining the constraint on the vast majority of normal operating conditions.

[0020] When the calculated spacing between adjacent screws is not lower than the preset minimum spacing, the current screw position coordinates are determined to meet the spatial distribution uniformity requirement, and the original screw position coordinates are directly output as the correction position coordinates. When the spacing between adjacent screws of a certain screw is lower than the preset minimum spacing, it indicates that there is a risk of the screw being too close to its adjacent screws in space. In this case, the scanning angle adjustment mechanism is triggered. The specific implementation of adjusting the scanning angle is to send a deflection command to the gimbal control unit of the industrial camera. The deflection angle is taken as a preset angle increment value, such as increasing the current pitch angle and yaw angle by a fixed angle step value. The gimbal drives the industrial camera and the structured light scanning device to re-image the tightening area from a new perspective, and the edge detection, ellipse fitting and coordinate transformation process in step S11 is executed again to obtain the screw position coordinates again and recalculate the spacing between adjacent screws. The process is executed in a loop. Each loop checks whether the recalculated spacing between adjacent screws is no less than the minimum spacing threshold. If it is still below the threshold, the scanning angle is adjusted and the data is acquired again until the spacing between adjacent screws meets the threshold condition. At this point, the last acquired screw position coordinates are output as the correction position coordinates. The angle increment is preset to a fixed step size of one degree. This step size is determined through comparative experiments using multiple sets of coordinate measurements with different angle step sizes during the prototype assembly stage, selecting the smallest step size that allows for stable reproduction of the adjacent screw spacing measurement value. A maximum loop count is set during the loop process to prevent infinite looping due to hardware malfunctions. When the maximum number of loops is reached, the system outputs an alarm and pauses the process. The maximum number of loops is preset to five times. This value is determined based on the physical travel range of the industrial camera gimbal's single-wheel scanning angle adjustment. Five loops can cover the entire effective viewing angle change range; exceeding this range will indicate an abnormality.

[0021] This step eliminates misjudgments of position caused by single-shot perspective occlusion, screw reflection, or edge extraction errors through spacing constraint correction, ensuring the geometric rationality of the spatial coordinates input for subsequent priority weight allocation and tightening queue sorting. The corrected position coordinates, as screw position data verified for spatial uniformity, provide a reliable spatial relationship basis for calculating center distances, filtering nearby screw lists, and constructing tightening sequence queues in step S14, thereby reducing the potential for improper tightening sequence and localized stress concentration caused by positional deviations.

[0022] In step S13, an initial friction coefficient is matched according to the screw material, and the initial friction coefficient is corrected for vacuum influence using the ambient vacuum level to obtain a verified friction coefficient, including: Based on the screw material, an initial friction coefficient is obtained by linearly mapping it to a pre-stored material friction coefficient library. Based on the initial friction coefficient and the ambient vacuum level, the influence of vacuum level on the friction coefficient is quantified using a pre-constructed friction adaptation model to obtain the vacuum influence coefficient; The vacuum influence coefficient is added to the initial friction coefficient to obtain the corrected friction coefficient; When the corrected friction coefficient is greater than the preset lower limit of the friction coefficient and less than the preset upper limit of the friction coefficient, the corrected friction coefficient is output as the verification friction coefficient. When the corrected friction coefficient is not greater than the preset lower limit of the friction coefficient, the lower limit of the friction coefficient is output as the verification friction coefficient; when the corrected friction coefficient is not less than the preset upper limit of the friction coefficient, the upper limit of the friction coefficient is output as the verification friction coefficient.

[0023] Specifically, the initial friction coefficient is matched according to the screw material, and the initial friction coefficient is corrected for vacuum influence using the ambient vacuum level to obtain the verification friction coefficient. A pre-stored material friction coefficient library is stored in the controller's non-volatile memory in the form of a two-dimensional data table. The data table includes a material name field and a corresponding reference value field for the threaded pair friction coefficient under standard atmospheric conditions. The data source for this library is a compilation of experimentally measured friction coefficient values ​​for commonly used metallic materials in ultra-high vacuum under standard atmospheric conditions. During the compilation process, several batches of samples from stainless steel, titanium alloy, and copper alloy were selected and subjected to paired friction tests on a friction and wear testing machine at a constant loading rate and constant sliding speed. Each test was repeated multiple times, and the arithmetic mean of the friction coefficient in the steady-state phase was recorded. This average value was written into the record row corresponding to the material name in the material friction coefficient library. When the screw material obtained in step S11 is input, the system performs a string matching query, locates the record with the same material name in the material friction coefficient library, reads the reference value of the friction coefficient stored in that record, and directly assigns this value to the initial friction coefficient variable, completing the linear mapping process.

[0024] The pre-built friction adaptation model employs a feedforward neural network structure to quantify the nonlinear influence of ultra-high vacuum environment on the friction coefficient of threaded pairs. The network input layer contains two neurons, corresponding to the initial friction coefficient and the ambient vacuum level, respectively. Before entering the network, the input data undergoes min-max normalization preprocessing to linearly map the values ​​to the zero-to-one range. The network contains two hidden layers: the first hidden layer has eight neurons, and the second hidden layer has four neurons. Each hidden layer uses the hyperbolic tangent function as the activation function to introduce nonlinear expressive power. The output layer contains one neuron, directly outputting the predicted value of the vacuum influence coefficient using a linear activation function. The model's training data comes from experimental records of threaded pair friction coefficient calibration under ultra-high vacuum conditions. The experiment tested the relationship between tightening torque and preload for screws of various materials under different vacuum levels. The actual friction coefficient was calculated using the torque-preload conversion formula. The difference between the actual friction coefficient and the friction coefficient under corresponding standard atmospheric conditions was recorded as the historical vacuum influence coefficient. Simultaneously, the historical ambient vacuum level and historical initial friction coefficient were recorded, forming a training dataset containing hundreds of samples. During training, the historical initial friction coefficient and historical environmental vacuum level are input into the network for forward propagation to obtain the output value. The output value and the historical vacuum influence coefficient are substituted into the mean squared error loss function to calculate the loss value. The gradient of the loss value with respect to the weights and biases of each layer is obtained through backpropagation. Stochastic gradient descent is used to update the parameters. The learning rate is set to a fixed small value, such as 0.001. This value is a conservative value recommended in neural network training, which can achieve high training efficiency while ensuring convergence stability. Each iteration uses the full training data to calculate the gradient and update the parameters. Training iterations continue until the loss value is lower than the preset loss threshold or the number of training rounds reaches the preset maximum number of rounds. The network weights and biases at this point are saved as the parameters of the completed friction adaptation model.

