Assembly height measurement and evaluation method and system, electronic equipment and storage medium

Abstract

The invention discloses an assembly height measurement and evaluation method and system, electronic equipment and a storage medium, and relates to the technical field of data processing, and the method comprises the steps: obtaining a signal curve graph representing a screw assembly time sequence signal, and carrying out the multi-dimensional feature extraction to obtain a first multi-dimensional feature; a pre-trained first model is utilized, a first height prediction value of the screw is predicted according to the first multi-dimensional features, and the first model comprises two different basic models used for predicting the first multi-dimensional features and a meta-model used for predicting according to prediction results of the two different basic models; in response to the fact that the first height prediction value is not matched with a preset target height value, it is determined that the assembly height of the screw is abnormal, and a preset correction instruction is output to correct the screw assembly parameter of the next screw; and in response to matching of the first height prediction value and a preset target height value, determining that the assembly height of the screw is qualified. According to the invention, the accuracy of measuring and evaluating the screw assembly height can be improved, and the cost is reduced.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to assembly height measurement and evaluation methods, systems, electronic devices, and storage media. Background Technology

[0002] In screw assembly production lines, to ensure that the screw assembly (such as screw fastening) height meets the process requirements (such as avoiding floating locks, stripped threads, or overpressure), a rear-mounted laser height measuring station is commonly used for 100% inspection. However, this method has certain drawbacks. For example, it requires the use of a professional laser displacement sensor with a servo moving mechanism, which results in high hardware costs and occupies workstation space in the screw assembly production line, leading to high space costs and making it unfriendly to the layout of compact screw assembly production lines.

[0003] Therefore, how to improve the accuracy of screw assembly height measurement and evaluation while reducing costs has become an urgent problem to be solved.

[0004] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this application is to provide a method, system, electronic device, and storage medium for measuring and evaluating assembly height, aiming to solve the technical problem of improving the accuracy of screw assembly height measurement and evaluation while reducing costs.

[0006] To achieve the above objectives, this application proposes an assembly height measurement and evaluation method, which includes the following steps: Obtain the signal curve diagram representing the screw assembly timing signal, and perform multi-dimensional feature extraction based on the signal curve diagram to obtain the first multi-dimensional feature. Using a pre-trained first model, the first height of the screw is predicted based on the first multidimensional features. The first model includes two different base models for predicting the first multidimensional features, and a meta-model for making predictions based on the prediction results of the two different base models. In response to the mismatch between the first height prediction value and the preset target height value, it is determined that the screw assembly height is abnormal, and a preset correction command is output to correct the screw assembly parameters of the next screw. In response to the matching of the first height prediction value and the preset target height value, the assembly height of the screw is determined to be qualified.

[0007] In one embodiment, the step of extracting multi-dimensional features based on the signal curve to obtain the first multi-dimensional feature includes: The second multidimensional feature of a single screw assembly corresponding to the signal curve is determined, wherein the second multidimensional feature includes mechanical dimension feature, kinematic dimension feature and material process coupling dimension feature; Calculate the SHAP value corresponding to each dimension feature in the second multidimensional feature, and select a preset number of different dimension features from the multidimensional features based on the SHAP value to obtain the first multidimensional feature.

[0008] In one embodiment, the mechanical dimension features include peak torque ratio, steady-state slope, and energy dissipation integral; the kinematic dimension features include rotational speed fluctuation entropy and revolution-time nonlinearity; and the material-process coupling dimension features include theoretical axial position and hysteresis area.

[0009] In one embodiment, the two distinct base models include a decision tree model and a linear model located at the first model architecture layer, and the meta-model includes a Ridge regression model located at the second model architecture layer. The steps for predicting the first height of a screw and its confidence interval using a pre-trained first model based on the first multidimensional features include: The first multidimensional feature is input into the pre-trained first model. The first multidimensional feature is predicted in parallel according to the decision tree model and the linear model, and the second and third height prediction values ​​of the screw are output respectively. The predicted values ​​of the second and third altitudes were fused using the Ridge regression model to obtain the predicted value of the first altitude.

[0010] In one embodiment, after the step of predicting the first height prediction value of the screw based on the first multidimensional features using a pre-trained first model, the method includes: Detect whether the absolute value of the difference between the first predicted height value and the preset target height value is greater than a preset difference threshold; If the difference exceeds the preset threshold, it is determined that the first height prediction value and the preset target height value do not match. If the difference is less than or equal to the preset difference threshold, then the first height prediction value is determined to match the preset target height value.

[0011] In one embodiment, after the step of obtaining the signal curve representing the screw assembly timing signal, the method includes: The signal curve is preprocessed, including a moving average filtering process to remove the first frequency jitter signal in the signal curve, a dynamic time warping process to eliminate random phase shifts in the signal curve based on a preset theoretical speed curve, and an abnormal segment removal process to eliminate abnormal segments in the signal curve. Based on the signal curve after data preprocessing, perform multi-dimensional feature extraction based on the signal curve to obtain the first multi-dimensional feature.

[0012] In one embodiment, the assembly height measurement and evaluation method further includes: Construct a training dataset containing at least one training data point, wherein the training data includes the historical multidimensional features of the screw and the historical assembly height value of the screw; The training data is input into the preset first model, and the training data is predicted by two different base models in the first model to obtain the fourth height prediction value and the fifth height prediction value. The fourth and fifth height prediction values ​​are input into the meta-model in the first model for fusion prediction to obtain the sixth height prediction value. The meta-model is a meta-model that has undergone weight and activation function quantization and is accelerated by the TensorRT engine. The predicted sixth height and the historical assembly height are input into the preset loss function to calculate the loss function value. The model parameters of the first model are updated based on the loss function value until the preset training termination condition is met, and the pre-trained first model is obtained.