[0025] The initial friction coefficient obtained in step S13 and the ambient vacuum level obtained in step S11 are used as inputs to the friction adaptation model. The model performs forward calculation, sequentially passing through input layer weighted summation, first hidden layer weighted summation and hyperbolic tangent activation, second hidden layer weighted summation and hyperbolic tangent activation, and output layer weighted summation and linear activation, finally obtaining the vacuum influence coefficient. The vacuum influence coefficient is added to the initial friction coefficient, and the result is recorded as the corrected friction coefficient. This corrected friction coefficient reflects a quantitative estimate of the friction behavior of the threaded pair under a given vacuum level, but is limited by the physically reasonable range of the friction coefficient, requiring further boundary constraint processing.

[0026] The preset lower and upper limits of the friction coefficient are derived from extreme value analysis of the statistical distribution of historical actual friction coefficients. In all historical experimental friction coefficient data collected during the friction adaptation model training phase, the minimum observed value of the friction coefficient corresponding to each material is taken as the lower limit of the friction coefficient for that material, and the maximum observed value is taken as the upper limit of the friction coefficient for that material. These upper and lower limits are stored in the boundary constraint parameter table, indexed by the material. During step S13, the corresponding lower and upper limits of the friction coefficient are retrieved from the boundary constraint parameter table based on the current screw material. When the corrected friction coefficient is greater than the lower limit and less than the upper limit, the corrected friction coefficient is directly output as the verification friction coefficient. When the corrected friction coefficient is not greater than the lower limit, it indicates that the friction coefficient after vacuum correction is below the physically observable range, and the lower limit is output as the verification friction coefficient. When the corrected friction coefficient is not less than the upper limit, it indicates that the corrected friction coefficient exceeds the physically observable range, and the upper limit is output as the verification friction coefficient. The verified friction coefficient after boundary constraints serves as the basis for calculating torque requirements in subsequent steps S14 and S15, ensuring that torque planning is always based on friction characteristic estimations that conform to actual physical laws.

[0027] This step combines the inherent frictional properties of materials with the dynamic changes in frictional behavior under ultra-high vacuum conditions. It quantifies the nonlinear influence of the ambient vacuum level on the friction coefficient through a neural network model and reasonably constrains the statistical boundaries based on historical experimental data to obtain a verified friction coefficient. This provides a friction parameter benchmark that conforms to actual vacuum conditions for subsequent torque distribution and graded compensation, fundamentally improving the accuracy of tightening torque planning.

[0028] In step S14, priority weight allocation is calculated based on the corrected position coordinates and the verified friction coefficient to obtain the allocation ratio coefficient, including: Based on the corrected position coordinates, the Euclidean distance between adjacent screws is calculated as the center distance. Screws with a center distance lower than a preset distance threshold are selected to obtain a list of close-range screws. The vibration frequency of the screw is obtained, and the screws with vibration frequencies higher than a preset vibration frequency threshold are selected to obtain a list of high-frequency vibration screws. Based on the verified friction coefficient, the basic priority value of each screw is calculated using a preset priority formula; Add the basic priority value corresponding to the list of close-range screws to the preset priority amount, and add the basic priority value corresponding to the list of high-frequency vibration screws to the priority amount to obtain the adjustment priority value; All screws are sorted from high to low according to the aforementioned adjustment priority value to obtain a tightening sequence; The adjustment priority values ​​of all screws are added together to obtain a total priority value, and the adjustment priority value is divided by the total priority value to obtain the allocation ratio coefficient of the corresponding screw.

[0029] Specifically, firstly, based on the corrected position coordinates, the Euclidean distance between each screw and its nearest spatial neighbor is calculated. The calculation method is to take the square root of the sum of the squares of the differences in the coordinates of the two screws in the three dimensions (horizontal, vertical, and vertical), and the resulting value is recorded as the center distance. The center distances of all screws are then compared with a preset distance threshold, which is determined based on the simulation analysis results of the deformation of the sealing surface of the same type of cavity flange. During the process design phase, finite element analysis software is used to simulate the contact stress distribution of the flange sealing surface under tightening load, extracting the critical screw spacing value corresponding to areas with significant stress gradient changes. This critical value is then fixed as the distance threshold in the system configuration parameters. Screws with center distances lower than the distance threshold indicate that they are in a densely distributed area in space. During tightening, the elastic interaction between adjacent screws will have a significant coupling effect on the preload. These screws are included in the list of nearby screws.

[0030] The process of obtaining the screw vibration frequency involves integrating an accelerometer at the front end of the tightening shaft in the tightening system. When the tightening shaft rotates unloaded, the sensor collects radial vibration signals from the shaft end. A Fast Fourier Transform (FFT) is performed on the vibration signals to obtain the spectral distribution, and the frequency value corresponding to the highest amplitude point in the spectrum is extracted as the vibration frequency at that screw location. The preset vibration frequency threshold is determined through historical unloaded vibration spectrum statistics. The statistical process collects vibration spectrum data from multiple batches of cavity screw holes under the same rotational speed conditions using the same type of tightening shaft. The maximum vibration frequencies of each hole location are summed, and the arithmetic mean is taken as the vibration frequency threshold and written into the system configuration file. Screws with vibration frequencies higher than the vibration frequency threshold indicate additional disturbances in the fit between the tightening shaft and the screw head at that hole location, making them more prone to torque fluctuations during tightening. These screws are included in the high-frequency vibration screw list.

[0031] A preset priority formula is used to calculate the basic priority value of each screw based on the verified friction coefficient. The priority formula is defined as follows: the basic priority value equals the verified friction coefficient multiplied by the friction weight factor plus a constant term offset. The friction weight factor and constant term offset are determined by linear regression fitting of the final stress deviation rate of screws with different friction coefficients in historical tightening cases. During the fitting process, the verified friction coefficient from multiple batches of tightening data is used as the independent variable, and the stress deviation rate of the corresponding screw after tightening is used as the dependent variable. The slope and intercept of the univariate linear regression equation are solved using the least squares method. The negative of the slope is taken as the friction weight factor, and the negative of the intercept plus a preset constant adjustment value is taken as the constant term offset. This ensures that the basic priority value decreases appropriately as the friction coefficient increases, reflecting the experience that screws with larger friction coefficients are more sensitive to torque response during tightening and should be tightened later. The constant adjustment value is set to 0.5, which is determined by comparing the stress uniformity simulation results under different constants, ensuring that the basic priority value maintains a reasonable distinction between different friction coefficients without excessive ranking fluctuations.

[0032] The basic priority value of screws in the near-field screw list is added to a preset priority value. Similarly, the basic priority value of screws in the high-frequency vibration screw list is added to the preset priority value to obtain the adjusted priority value. The preset priority value is set to a fixed integer value. This value is determined during the system process parameter calibration phase by constructing a co-simulation environment that includes a finite element simulation model of the tightening process and sequential tightening control logic. The finite element simulation model establishes a three-dimensional solid mesh based on the actual geometric dimensions of the cavity flange, assigning elastic-plastic constitutive parameters to the screw material and contact properties to the thread pair. The friction coefficient in the contact properties is set to a typical verification friction coefficient value selected from historical data. The co-simulation environment performs multiple iterative simulation runs with different priority values ​​as input variables. The priority value increases from zero with a preset step size. Each simulation run fully executes the control flow described in steps S11 to S18 and records the final non-uniformity value at the end of the simulation. The preset step size is set to one, meaning it increases by one unit value each time from zero. This step size corresponds to the smallest resolvable unit of the basic priority value, ensuring the exhaustiveness of the search process and preventing the omission of optimal values.

[0033] After iterative simulation, a dataset mapping the priority values ​​to the corresponding final non-uniformity is obtained. Analyzing the trend of final non-uniformity with increasing priority in this dataset reveals that when the priority value is small, the ranking improvement of close-range screws and high-frequency vibrating screws in the tightening queue is not significant, resulting in a high level of final non-uniformity. As the priority value gradually increases, the ranking positions of close-range screws and high-frequency vibrating screws gradually shift forward, and the final non-uniformity shows a decreasing trend. When the priority value increases to a certain critical value, further increases in priority value tend to saturate the effect on the ranking results, and the change in final non-uniformity is less than the preset convergence threshold. This critical value is written into the system configuration file for permanent storage; this critical value is typically an integer value between 10% and 30% of the maximum fluctuation range of the basic priority value. Once the priority amount is calibrated, it is directly called as a fixed constant during the execution of step S14 and does not change dynamically with each tightening task. The convergence judgment threshold is preset to be less than 0.5 percent of the final non-uniformity change. This threshold is set according to the process accuracy requirements. When the non-uniformity change is less than this value, it is considered that increasing the priority amount has no practical significance for improving uniformity.

[0034] All screws are sorted in descending order of their adjustment priority values. The resulting order is the tightening priority queue, and screws at the beginning of the queue will be tightened with priority in subsequent steps. The adjustment priority values ​​of all screws are summed to obtain a total priority value. For each screw, its adjustment priority value is divided by the total priority value. The quotient is the allocation ratio coefficient for that screw, which represents the torque allocation weight that screw should receive in the total torque requirement.

[0035] This step integrates the spatial positional relationship, vibration response characteristics, and frictional characteristics of screws into a unified priority quantification framework. By calculating the center distance and vibration frequency, it identifies individual screws significantly affected by spatial coupling and dynamic disturbances. Through priority formulas and priority adjustment mechanisms, it rationally increases the tightening priority of these screws, avoiding preload attenuation in already tightened screws due to subsequent tightening. The tightening sequence provides the order for generating the initial tightening command in step S15, and the allocation ratio coefficient provides a weighting basis for torque allocation, ensuring that total torque resources are allocated to each screw as needed during the serialized tightening process. This optimizes the load balance of the tightening process from both spatial temporal and torque ratio dimensions.

[0036] In step S15, an initial tightening command is generated based on the verified friction coefficient and the distribution ratio coefficient, and feedback is performed to calculate the torque deviation rate. Then, graded torque compensation processing is performed based on the torque deviation rate to obtain a depth compensation list, including: The original torque requirement is obtained by multiplying the verified friction coefficient by a preset torque reference constant. The original torque requirement for each screw is then summed to obtain the total torque requirement. Multiply the total torque requirement by the allocation ratio coefficient to obtain the allocated torque value, and output the tightening sequence and the allocated torque value as the initial tightening command. Execute the initial tightening command and obtain the actual torque value of the screw; The torque deviation rate is obtained by taking the absolute value of the difference between the actual torque value and the distributed torque value and then dividing it by the distributed torque value. When the torque deviation rate is lower than a preset first deviation threshold, the screw is determined to be tightened properly. When the torque deviation rate is not lower than the preset first deviation threshold and not higher than the preset second deviation threshold, the difference between the actual torque value and the allocated torque value is calculated to obtain the torque deviation value. The tightening order is based on the torque deviation rate from large to small, and the torque deviation value is used as the torque increment. The screw is iteratively tightened. After each tightening is completed, the torque deviation rate is recalculated until it is lower than the first deviation threshold. When the torque deviation rate is higher than the preset second deviation threshold, the corresponding screws will be output as a depth compensation list.

[0037] Specifically, the verification friction coefficient is first multiplied by a preset torque reference constant to obtain the initial torque requirement. This torque reference constant is obtained by conducting multiple calibration experiments in an ultra-high vacuum environment simulation chamber using screws of different materials on the same type of cavity flange. The verification friction coefficient and the actual torque value required to achieve the target preload value are recorded in each experiment. The ratio of the actual torque value to the verification friction coefficient is used as the sample value of the torque reference constant. The arithmetic mean of all valid sample values ​​is taken as the torque reference constant and written into the system configuration file. Since the verification friction coefficient is dimensionless, the torque reference constant is measured in torque units, thus the initial torque requirement has torque dimensions. After performing the above multiplication operation on each screw to obtain its own initial torque requirement, the initial torque requirements of all screws are summed to obtain the total torque requirement.