[0013] Furthermore, to achieve the above objectives, this application also proposes an assembly height measurement and evaluation system. The assembly height measurement and evaluation system includes a data acquisition system and an edge system. The data acquisition system is used to acquire screw assembly timing signals and send the screw assembly timing signals to the edge system. The edge system includes: The acquisition module is used to acquire a signal curve representing the screw assembly timing signal, and to perform multi-dimensional feature extraction based on the signal curve to obtain the first multi-dimensional feature. A prediction model is used to predict the first height of a screw based on a first multidimensional feature using a pre-trained first model. The first model includes two different base models for predicting the first multidimensional feature and a meta-model for making predictions based on the prediction results of the two different base models. The exception module is used to respond to the mismatch between the first height prediction value and the preset target height value, determine that the screw assembly height is abnormal, and output a preset correction command to correct the screw assembly parameters of the next screw. The qualification module is used to determine that the screw assembly height is qualified in response to the matching of the first height prediction value and the preset target height value.

[0014] In addition, to achieve the above objectives, this application also proposes an electronic device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the assembly height measurement and evaluation method as described above.

[0015] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the assembly height measurement and evaluation method as described above.

[0016] In this application, a signal curve representing the screw assembly timing signal is obtained, and multi-dimensional features are extracted to obtain the first multi-dimensional feature. Then, a pre-trained first model is used to predict the first height of the screw based on the first multi-dimensional feature. Thus, when measuring and evaluating the screw assembly height, the multi-dimensional features of the screw assembly can be considered, and the advantages of the model's prediction accuracy can be comprehensively utilized, thereby improving the accuracy of the predicted first height of the screw assembly. In order to further improve the accuracy of the predicted first height, the first model can be set up to include two different base models for predicting the first multi-dimensional feature, and a meta-model for predicting based on the prediction results of the two different base models. Thus, multiple models can be combined to predict the screw assembly height from different perspectives, thereby improving the accuracy of the predicted first height. When the predicted first height does not match the preset target height value, the screw assembly height is determined to be abnormal, and a preset correction command is output to correct the screw assembly parameters of the next screw. Thus, abnormalities can be detected early without affecting the subsequent screw assembly process. When the predicted height value matches the preset target height value, the assembly height of the screw is determined to be qualified. This avoids the need to use a rear-mounted laser height measurement station for inspection, which would increase hardware and space costs. Furthermore, when predicting the screw assembly height, multi-dimensional features are comprehensively considered, and the advantages of the model are utilized to improve the accuracy of screw assembly height measurement and evaluation. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the first embodiment of the assembly height measurement and evaluation method of this application; Figure 2 This is a flowchart illustrating the second embodiment of the assembly height measurement and evaluation method in this application; Figure 3 This is a flowchart illustrating the third embodiment of the assembly height measurement and evaluation method in this application; Figure 4 This is a schematic diagram of the system modules of the height measurement and evaluation system of this application; Figure 5 This is a schematic diagram of the equipment structure of the hardware operating environment involved in the assembly height measurement and evaluation method in this application embodiment.

[0020] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0022] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0023] In high-density screw assembly scenarios such as automotive electronics and home appliances, to ensure that the screw assembly (e.g., taking fastening as an example) height meets process requirements (such as avoiding floating locks, stripped threads, or overpressure), a rear-mounted laser height measuring station is commonly used for 100% measurement. However, this approach has the following drawbacks: high hardware cost, with a single laser displacement sensor costing approximately 30,000-50,000 RMB and requiring a supporting servo movement mechanism; cycle time loss, as the laser height measuring station requires an additional 2-3 seconds of inspection time per screw, becoming a bottleneck in the screw assembly production line (e.g., a mobile phone motherboard requires fastening 30 screws, increasing the cycle time per unit by 60-90 seconds); space occupation, as a separate inspection station needs to be reserved for the laser height measuring station, which is not friendly to the layout of compact screw assembly production lines; and data silos, as the height measurement results of the laser height measuring station cannot be used to optimize screw machine parameters and form a closed-loop control.

[0024] Based on this, this embodiment provides a method for inferring screw assembly height through indirect parameters. However, this method has certain limitations. It only uses a single current parameter and does not incorporate speed, number of turns, or material characteristics, resulting in insufficient accuracy in predicting screw assembly height. Furthermore, it lacks an edge computing architecture and requires uploading to the cloud for processing, leading to high latency and failing to meet the real-time requirements of screw assembly production lines.

[0025] Therefore, to circumvent the aforementioned shortcomings, this application provides an assembly height measurement and evaluation method, consisting of a three-level closed loop: a data acquisition system (field acquisition subsystem), an edge system (such as an edge computing subsystem), and a twin interface. These three subsystems are connected in series via an industrial Ethernet network and use MQTT-TLS encrypted messages as the data exchange carrier. The field acquisition subsystem outputs timing signals such as raw torque / speed / number of revolutions. The edge computing subsystem completes the entire process of data preprocessing, feature engineering, virtual measurement, and closed-loop feedback, thus providing a predicted screw height value (i.e., the first predicted height value) and its corresponding confidence interval within a very short time. This allows for the replacement of hardware measurement of screw assembly height (e.g., eliminating the need for a rear-mounted laser height measuring station, reducing hardware costs, and saving production line space in the screw assembly line). It also enables rapid virtual measurement of screw assembly height after screw fastening, with the detection time overlapping with the screw machine's cycle time, achieving zero cycle time loss. Furthermore, multi-dimensional dynamic feature engineering and material-process coupled modeling can reduce the prediction error of screw assembly height, achieving a breakthrough in the accuracy of screw assembly height estimation. It can also achieve local data closed loop based on lightweight regression model and edge gateway heterogeneous computing, avoid cloud latency, and thus meet the real-time requirements of screw assembly scenarios through edge devices.