[0038] The total torque requirement is multiplied by the allocation ratio coefficient corresponding to each screw, and the product is the allocated torque value of each screw. The allocated torque value represents the torque share that each screw should receive according to its priority weight under the current working condition. The tightening sequence queue generated in step S14 is combined with the allocated torque value of each screw into a structured data frame, which is sent to each tightening axis controller via fieldbus. This data frame is the initial tightening command. Each tightening axis controller drives the servo motor to output torque in sequence according to the order in the tightening sequence, and collects torque feedback signals in real time through dynamic torque sensors installed at the output end of the tightening axis during the tightening process. Loading stops when the torque feedback signal reaches the allocated torque value. After each screw is tightened, the system records the actual torque value of the screw. The actual torque value is the sampled value of the torque sensor at the end of the loading process.

[0039] The torque deviation rate is calculated by taking the absolute value of the difference between the actual torque value and the distributed torque value, and then dividing this absolute value by the distributed torque value. The torque deviation rate reflects the relative deviation between the actual applied torque and the target torque in a single tightening operation. The preset first and second deviation thresholds are determined by statistically analyzing the torque deviation rate distribution of historical successful tightening cases. During the process verification phase before system deployment, historical torque deviation rate data for each screw of the same type of cavity flange under standard tightening procedures are collected. All historical torque deviation rate samples are sorted from smallest to largest. The value at the 90th percentile after sorting is taken as the first deviation threshold, and the value at the 95th percentile after sorting is taken as the second deviation threshold. Both thresholds are written into the system configuration file and called as fixed constants during operation. Using the percentile method to determine the deviation thresholds effectively distinguishes between minor deviations acceptable under normal operating conditions and significant deviations requiring special attention.

[0040] When the torque deviation rate is below the first deviation threshold, it indicates that the deviation between the actual torque and the distributed torque value is within the normal fluctuation range, and the tightening quality of the screw is deemed qualified. The system then directly proceeds to the tightening process of the next screw. When the torque deviation rate is not lower than the first deviation threshold and not higher than the second deviation threshold, it indicates that the screw has a moderate degree of torque deviation. In this case, the difference between the actual torque value and the distributed torque value is calculated as the torque deviation value, which is marked with a positive or negative sign to indicate the direction of insufficient or overshoot torque. For this type of screw, the tightening order is rearranged from largest to smallest based on the torque deviation rate, with screws having larger torque deviation rates being tightened first. Each tightening operation applies the torque deviation value as a torque increment to the original tightening operation. After each tightening operation, the actual torque value is re-acquired and the torque deviation rate is recalculated. If it is still not lower than the first deviation threshold, tightening continues until the torque deviation rate is lower than the first deviation threshold. The cyclic process sets a maximum number of tightening attempts to prevent infinite loops. The maximum number of tightening attempts is preset to three. This value is based on the statistical results of a large number of calibration experiments. Experiments show that screws with torque deviation rates between the first and second thresholds can converge to the qualified range after three iterations of tightening.

[0041] It should be noted that after each tightening operation is completed, the actual torque value of the screw is obtained again, and the difference between the actual torque value and the allocated torque value is recalculated to obtain an updated torque deviation value. Subsequent tightening operations will continue to be performed with this updated torque deviation value as the torque increment until the torque deviation rate is lower than the first deviation threshold.

[0042] When the torque deviation rate exceeds the second deviation threshold, it indicates that the screw has a serious torque deviation. Simply tightening it once using the torque deviation value is unlikely to converge it to the acceptable range within a reasonable number of tightening cycles, and there may be abnormal factors such as thread jamming, foreign object embedding, or sensor malfunction. At this point, the screw's identification information, allocated torque value, actual torque value, and torque deviation rate are packaged and output as a depth compensation list. This depth compensation list serves as input to step S16, triggering a deeper torque compensation mechanism.

[0043] This step divides torque deviation into three levels: slightly acceptable, moderately compensable, and severely requiring deep compensation, by setting two levels of deviation thresholds. Different strategies are adopted for different levels: no processing required, single-time tightening iteration, and transition to deep compensation. This avoids the drawbacks of slight deviations consuming too many computing resources and severe deviations not being adequately processed, and realizes graded and refined adjustment of torque closed-loop control.

[0044] In step S16, torque depth compensation is performed according to the depth compensation list and the torque deviation rate to obtain a compensated tightening command, including: Based on the depth compensation list and the torque deviation rate, torque compensation is performed using a preset torque compensation function to obtain the compensated torque value; The depth compensation list is reordered from largest to smallest according to the torque deviation rate to obtain the updated tightening sequence; The compensated torque value and the updated tightening sequence are output as a compensated tightening command.

[0045] Specifically, the compensation torque value is first calculated for each screw in the depth compensation list using a preset torque compensation function. The preset torque compensation function adopts a piecewise linear compensation form, and its construction process is as follows: During the system calibration phase, case data of severe torque deviations occurring during the tightening and debugging of the same type of cavity flange are collected. The torque deviation rate of each case and the actual compensation torque value required to achieve the qualified preload force for the screw are recorded. The collected torque deviation rate and corresponding compensation torque value data are sorted by torque deviation rate from smallest to largest, dividing them into several deviation rate intervals. Within each interval, a linear regression method is used to fit the slope and intercept between the torque deviation rate and the compensation torque value. Specifically, for each sample point within an interval, with the torque deviation rate as the independent variable and the compensation torque value as the dependent variable, the slope and intercept parameters of the univariate linear equation are solved using the least squares method to minimize the sum of squares of the differences between the predicted and actual compensation torque values ​​for each sample point within that interval. The slope and intercept parameters of each interval, as well as the interval boundary values, are stored in the system configuration file in the form of a parameter table. During step S16, the torque compensation function reads the current screw's torque deviation rate, determines which deviation rate interval it falls into, calls the slope and intercept parameters corresponding to that interval, multiplies the torque deviation rate by the slope, and adds the intercept; the result is the compensated torque value. If the torque deviation rate exceeds the maximum deviation rate interval covered by historical samples, the parameters of the last interval are called for extrapolation calculation.

[0046] After obtaining the compensated torque value for each screw, the screw records in the depth compensation list are sorted in descending order of torque deviation rate, with the screw with the largest torque deviation rate placed at the beginning of the sequence, and so on down. The resulting list is called the updated tightening sequence. The compensated torque value of each screw in the updated tightening sequence is combined with its corresponding identifier to form structured instruction data, which is sent to the corresponding tightening axis controller via the fieldbus. This data frame is the compensated tightening instruction. The compensated tightening instruction replaces the target torque value of the corresponding screw in the original initial tightening instruction. In subsequent execution stages, the tightening axis controller loads the compensated torque values ​​sequentially according to the order specified in the updated tightening sequence.

[0047] This step introduces a piecewise linear compensation function fitted based on historical compensation data. This eliminates the reliance on simple deviation scaling when determining the compensation torque value. Instead, it uses targeted compensation based on the statistical patterns of severe deviations in actual working conditions, improving the efficiency of correcting severely deviated screws. Arranging the compensation tightening sequence in descending order of torque deviation rate, prioritizing the screws with the largest deviations, avoids further disturbance of the preload of surrounding compensated screws due to post-processing of severely deviated screws, thus enhancing the convergence stability of the deep compensation process.

[0048] In step S17, the compensation tightening command is executed and stress safety status identification is performed to obtain a stress safety list and a stress anomaly list. The stress anomaly list is then subjected to stress deviation quantification to calculate an initial stress deviation rate. This initial stress deviation rate is used for stress compensation tightening iterative optimization to obtain the final stress value, including: Execute the compensation tightening command to obtain the stress value around the screw; When the surrounding stress value is not lower than the preset lower limit of stress safety and not higher than the preset upper limit of stress safety, the corresponding screw stress safety is marked, and a stress safety list is obtained. When the surrounding stress value is lower than the preset lower limit of stress safety or higher than the preset upper limit of stress safety, the corresponding screw stress is marked as abnormal, and a list of stress anomalies is obtained. The arithmetic mean of the surrounding stress values ​​corresponding to the stress safety list is calculated as the average stress value. According to the stress anomaly list, the difference between the surrounding stress value and the average stress value is divided by the average stress value to obtain the initial stress deviation rate. Multiply the initial stress deviation rate by the compensation torque value to obtain the stress-optimized torque value; Using the initial stress deviation rate from largest to smallest as the tightening order, and the stress-optimized torque value as the torque increment, the screws are iteratively tightened, and the peripheral stress value is reacquired until the peripheral stress value is not lower than the preset lower limit of stress safety and not higher than the preset upper limit of stress safety. The peripheral stress value is then output as the final stress value.

[0049] Specifically, each tightening shaft controller first loads the compensation torque value sequentially according to the updated tightening sequence, completing the torque application to the screws in the depth compensation list. After all compensation tightening commands are executed, the system acquires stress signals through a thin-film strain gauge array embedded near the screw holes below the sealing surface of the cavity flange. The thin-film strain gauges are connected to the signal conditioning circuit in the form of a Wheatstone bridge. The bridge output voltage is differentially amplified by an instrumentation amplifier and then converted into a digital quantity by an analog-to-digital converter. Based on the factory calibration parameters of the strain gauges, the voltage value is then converted into a peripheral stress value in Pascals, with one peripheral stress value obtained for each screw.

[0050] The process for determining the preset lower and upper limits of stress safety is as follows: During the design phase of the ultra-high vacuum cavity, based on the yield strength and fatigue limit data of the sealing surface material under vacuum conditions, combined with the stress distribution range extracted from the finite element simulation analysis of the flange sealing contact, the minimum compressive stress value required to ensure sealing performance is determined as the lower limit of stress safety, and the maximum compressive stress value that will not cause micro-plastic deformation of the material or damage to the sealing surface is determined as the upper limit of stress safety. Both thresholds are written into the configuration file as fixed constants during system deployment. The system reads the surrounding stress value of each screw and compares it with the thresholds. When the surrounding stress value is not lower than the lower limit of stress safety and not higher than the upper limit of stress safety, the screw is marked as stress-safe and added to the stress-safe list. When the surrounding stress value is lower than the lower limit of stress safety or higher than the upper limit of stress safety, the screw and its surrounding stress value are marked as stress-abnormal and added to the stress-abnormal list.

[0051] The sum of the peripheral stress values ​​of all screws in the stress safety list is divided by the number of screws in the stress safety list. The quotient is recorded as the average stress value, which represents the stress level benchmark of the current stress safety screw group. For each screw in the stress anomaly list, the difference between its peripheral stress value and the average stress value is calculated, and then the difference is divided by the average stress value. The quotient is recorded as the initial stress deviation rate. The initial stress deviation rate can be positive or negative. A positive value indicates that the peripheral stress value is higher than the average stress value, and a negative value indicates that the peripheral stress value is lower than the average stress value. The initial stress deviation rate of each stress anomaly screw is multiplied by the compensation torque value corresponding to that screw in step S16. The product is recorded as the stress optimization torque value, which serves as the torque increment for subsequent tightening operations.

[0052] It should be noted that if the stress safety list is empty, that is, the stress values ​​around all screws do not meet the upper and lower limits of stress safety, then the average stress value of the group of safety screws cannot be used as a benchmark. In this case, the arithmetic mean of the stress values ​​around all screws is calculated as the average stress value, and the initial stress deviation rate of each screw is calculated based on the average stress value. The subsequent iterative optimization process remains unchanged.

[0053] The stress-compensated tightening iterative optimization process uses a stress anomaly list as the operating object. First, screws in the list are sorted in descending order of the absolute value of their initial stress deviation rate; the larger the absolute value, the more severe the stress deviation, and these are processed first. The tightening shaft controller performs a single tightening operation on the screws in this order, using the stress-optimized torque value as the torque increment. The direction of the torque increment is determined by the sign of the initial stress deviation rate: when the deviation rate is positive, the torque increment is applied in the positive direction; when the deviation rate is negative, the torque increment is applied in the negative direction. After tightening, the peripheral stress value of the screw is re-acquired and compared with the lower and upper safety limits. If it is still outside the safe range, the stress-optimized torque value is recalculated based on the recalculated stress deviation rate, and tightening continues until the peripheral stress value falls between the lower and upper safety limits. At this point, the last acquired peripheral stress value is output as the final stress value of the screw. For screws in the stress safety list that already meet the safety range requirements in the first detection, their final stress value is directly taken from the initially acquired peripheral stress value.