[0026] Based on this, embodiments of this application provide a method for measuring and evaluating assembly height, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the assembly height measurement and evaluation method of this application.

[0027] In this embodiment, the assembly height measurement and evaluation method includes steps S10 to S40.

[0028] Step S10: Obtain the signal curve diagram representing the screw assembly timing signal, and perform multi-dimensional feature extraction based on the signal curve diagram to obtain the first multi-dimensional feature; Optionally, the assembly height can be the actual height or depth of the top of the screw relative to the assembly reference surface after the screw is tightened, which is a key geometric dimension for measuring whether the screw is tightened properly.

[0029] Optionally, the screw assembly timing signal can be a time-varying process parameter signal acquired by the acquisition system through various sensors during the screw tightening process. The screw assembly timing signal can include the initial torque signal, rotation speed signal, and number of rotations signal during screw assembly. The initial torque signal can characterize the torque applied by the screwdriver or tightening shaft. The rotation speed signal can characterize the rotational speed of the screwdriver or tightening shaft. The number of rotations signal can characterize the cumulative rotation angle of the screwdriver or tightening shaft.

[0030] Optionally, the signal graph can be a signal graph, and can be one or more sets of waveforms or data sequences that change over time, consisting of screw assembly timing signals. One signal graph can correspond to a complete digital record of a single screw tightening process.

[0031] Optionally, multi-dimensional feature extraction can be a process of extracting fixed-dimensional numerical feature vectors from the original screw assembly timing signal through a series of mathematical transformations and statistical algorithms. These vectors characterize different physical aspects of the tightening process (such as mechanical properties, motion characteristics, and material-process interaction characteristics). The aim is to reduce the dimensionality of high-dimensional, redundant time-series data and transform it into a representation that is more conducive to model learning.

[0032] Optionally, the first multidimensional feature may include features of multiple dimensions during the screw tightening process, such as mechanical dimension features, kinematic dimension features, and material-process coupling dimension features.

[0033] Optionally, the acquisition system can acquire screw assembly timing signals for each screw being assembled in the screw assembly scenario, encrypt the acquired screw assembly timing signals using MQTT-TLS to form MQTT-TLS encrypted messages, and then send them to the edge system. The edge system decrypts the messages and generates a signal curve representing the screw assembly timing signals.

[0034] Optionally, the acquisition system can acquire screw assembly timing signals using sensors installed on the screw assembly production line. The edge system can be deployed on edge devices, which can be electronic devices, terminal devices, servers, etc.

[0035] Optionally, multiple buffers can be set in the edge device, such as dual buffers and triple buffers. Taking a dual buffer as an example, the screw assembly timing signals corresponding to two screws can be received simultaneously through the dual buffers, and steps S10-S40 can be executed simultaneously using two threads based on the screw assembly timing signals corresponding to the two screws in the dual buffers. Alternatively, a single thread can be used to execute steps S10-S40 based on the screw assembly timing signal corresponding to one screw in one buffer, and after execution, steps S10-S40 can be executed based on the screw assembly timing signal corresponding to another screw in another buffer. At the same time, the first buffer receives the screw assembly timing signal of the new screw sent by the acquisition system.

[0036] Optionally, the CPU side in the edge device can maintain an A / B buffer. When the inference processing thread consumes the timing signal data in the A buffer, the data receiving thread writes the next timing signal data into the B buffer to form a new timing signal curve, thereby achieving zero waiting time in the screw assembly line.

[0037] Optionally, after acquiring the screw assembly timing signal, the edge device can represent the screw assembly timing signal in chronological order as an image to obtain a signal curve representing the screw assembly timing signal.

[0038] Optionally, after obtaining the signal curve, features can be extracted from the signal curve in various dimensions to obtain the first multidimensional features.

[0039] Optionally, after obtaining the signal curve diagram representing the screw assembly timing signal in step S10, steps a10-a20 are included.

[0040] Step a10: Perform data preprocessing on the signal curve; It should be noted that the data preprocessing includes moving average filtering to remove the first frequency jitter signal in the signal curve, dynamic time warping to eliminate random phase shifts in the signal curve based on the preset theoretical speed curve, and abnormal segment removal to eliminate abnormal segments in the signal curve. Optionally, after obtaining the signal curve during screw assembly, the signal curve can be preprocessed using the data preprocessing engine in the edge device.

[0041] Optionally, when performing moving average filtering on the signal curve, a sliding time window can be set on the display interface when displaying the signal curve. The data can be smoothed by calculating the average value of each screw assembly timing signal in the signal curve within the sliding time window, so as to effectively suppress high-frequency noise or jitter, thereby suppressing and removing the first frequency jitter signal (i.e., screw assembly timing signal with a frequency greater than or equal to a preset frequency value (e.g., 1kHz)) in the signal curve. The first frequency jitter signal can be a high-frequency jitter signal.

[0042] Optionally, when performing dynamic time warping on the signal curve, an algorithm for two time series of different lengths can be used, and the first signal curve in the signal curve can be nonlinearly aligned with a preset theoretical speed curve (such as the signal curve of the ideal screw tightening speed change) to eliminate the random phase shift of the first signal curve.