[0054] This step extends the feedback parameters of tightening control from the torque level to the stress level. It distinguishes between screws that have met the stress safety standards and those with stress deviations by identifying the stress safety status. The deviation of abnormal screws is quantified based on the average stress value of the stress-safe screw group, achieving progressive control from torque closed-loop to stress closed-loop. The stress-compensated tightening iterative optimization prioritizes correcting the screws with the largest deviations and determines the torque increment based on the proportion of the compensation torque value. This ensures that each iterative adjustment has a clear physical basis and convergence direction, avoiding stress oscillations caused by blind trial and error. The final stress value, as the result after all screw stress states have converged to the safe range, provides iteratively optimized stress distribution data for the stress uniformity quantification assessment in step S18.

[0055] In step S18, based on the final stress value and the initial stress deviation rate, a quantitative assessment of stress uniformity is performed to obtain an optimization report, including: Obtain the torque applied to the screw and the final torque value; The arithmetic mean of the final stress values ​​is calculated as the final average stress value. Then, the absolute value of the difference between the final stress value and the final average stress value of each screw is calculated and divided by the final average stress value to obtain the final stress deviation rate of each screw. The absolute values ​​of all the final stress deviation rates are summed and divided by the total number of screws to obtain the final non-uniformity. The output includes an optimization report containing the final torque application record, the final torque value, and the percentage improvement in uniformity; The calculation process for the uniformity improvement percentage is as follows: sum the absolute values ​​of the initial stress deviation rates of each screw and divide by the total number of screws to obtain the initial non-uniformity; subtract the final non-uniformity from the initial non-uniformity and divide by the initial non-uniformity; then multiply by 100% to obtain the uniformity improvement percentage.

[0056] Specifically, the final stress value is derived from the stress values ​​around each screw output after the stress compensation tightening iteration optimization in step S17. The initial stress deviation rate is derived from the stress deviation quantification value calculated for each screw with abnormal stress after the first stress safety state identification in step S17. The torque application record and the final torque value are extracted from the data log module of the tightening shaft controller. The torque application record includes the timestamp, target torque value, actual torque value, and loading direction of each loading operation for each screw throughout the entire tightening process. The final torque value is the actual torque value reached by the screw in the last torque loading operation.

[0057] The final average stress value is calculated by summing the final stress values ​​of all screws and dividing the sum by the total number of screws. This quotient reflects the overall stress level around each screw hole on the sealing surface after a complete tightening control process. For each screw, the difference between its final stress value and the final average stress value is calculated, and the absolute value is taken. This absolute value is then divided by the final average stress value, and the quotient is recorded as the final stress deviation rate. The final stress deviation rate indicates the degree of deviation of a single screw from the overall average level under its final stress state. The absolute values ​​of the final stress deviation rates of all screws are summed and divided by the total number of screws. The quotient is recorded as the final non-uniformity. The smaller the final non-uniformity value, the more uniform the stress distribution at each point on the sealing surface.

[0058] The initial non-uniformity is calculated as follows: The initial stress deviation rate of each screw in the stress anomaly list from step S17 is taken. Since the screws in the stress safety list already meet the safety range requirements during initial testing and are not assigned an initial stress deviation rate value, their initial stress deviation rate is calculated as zero. The absolute values ​​of the initial stress deviation rates corresponding to all screws are summed, and then divided by the total number of screws. The quotient is recorded as the initial non-uniformity. The initial non-uniformity characterizes the degree of non-uniformity of the stress distribution on the sealing surface before the stress compensation tightening iteration optimization is executed. If the calculated initial non-uniformity is 0, it indicates that the initial stress is completely uniform. In this case, the uniformity improvement percentage is defined as 0, and this is noted in the optimization report.

[0059] The calculation process for the uniformity improvement percentage is as follows: subtract the final non-uniformity from the initial non-uniformity to obtain the difference, then divide the difference by the initial non-uniformity and multiply by 100% to obtain the uniformity improvement percentage. When the value is positive, it indicates that the stress uniformity has been improved, and the larger the value, the more significant the improvement.

[0060] The optimization report is output in structured text format to the human-computer interaction interface and stored in the historical database. The report includes the screw identification number sequence, the torque application record for each screw, the final torque value for each screw, the final stress value for each screw, the final average stress value, the final non-uniformity, and the percentage improvement in uniformity. The report concludes with the numerical value of the uniformity improvement percentage and its positive or negative sign.

[0061] This step, by introducing a quantitative comparison between initial and final non-uniformity, transforms the improvement effect of the entire tightening control process on stress uniformity into a single, comparable percentage indicator. This provides an objective, quantitative evaluation basis for quality traceability and iterative optimization of process parameters in serialized screw tightening processes under ultra-high vacuum conditions. The optimization report allows operators to intuitively grasp the stress distribution quality of this tightening operation and provides data support for subsequent process improvements.

[0062] Reference Figure 2 The second embodiment of the present invention provides an intelligent control system for a serialized screw tightening process under ultra-high vacuum conditions, comprising: The data acquisition module is used to acquire the screw position coordinates, screw material, and ambient vacuum level; The position correction module is used to perform spacing constraint position correction based on the screw position coordinates to obtain the corrected position coordinates; The vacuum influence module is used to match the initial friction coefficient according to the screw material, and to perform vacuum influence correction on the initial friction coefficient using the ambient vacuum level to obtain the verification friction coefficient. The allocation calculation module is used to perform priority weight allocation calculation based on the corrected position coordinates and the verified friction coefficient to obtain the allocation ratio coefficient; The graded processing module is used to generate an initial tightening command based on the verified friction coefficient and the distribution ratio coefficient, perform feedback calculation to determine the torque deviation rate, and perform graded torque compensation processing based on the torque deviation rate to obtain a depth compensation list. The torque compensation module is used to perform torque depth compensation based on the depth compensation list and the torque deviation rate to obtain a compensation tightening command; The stress optimization module is used to execute the compensation tightening command and perform stress safety status identification to obtain a stress safety list and a stress anomaly list. It then performs stress deviation quantification on the stress anomaly list to calculate the initial stress deviation rate and uses the initial stress deviation rate to perform stress compensation tightening iterative optimization to obtain the final stress value. The output module is used to perform a quantitative evaluation of stress uniformity based on the final stress value and the initial stress deviation rate, and to obtain an optimization report.