[0043] Optionally, when performing abnormal segment removal processing on the signal curve, abnormal timing signals in the signal curve can be identified, marked, and removed, such as being marked as stripped threads or stuck screws. The standard deviation of each screw assembly timing signal in the signal curve can be determined, and screw assembly timing signals in the signal curve that are greater than a certain multiple of the standard deviation (such as 3 times the standard deviation) can be regarded as abnormal timing signals to avoid subsequent contamination of the first model and the phenomenon that leads to a decrease in the prediction accuracy of the first model.

[0044] Step a20: Based on the signal curve after data preprocessing, perform multi-dimensional feature extraction based on the signal curve to obtain the first multi-dimensional feature.

[0045] Optionally, after preprocessing the signal curve, multi-dimensional features of the screw assembly process can be extracted based on the signal curve, and the features of each dimension can be summarized to form the first multi-dimensional feature.

[0046] In this embodiment, by preprocessing the signal curve, extracting multi-dimensional features, and then using the first model to predict the height of the screw assembly, abnormal factors in the signal curve can be eliminated, so that the first model can more accurately predict the height of the screw assembly.

[0047] Step S20: Using the pre-trained first model, predict the first height of the screw based on the first multidimensional features; It should be noted that the first model includes two different base models used to predict the first multidimensional feature, and a meta-model used to make predictions based on the prediction results of the two different base models. Optionally, the first model can be a machine learning model, and can be integrated with the corresponding model architecture using Stacking (Stacked Generalization), such as a first model architecture layer containing two different base models and a second model architecture layer containing a meta-model.

[0048] Optionally, the base model can be a model set in the first model architecture layer, such as a decision tree model (e.g., LightGBM (Light Gradient Boosting Machine)), a gradient boosting tree model (e.g., XGBoost (eXtreme Gradient Boosting) or CatBoost (Categorical Boosting)), or a multinomial regression model (e.g., a third-order multinomial regression model). Multiple base models can be set in the first model architecture layer, such as two base models, and the two base models can perform prediction processing in parallel.

[0049] Alternatively, the metamodel can be a model set in the second model architecture layer, such as a linear regression model, a ridge regression model, etc.

[0050] Optionally, the input to the second model architecture layer can be the output of the first model architecture layer, and the output of the second model architecture layer can be the output of the first model.

[0051] Optionally, the first height prediction value can be a height prediction value, and can be the assembly height value of the screw in this assembly process predicted by the first model.

[0052] Optionally, a first model can be pre-configured and trained to obtain a pre-trained first model.

[0053] Optionally, the first multidimensional feature can be input into the pre-trained first model. In the first model, the first multidimensional feature can be predicted by two different base models in the first model architecture layer to obtain two prediction results. These two prediction results are then input into the meta-model in the second model architecture layer. The meta-model is used to fuse and predict the two prediction results to obtain the predicted value representing the height of the screw assembly, which is then used as the first height prediction value.

[0054] Optionally, while the first model predicts the height of the screw assembly and obtains the first height prediction value, the corresponding confidence interval can also be determined by combining the historical prediction data of the first model (such as the height prediction values ​​of the previous 100 measurements of the screw assembly).

[0055] Optionally, the confidence interval is a process requirement of the equipment itself and can be directly output by the equipment. The confidence interval includes the upper and lower limits of the height measurement value (such as the first predicted height value). Furthermore, different confidence intervals can be used for different assembly processes and different screws.

[0056] Optionally, the confidence interval of the screw corresponding to the first height prediction value can also be detected. For example, it can be detected whether the half-width of the confidence interval is greater than a preset width threshold, thereby determining whether the screw used in this assembly process is correct. The preset width threshold can be determined based on the half-width of the confidence interval corresponding to the screw that should theoretically be used in the current screw assembly process, for example, the two are equal.

[0057] Optionally, if the half-width of the confidence interval does not match the preset width threshold (e.g., the half-width of the confidence interval is greater than the preset width threshold), a corresponding alarm message can be triggered, and a re-inspection can be performed through a vision station (such as a vision sensor) to determine whether the assembled screw is abnormal. If the half-width of the confidence interval matches the preset width threshold, subsequent process steps can continue.

[0058] Optionally, after step S20, which uses the pre-trained first model to predict the first height of the screw based on the first multidimensional features, steps b10-b30 are included.

[0059] Step b10: Detect whether the absolute value of the difference between the first height prediction value and the preset target height value is greater than the preset difference threshold. Step b20: If the difference is greater than the preset difference threshold, then it is determined that the first height prediction value and the preset target height value do not match. Step b30: If the difference is less than or equal to the preset difference threshold, then the first height prediction value and the preset target height value are determined to match.

[0060] Optionally, the preset target height value can be a pre-set standard height value for screw assembly, such as the ideal height value that the screws should reach after assembly, preset according to product design and process requirements.

[0061] Optionally, the preset difference threshold can be the maximum deviation value of the screw assembly height set in advance, and different preset difference thresholds can be set according to different process assembly scenarios.

[0062] Optionally, after predicting the first height of the screw assembly using the pre-trained first model, the absolute value of the difference between the first height value and the target height value can be calculated, and it can be checked whether the absolute value of the difference is greater than a preset difference threshold. If the absolute value of the difference is greater than the preset difference threshold, it can be determined that the first height prediction value of the screw assembly predicted by the first model differs significantly from the target height value, and therefore the two do not match. However, if the absolute value of the difference is less than or equal to the preset difference threshold, it can be determined that the first height prediction value matches the preset target height value.

[0063] In this embodiment, after obtaining the first height prediction value of the screw assembly using the first model, the first height prediction value can be evaluated using a preset target height value to determine whether the first height prediction value is qualified and valid. If the absolute value of the difference between the first height prediction value and the preset target height value is greater than a preset difference threshold, it can be determined that the two do not match, and it can be considered that the height of the screw assembly may be abnormal. However, if the absolute value of the difference between the first height prediction value and the preset target height value is less than or equal to the preset difference threshold, it can be determined that the two match, thereby achieving an effective evaluation of the first height prediction value using the first model.

[0064] Step S30: In response to the mismatch between the first height prediction value and the preset target height value, it is determined that the screw assembly height is abnormal, and a preset correction command is output to correct the screw assembly parameters of the next screw. Step S40: In response to the matching of the first height prediction value and the preset target height value, the assembly height of the screw is determined to be qualified.

[0065] Alternatively, the edge system can be connected to control devices (such as PLCs) in an external screw assembly process line via a twin interface.

[0066] Optionally, in response to a mismatch between the first height prediction value and the preset target height value, it can be determined that the assembly height of the screw is abnormal. Therefore, a preset correction command, such as a torque correction character command, can be output to an external control device to correct the screw assembly parameters of the next screw, such as automatically adjusting the pressing torque threshold of the next screw.

[0067] Optionally, in response to the matching of the first height prediction value and the preset target height value, it can be determined that the assembly height of the screw is qualified, and the process of steps S10 to S40 can be performed for the next screw until all screws in the screw assembly production line are assembled.

[0068] In this embodiment, by acquiring the signal curve representing the screw assembly timing signal and extracting multi-dimensional features, a first multi-dimensional feature is obtained. Then, a pre-trained first model is used to predict the first height of the screw based on the first multi-dimensional feature. This allows for consideration of the multi-dimensional features of the screw assembly when measuring and evaluating the screw assembly height, and leverages the model's high prediction accuracy to improve the accuracy of the predicted first height. To further improve the accuracy of the predicted first height, the first model can be configured to include two different base models for predicting the first multi-dimensional feature and a meta-model for predicting based on the prediction results of the two base models. This allows for the integration of multiple models to predict the screw assembly height from different perspectives, thereby improving the accuracy of the predicted first height. When the predicted first height does not match the preset target height, an anomaly in the screw assembly height is identified, and a preset correction command is output to correct the screw assembly parameters of the next screw. This allows for early detection of anomalies without affecting the subsequent screw assembly process. When the predicted height value matches the preset target height value, the assembly height of the screw is determined to be qualified. This avoids the need to use a rear-mounted laser height measurement station for inspection, which would increase hardware and space costs. Furthermore, when predicting the screw assembly height, multi-dimensional features are comprehensively considered, and the advantages of the model are utilized to improve the accuracy of screw assembly height measurement and evaluation.

[0069] Based on the first embodiment of this application, a second embodiment of this application is proposed. In this second embodiment, content that is the same as or similar to the above embodiment can be referred to the above description, and will not be repeated hereafter. Based on this, refer to... Figure 2 In step S10, the step of extracting multi-dimensional features based on the signal curve to obtain the first multi-dimensional feature also includes steps c10-c20.

[0070] Step c10: Determine the second multidimensional feature of a single screw assembly corresponding to the signal curve graph; It should be noted that the second multidimensional feature includes mechanical dimension features, kinematic dimension features, and material-process coupling dimension features.

[0071] Optionally, mechanical dimensions can characterize the strength and stability of the connection within a single screw assembly (e.g., a single screw tightening cycle). Kinematic dimensions can characterize the smoothness and consistency of the screw tightening process. Material-process coupling dimensions can characterize the physical relationship between the motion during screw assembly and the final geometric result.

[0072] Optionally, the mechanical dimension features include peak torque ratio, steady-state slope, and energy dissipation integral; the kinematic dimension features include rotational speed fluctuation entropy and revolution-time nonlinearity; and the material-process coupling dimension features include theoretical axial position and hysteresis area.

[0073] Optionally, the peak torque ratio can be determined according to Formula 1, and the peak torque ratio can be used to determine whether the screw is over-tightened or under-tightened during the tightening process. For example, if the peak torque ratio is greater than 1, it means that the torque is excessive and may damage the screw thread. If the peak torque ratio is less than 1, it means that the torque is insufficient and the connection may be unreliable.

[0074] Formula 1; in, This refers to the peak torque ratio; The maximum torque value measured during the actual screw tightening process can be determined when the signal curve includes a torque-time curve. The global maximum value of the torque-time curve can then be used as the maximum torque value. The target torque threshold set for the screw assembly process.

[0075] Optionally, the steady-state slope can be determined according to Formula 2. The steady-state slope can be obtained by identifying the stable segment in the torque-time curve, performing linear regression, and then calculating the slope of the linear regression line. The steady-state slope characterizes the stiffness properties of the screw material in the elastic stage; the larger the slope, the higher the stiffness.

[0076] Formula 2; Where k is the steady-state slope; This represents the torque T of the screw relative to the rotation angle. The first derivative; This indicates the rotational angle at which the torque T reaches a stable state.

[0077] Optionally, the energy dissipation integral can be determined according to Formula 3, which can reflect the total mechanical work consumed in the screw assembly process (such as tightening), and can be obtained by integrating the torque-time curve.

[0078] Formula 3; Where E is the integral of energy dissipation; T is the instantaneous torque value; This represents the instantaneous angle increment.

[0079] Optionally, the speed fluctuation entropy can be determined according to Formula 4. The larger the speed fluctuation entropy, the more unstable the speed.

[0080] Formula 4; in, The entropy of rotational speed fluctuation; This represents the probability that the rotational speed value falls within the i-th interval, where i is a positive integer.

[0081] Optionally, the rotation time nonlinearity can be determined according to Formula 5 to measure the degree to which the actual rotation process deviates from the ideal uniform rotation. When the signal curve includes the rotation-time curve, a linear regression calculation can be performed to fit a straight line, and the rotation time nonlinearity can be calculated using Formula 5.