[0063] It should be noted that the intelligent control system for serialized screw tightening process under ultra-high vacuum environment provided in the embodiments of the present invention is used to execute all process steps of the intelligent control method for serialized screw tightening process under ultra-high vacuum environment in the above embodiments. The working principles and beneficial effects of the two are one-to-one, so they will not be described again.

[0064] It should be noted that the system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0065] 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 scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for intelligent control of a serialized screw tightening process under ultra-high vacuum conditions, characterized in that, include: Obtain the screw's location coordinates, screw material, and ambient vacuum level; Based on the screw position coordinates, perform spacing constraint position correction to obtain the corrected position coordinates; The initial friction coefficient is matched according to the screw material, and the initial friction coefficient is corrected for vacuum influence using the ambient vacuum level to obtain the verified friction coefficient. Based on the corrected position coordinates and the verified friction coefficient, priority weight allocation calculation is performed to obtain the allocation ratio coefficient; An initial tightening command is generated based on the verified friction coefficient and the distribution ratio coefficient, and feedback is performed to calculate the torque deviation rate. Then, a graded torque compensation process is performed based on the torque deviation rate to obtain a depth compensation list. Based on the depth compensation list and the torque deviation rate, torque depth compensation is performed to obtain a compensated tightening command; The compensation tightening command is executed and stress safety status is identified to obtain a stress safety list and a stress anomaly list. The stress deviation of the stress anomaly list is quantified to calculate the initial stress deviation rate. The initial stress deviation rate is used to perform stress compensation tightening iterative optimization to obtain the final stress value. Based on the final stress value and the initial stress deviation rate, a quantitative assessment of stress uniformity is performed to obtain an optimization report.

2. The intelligent control method for serialized screw tightening process under ultra-high vacuum environment according to claim 1, characterized in that, The step of performing spacing constraint position correction based on the screw position coordinates to obtain the corrected position coordinates includes: Based on the screw position coordinates, the Euclidean distance between adjacent screws is calculated as the spacing between adjacent screws; When the spacing between adjacent screws is not less than the preset minimum spacing, the screw position coordinates are output as correction position coordinates; When the distance between adjacent screws is lower than the preset minimum distance, the scanning angle is adjusted to reacquire the screw position coordinates and the distance between adjacent screws is recalculated until the distance between adjacent screws is not lower than the minimum distance. Then, the screw position coordinates are output as the corrected position coordinates.

3. The intelligent control method for serialized screw tightening process under ultra-high vacuum environment according to claim 1, characterized in that, The process of matching an initial friction coefficient based on the screw material and correcting for vacuum effects on the initial friction coefficient using the ambient vacuum level to obtain a verified friction coefficient includes: Based on the screw material, an initial friction coefficient is obtained by linearly mapping it to a pre-stored material friction coefficient library. Based on the initial friction coefficient and the ambient vacuum level, the influence of vacuum level on the friction coefficient is quantified using a pre-constructed friction adaptation model to obtain the vacuum influence coefficient; The vacuum influence coefficient is added to the initial friction coefficient to obtain the corrected friction coefficient; When the corrected friction coefficient is greater than the preset lower limit of the friction coefficient and less than the preset upper limit of the friction coefficient, the corrected friction coefficient is output as the verification friction coefficient. When the corrected friction coefficient is not greater than the preset lower limit of the friction coefficient, the lower limit of the friction coefficient is output as the verification friction coefficient; when the corrected friction coefficient is not less than the preset upper limit of the friction coefficient, the upper limit of the friction coefficient is output as the verification friction coefficient.

4. The intelligent control method for serialized screw tightening process under ultra-high vacuum environment according to claim 3, characterized in that, The process of constructing the friction adaptation model includes: Obtain historical initial friction coefficient, historical environmental vacuum level, and historical vacuum influence coefficient; The historical initial friction coefficient and historical environmental vacuum level are input into the input layer of the initially constructed neural network model for training, and the predicted value of the vacuum influence coefficient output by the output layer of the neural network model is obtained. Substitute the predicted value of the vacuum influence coefficient and the historical vacuum influence coefficient into the loss function to calculate the loss value; The gradient of the output layer of the neural network model is calculated based on the loss value, and the gradient is passed forward layer by layer through the chain rule to calculate the gradient of the parameters of each layer and obtain the gradient data. Based on the gradient data and the preset learning rate, the parameters of each layer of the neural network model are updated using the gradient descent method. The parameters of each layer are iteratively updated until the number of training iterations of the neural network model is greater than a preset number of iterations, or the loss value of the neural network model is less than a preset loss threshold. At this point, the training is considered complete, and a trained friction adaptation model is obtained.

5. The intelligent control method for serialized screw tightening process under ultra-high vacuum environment according to claim 1, characterized in that, The step of calculating the priority weight allocation based on the corrected position coordinates and the verified friction coefficient to obtain the allocation ratio coefficient includes: Based on the corrected position coordinates, the Euclidean distance between adjacent screws is calculated as the center distance. Screws with a center distance lower than a preset distance threshold are selected to obtain a list of close-range screws. The vibration frequency of the screw is obtained, and the screws with vibration frequencies higher than a preset vibration frequency threshold are selected to obtain a list of high-frequency vibration screws. Based on the verified friction coefficient, the basic priority value of each screw is calculated using a preset priority formula; Add the basic priority value corresponding to the list of close-range screws to the preset priority amount, and add the basic priority value corresponding to the list of high-frequency vibration screws to the priority amount to obtain the adjustment priority value; Sort all screws from high to low according to the aforementioned adjustment priority value to obtain a tightening sequence; The adjustment priority values ​​of all screws are added together to obtain a total priority value, and the adjustment priority value is divided by the total priority value to obtain the allocation ratio coefficient of the corresponding screw.