[0082] Formula 5; in, SSres is the nonlinearity of lap time; SSres is the sum of squares of the residuals between the lap-time curve and the fitted straight line; SStot is the sum of squares of the total deviation of the laps from its mean.

[0083] Alternatively, the theoretical axial displacement can be determined according to Formula 6. Formula 6; in, denoted as axial displacement, and p as the thread pitch during screw assembly.

[0084] Alternatively, the hysteresis area can be determined according to Formula 7.

[0085] Formula 7; Where S is the hysteresis area; ∮ represents the line integral along the closed loop; This is the differential of the theoretical axial displacement.

[0086] Step c20: Calculate the SHAP value corresponding to each dimension feature in the second multidimensional feature, and filter a preset number of different dimension features in the multidimensional feature based on the SHAP value to obtain the first multidimensional feature.

[0087] Optionally, each dimension feature in the second multidimensional feature can be at least one of the mechanical dimension feature, the kinematic dimension feature, and the material-process coupling dimension feature.

[0088] Optionally, the SHAP (SHapley Additive exPlanations) value can be used to quantify the contribution of each dimension feature in the second multidimensional feature to the predicted height. The larger the absolute value of the SHAP value, the greater the contribution.

[0089] Optionally, after obtaining the second multidimensional features, feature selection processing can be performed. The SHAP value corresponding to each feature dimension can be calculated and sorted from largest to smallest according to the absolute value of the SHAP values. Then, a preset number (used for pre-setting) of different features can be selected based on the sorting results to obtain the first multidimensional features. For example, if the number of second multidimensional features is 20, then 15 features with larger absolute SHAP values ​​can be selected as the first multidimensional features.

[0090] In this embodiment, when extracting multi-dimensional features for a single screw assembly, a second multi-dimensional feature, including mechanical, kinematic, and material-process coupling features, can be obtained based on the signal curve. Then, the SHAP value corresponding to each dimension is calculated, and feature filtering is performed based on the magnitude of the SHAP value to obtain a first multi-dimensional feature that is more useful for predicting the screw assembly height. This facilitates improving the accuracy of the first height prediction value of the screw assembly obtained by the first model.

[0091] Optionally, step S20, which involves using a pre-trained first model to predict the first height prediction value and confidence interval of the screw based on the first multidimensional features, includes steps d10-d20.

[0092] Step d10: Input the first multidimensional feature into the pre-trained first model, and predict the first multidimensional feature in parallel according to the decision tree model and the linear model, and output the second height prediction value and the third height prediction value of the screw respectively. Step d20: Based on the Ridge regression model, the predicted values ​​of the second and third altitudes are fused and predicted to obtain the predicted value of the first altitude.

[0093] It should be noted that in the first model, the two different base models include the decision tree model and the linear model located in the first model architecture layer, and the meta-model includes the Ridge regression model located in the second model architecture layer.

[0094] Optionally, the second height prediction value can be a height prediction value, and can be the assembly height value of the screw in this assembly process predicted by the base model (such as a decision tree model).

[0095] Optionally, the third height prediction value can be a height prediction value, and can be the assembly height value of the screw in this assembly process predicted using another base model (such as a linear model).

[0096] Optionally, the decision tree model can be a LightGBM model. Optionally, the linear model can be a third-order multinomial regression model.

[0097] Optionally, the first multidimensional feature can be input into the pre-trained first model. Multiple different base models can be set in the first model architecture layer, such as the LightGBM model, XGBoost model, CatBoost model, and third-order multinomial regression model. This embodiment only uses two different base models (i.e., the decision tree model and the linear model) as examples.

[0098] Optionally, in the first model architecture layer, two parallel channels can be used to run the decision tree model and the linear model simultaneously. The first multidimensional feature is input into the decision tree model, such as the LightGBM model, to predict the height of the screw assembly, outputting a second height prediction value. The first multidimensional feature is then input into the linear model, such as a third-order multinomial regression model, to predict the height of the screw assembly, outputting a third height prediction value.

[0099] Optionally, the second and third height predictions can be height predictions with different perspectives.

[0100] Optionally, the second and third height predictions can be used as inputs to the meta-model (such as the Ridge regression model) in the second model architecture layer, and the Ridge regression model can be used to perform fusion prediction on the second and third height predictions to output the first height prediction.

[0101] In this embodiment, by using a decision tree model and a linear model in parallel to predict the first multidimensional features in a pre-trained first model, the height of the screw assembly can be predicted from different angles, resulting in a second height prediction value and a third height prediction value. Then, a Ridge regression model is used to fuse the second and third height prediction values ​​to obtain the first height prediction value. This allows for a comprehensive consideration of height prediction values ​​from two different angles, making the final predicted first height value more accurate.

[0102] Based on the first or second embodiment of this application, a third embodiment of this application is proposed. In this third embodiment, content that is the same as or similar to the above embodiments can be referred to the above description, and will not be repeated hereafter. Based on this, refer to... Figure 3 The assembly height measurement and evaluation method also includes steps e10-e40.

[0103] Step e10: Construct a training dataset containing at least one training data point, wherein the training data includes the historical multidimensional features of the screw and the historical assembly height value of the screw; Optionally, the training dataset may contain multiple sets of labeled training data, and the training data may include historical screw assembly data, namely the actual height value after successful historical screw assembly (i.e., historical assembly height value), and the multidimensional features of historical screws during the assembly process (i.e., historical multidimensional features). The number of historical multidimensional features may be the same as that of the first multidimensional features, and they are also selected from the mechanical dimension features, kinematic dimension features, and material process coupling dimension features corresponding to the historical screws.

[0104] Step e20: Input the training data into the preset first model, and use two different base models in the first model to predict the training data to obtain the fourth height prediction value and the fifth height prediction value. Optionally, training data (such as historical multidimensional features) can be input into a pre-set first model, and in the first model architecture layer of the first model, the training data can be simultaneously input into two different base models for model training, so as to predict the height prediction value of the screw assembly corresponding to the training data, and obtain the fourth height prediction value and the fifth height prediction value.

[0105] Step e30: Input the fourth altitude prediction value and the fifth altitude prediction value into the meta-model in the first model for fusion prediction to obtain the sixth altitude prediction value; It should be noted that the meta-model has undergone weight and activation function quantization and is accelerated using the TensorRT engine.

[0106] Optionally, the meta-model (such as the Ridge regression model) can be quantized with weights and activation functions, such as INT8 quantization or FP16 quantization, to reduce the model size.

[0107] Optionally, the metamodel can be accelerated using the ensorRT engine, such as by building a CUDA engine, enabling the fp16 accumulator, and setting the batch size to 8, thereby improving the utilization of the GPU in the device.

[0108] Optionally, the output of the first model architecture layer can be used as the input of the second model architecture layer. That is, the fourth height prediction value and the fifth height prediction value can be input into the metamodel in the second model architecture layer for fusion prediction processing to obtain the sixth height prediction value, and then used as the final height prediction value of the screw assembly corresponding to the training data by the first model.

[0109] Step e40: Input the sixth height prediction value and the historical assembly height value into the preset loss function to calculate the loss function value, and update the model parameters of the first model according to the loss function value until the preset training end condition is met, and obtain the pre-trained first model.

[0110] Optionally, the preset loss function can be a pre-set loss function, such as the MSE (root mean square error) function of five-fold cross-validation. The preset training termination condition can be reaching a certain number of training iterations, such as 60 iterations, the loss function value being less than the preset loss function value, or other conditions such as model convergence; there are no restrictions here.

[0111] Optionally, the sixth height prediction value and the historical assembly height value can be input into a preset loss function to calculate the loss function value. The loss function value can be used to update the model parameters of the first model, such as performing hyperparameter optimization. Then, new training data can be selected, the first model can be retrained, and steps e20-e40 can be executed again until the preset training termination condition is met, and the trained first model is obtained.

[0112] Optionally, when updating the model parameters of the first model, only the model parameters of the meta-model can be updated.

[0113] In this embodiment, a training dataset is constructed to train a preset first model. The loss function value is calculated by combining the historical assembly height value of the screw assembly with the predicted sixth height value. The model parameters of the first model are then updated based on the calculated loss function value until a well-trained first model is obtained. This facilitates the subsequent use of the first model to predict the height of the screw assembly and ensures the accuracy of the predicted height value.

[0114] In addition, this application also provides an assembly height measurement and evaluation system, referring to Figure 4 The assembly height measurement and evaluation system includes a data acquisition system A100 and an edge system A200. The data acquisition system A100 is used to acquire screw assembly timing signals and send the screw assembly timing signals to the edge system A200. The edge system A200 includes: The acquisition module A10 is used to acquire a signal curve diagram representing the screw assembly timing signal, and to perform multi-dimensional feature extraction based on the signal curve diagram to obtain the first multi-dimensional feature. Prediction model A20 is used to predict the first height of a screw based on a first multidimensional feature using a pre-trained first model. The first model includes two different base models for predicting the first multidimensional feature and a meta-model for making predictions based on the prediction results of the two different base models. The exception module A30 is used to respond to the mismatch between the first height prediction value and the preset target height value, determine that the screw assembly height is abnormal, and output a preset correction command to correct the screw assembly parameters of the next screw. The qualified module A40 is used to determine that the screw assembly height is qualified in response to the matching of the first height prediction value and the preset target height value.

[0115] The assembly height measurement and evaluation system provided in this application, employing the assembly height measurement and evaluation method described in the above embodiments, can improve the accuracy of screw assembly height measurement and evaluation while reducing costs. Compared with the prior art, the beneficial effects of the assembly height measurement and evaluation system provided in this application are the same as those of the assembly height measurement and evaluation method provided in the above embodiments, and other technical features of the assembly height measurement and evaluation system are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0116] Furthermore, this application provides an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the assembly height measurement and evaluation method in Embodiment 1 described above.

[0117] The following is for reference. Figure 5 The figure illustrates a structural diagram of an electronic device suitable for implementing embodiments of this application. The electronic devices in the embodiments of this application may include, but are not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. The devices shown in the figure are merely examples and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0118] The electronic device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for device operation. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. While electronic devices with various systems are shown in the figures, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0119] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0120] The electronic device provided in this application, employing the assembly height measurement and evaluation method in the above embodiments, can solve the technical problem of how to improve the accuracy of screw assembly height measurement and evaluation while reducing costs. Compared with the prior art, the beneficial effects of the electronic device provided in this application are the same as those of the assembly height measurement and evaluation method provided in the above embodiments, and other technical features of this electronic device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0121] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0122] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0123] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the assembly height measurement and evaluation method in the above embodiments.

[0124] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0125] The aforementioned computer-readable storage medium may be included in an electronic device or may exist independently without being assembled into an electronic device.

[0126] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by an electronic device, enable the electronic device to perform the steps in the aforementioned assembly height measurement and evaluation method.

[0127] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0128] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0129] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0130] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described assembly height measurement and evaluation method. This solves the technical problem of improving the accuracy of screw assembly height measurement and evaluation while reducing costs. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the assembly height measurement and evaluation method provided in the above embodiments, and will not be repeated here.