6. The intelligent control method for serialized screw tightening process under ultra-high vacuum environment according to claim 5, characterized in that, The process involves generating an initial tightening command based on the verified friction coefficient and the distribution ratio coefficient, performing feedback calculation to determine the torque deviation rate, and then performing graded torque compensation processing based on the torque deviation rate to obtain a depth compensation list, including: The original torque requirement is obtained by multiplying the verified friction coefficient by a preset torque reference constant. The original torque requirement of each screw is then summed to obtain the total torque requirement. Multiply the total torque requirement by the allocation ratio coefficient to obtain the allocated torque value, and output the tightening sequence and the allocated torque value as the initial tightening command. Execute the initial tightening command and obtain the actual torque value of the screw; The torque deviation rate is obtained by taking the absolute value of the difference between the actual torque value and the distributed torque value and then dividing it by the distributed torque value. When the torque deviation rate is lower than a preset first deviation threshold, the screw is determined to be tightened properly. When the torque deviation rate is not lower than the preset first deviation threshold and not higher than the preset second deviation threshold, the difference between the actual torque value and the allocated torque value is calculated to obtain the torque deviation value. The tightening order is based on the torque deviation rate from large to small, and the torque deviation value is used as the torque increment. The screw is iteratively tightened. After each tightening is completed, the torque deviation rate is recalculated until it is lower than the first deviation threshold. When the torque deviation rate is higher than the preset second deviation threshold, the corresponding screws will be output as a depth compensation list.

7. The intelligent control method for serialized screw tightening process under ultra-high vacuum environment according to claim 6, characterized in that, The step of performing torque depth compensation based on the depth compensation list and the torque deviation rate to obtain a compensated tightening command includes: Based on the depth compensation list and the torque deviation rate, torque compensation is performed using a preset torque compensation function to obtain the compensated torque value; The depth compensation list is reordered from largest to smallest according to the torque deviation rate to obtain the updated tightening sequence; The compensated torque value and the updated tightening sequence are output as a compensated tightening command.

8. The intelligent control method for serialized screw tightening process under ultra-high vacuum environment according to claim 7, characterized in that, The process involves executing the compensation tightening command and identifying the stress safety status to obtain a stress safety list and a stress anomaly list. The stress anomaly list is then subjected to stress deviation quantification to calculate an initial stress deviation rate. This initial stress deviation rate is used for iterative optimization of the stress compensation tightening process to obtain the final stress value, including: Execute the compensation tightening command to obtain the stress value around the screw; When the surrounding stress value is not lower than the preset lower limit of stress safety and not higher than the preset upper limit of stress safety, the corresponding screw stress safety is marked, and a stress safety list is obtained. When the surrounding stress value is lower than the preset lower limit of stress safety or higher than the preset upper limit of stress safety, the corresponding screw stress is marked as abnormal, and a list of stress anomalies is obtained. The arithmetic mean of the surrounding stress values ​​corresponding to the stress safety list is calculated as the average stress value. According to the stress anomaly list, the difference between the surrounding stress value and the average stress value is divided by the average stress value to obtain the initial stress deviation rate. Multiply the initial stress deviation rate by the compensation torque value to obtain the stress-optimized torque value; Using the initial stress deviation rate from largest to smallest as the tightening order, and the stress-optimized torque value as the torque increment, the screws are iteratively tightened, and the peripheral stress value is reacquired until the peripheral stress value is not lower than the preset lower limit of stress safety and not higher than the preset upper limit of stress safety. The peripheral stress value is then output as the final stress value.

9. The intelligent control method for serialized screw tightening process under ultra-high vacuum environment according to claim 1, characterized in that, The step of performing a quantitative assessment of stress uniformity based on the final stress value and the initial stress deviation rate to obtain an optimization report includes: Obtain the torque applied to the screw and the final torque value; The arithmetic mean of the final stress values ​​is calculated as the final average stress value. Then, the absolute value of the difference between the final stress value and the final average stress value of each screw is calculated and divided by the final average stress value to obtain the final stress deviation rate of each screw. The absolute values ​​of all the final stress deviation rates are summed and divided by the total number of screws to obtain the final non-uniformity. The output includes an optimization report containing the final torque application record, the final torque value, and the percentage improvement in uniformity; The calculation process for the uniformity improvement percentage is as follows: sum the absolute values ​​of the initial stress deviation rates of each screw and divide by the total number of screws to obtain the initial non-uniformity; subtract the final non-uniformity from the initial non-uniformity and divide by the initial non-uniformity; then multiply by 100% to obtain the uniformity improvement percentage.

10. An intelligent control system for a serialized screw tightening process under ultra-high vacuum environment, characterized in that, include: The data acquisition module is used to acquire the screw position coordinates, screw material, and ambient vacuum level; The position correction module is used to perform spacing constraint position correction based on the screw position coordinates to obtain the corrected position coordinates; The vacuum influence module is used to match the initial friction coefficient according to the screw material, and to perform vacuum influence correction on the initial friction coefficient using the ambient vacuum level to obtain the verification friction coefficient. The allocation calculation module is used to perform priority weight allocation calculation based on the corrected position coordinates and the verified friction coefficient to obtain the allocation ratio coefficient; The graded processing module is used to generate an initial tightening command based on the verified friction coefficient and the distribution ratio coefficient, perform feedback calculation to determine the torque deviation rate, and perform graded torque compensation processing based on the torque deviation rate to obtain a depth compensation list. The torque compensation module is used to perform torque depth compensation based on the depth compensation list and the torque deviation rate to obtain a compensated tightening command. The stress optimization module is used to execute the compensation tightening command and perform stress safety status identification to obtain a stress safety list and a stress anomaly list. It then performs stress deviation quantification on the stress anomaly list to calculate the initial stress deviation rate and uses the initial stress deviation rate to perform stress compensation tightening iterative optimization to obtain the final stress value. The output module is used to perform a quantitative evaluation of stress uniformity based on the final stress value and the initial stress deviation rate, and to obtain an optimization report.

Images