[0131] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the assembly height measurement and evaluation method described above.

[0132] The computer program product provided in this application solves the technical problem of improving the accuracy of screw assembly height measurement and evaluation while reducing costs. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the assembly height measurement and evaluation method provided in the above embodiments, and will not be repeated here.

[0133] The above are only some embodiments of this application and do not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the content of this application specification and drawings, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A method for measuring and evaluating assembly height, characterized in that, The assembly height measurement and evaluation method includes the following steps: Obtain a signal curve representing the screw assembly timing signal, and perform multi-dimensional feature extraction based on the signal curve to obtain the first multi-dimensional feature; Using a pre-trained first model, the first height prediction value of the screw is predicted based on the first multidimensional feature. The first model includes two different base models for predicting the first multidimensional feature, and a meta-model for making predictions based on the prediction results of the two different base models. In response to the mismatch between the first height prediction value and the preset target height value, it is determined that the assembly height of the screw is abnormal, and a preset correction command is output to correct the screw assembly parameters of the next screw. In response to the matching of the first height prediction value and the preset target height value, the assembly height of the screw is determined to be qualified.

2. The assembly height measurement and evaluation method as described in claim 1, characterized in that, The step of extracting multi-dimensional features based on the signal curve to obtain the first multi-dimensional feature includes: Determine the second multidimensional feature of a single screw assembly corresponding to the signal curve, wherein the second multidimensional feature includes mechanical dimension feature, kinematic dimension feature and material-process coupling dimension feature; Calculate the SHAP value corresponding to each dimension feature in the second multidimensional feature, and filter a preset number of different dimension features from the multidimensional feature based on the SHAP value to obtain the first multidimensional feature.

3. The assembly height measurement and evaluation method as described in claim 2, characterized in that, The mechanical dimension features include peak torque ratio, steady-state slope, and energy dissipation integral; the kinematic dimension features include rotational speed fluctuation entropy and revolution-time nonlinearity; and the material-process coupling dimension features include theoretical axial position and hysteresis area.

4. The assembly height measurement and evaluation method as described in claim 1, characterized in that, The two distinct base models include a decision tree model and a linear model located at the first model architecture layer, and the meta-model includes a Ridge regression model located at the second model architecture layer. The step of using a pre-trained first model to predict the first height prediction value and confidence interval of the screw based on the first multidimensional features includes: The first multidimensional feature is input into the pre-trained first model, and the first multidimensional feature is predicted in parallel according to the decision tree model and the linear model, respectively, and the second height prediction value and the third height prediction value of the screw are output. The second and third height prediction values ​​are fused and predicted based on the Ridge regression model to obtain the first height prediction value.

5. The assembly height measurement and evaluation method as described in claim 1, characterized in that, After the step of predicting the first height of the screw based on the first multidimensional features using the pre-trained first model, the following steps are included: Detect whether the absolute value of the difference between the first predicted height value and the preset target height value is greater than a preset difference threshold; If the difference is greater than a preset threshold, it is determined that the first height prediction value and the preset target height value do not match. If the difference is less than or equal to a preset difference threshold, then the first height prediction value and the preset target height value are determined to match.

6. The assembly height measurement and evaluation method as described in claim 1, characterized in that, After the step of obtaining the signal curve diagram representing the screw assembly timing signal, the following steps are included: The signal curve is preprocessed, including a moving average filtering process to remove the first frequency jitter signal in the signal curve, a dynamic time warping process to eliminate random phase shifts in the signal curve based on a preset theoretical speed curve, and an abnormal segment removal process to eliminate abnormal segments in the signal curve. Based on the signal curve after data preprocessing, the step of extracting multi-dimensional features based on the signal curve to obtain the first multi-dimensional feature is performed.

7. The assembly height measurement and evaluation method according to any one of claims 1-6, characterized in that, The assembly height measurement and evaluation method also includes: Construct a training dataset containing at least one training data point, wherein the training data includes the historical multidimensional features of the screw and the historical assembly height value of the screw; The training data is input into a preset first model, and the training data is predicted using two different base models in the first model to obtain the fourth height prediction value and the fifth height prediction value. The fourth height prediction value and the fifth height prediction value are input into the meta-model in the first model for fusion prediction to obtain the sixth height prediction value. The meta-model is a meta-model that has undergone weight and activation function quantization and is accelerated by the TensorRT engine. The sixth height prediction value and the historical assembly height value are input into a preset loss function to calculate the loss function value. The model parameters of the first model are updated according to the loss function value until the preset training termination condition is met, and the pre-trained first model is obtained.

8. An assembly height measurement and evaluation system, characterized in that, The assembly height measurement and evaluation system includes a data acquisition system and an edge system. The data acquisition system is used to acquire screw assembly timing signals and send the screw assembly timing signals to the edge system. The edge system includes: The acquisition module is used to acquire a signal curve diagram representing the screw assembly timing signal, and perform multi-dimensional feature extraction based on the signal curve diagram to obtain the first multi-dimensional feature. A prediction model is used to predict the first height of the screw based on the first multidimensional features using a pre-trained first model. The first model includes two different base models for predicting the first multidimensional features and a meta-model for making predictions based on the prediction results of the two different base models. An anomaly module is used to respond to a mismatch between the first height prediction value and the preset target height value, determine that the assembly height of the screw is abnormal, and output a preset correction command to correct the screw assembly parameters of the next screw. The qualification module is used to determine that the assembly height of the screw is qualified in response to the matching of the first height prediction value and the preset target height value.

9. An electronic device, characterized in that, The electronic device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the assembly height measurement and evaluation method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the assembly height measurement and evaluation method as described in any one of claims 1 to 7.

Images