Intelligent electric screwdriver bit calibration and identification method, device and equipment of intelligent electric screwdriver and storage medium

Abstract

The invention relates to the technical field of intelligent calibration, and discloses an intelligent calibration recognition method, device and equipment for an electric screwdriver head of an intelligent electric screwdriver and a storage medium, and the method comprises the steps: carrying out the data collection and feature extraction of the electric screwdriver head, and obtaining a feature data set; writing the equipment identification code and the feature data set into a storage chip of the electric screwdriver; constructing a projection neural network for back propagation training to obtain an intelligent calibration recognition model; when the electric screwdriver head is connected with a host, the equipment identification code and the feature data set in the storage chip are transmitted to the host; inputting the real-time working parameters and the feature data set into an intelligent calibration recognition model for parameter recognition and Kalman filtering calibration to obtain filtered calibration parameters; and judging a deviation value between the filtered calibration parameter and the initial calibration parameter, when the deviation value exceeds a preset threshold value, writing the filtered calibration parameter into the storage chip, and updating the calibration coefficient of the host, thereby solving the problem of inaccurate torsion caused by mixed use of the electric screwdriver heads.

Description

Technical Field

[0001] The present application relates to the field of intelligent calibration technology, and in particular to an intelligent calibration and identification method, device, equipment and storage medium for an electric screwdriver head of an intelligent electric screwdriver. Background Art

[0002] As a key equipment for industrial automation assembly, intelligent electric screwdrivers are widely used in precision manufacturing, electronic assembly and other fields. In the actual production process, the electric screwdriver bit is the core component that affects the assembly quality, and the accuracy and stability of its torque value directly determine the assembly effect. However, due to the high similarity in appearance of electric screwdriver bits, confusion is easy to occur during disassembly and replacement, resulting in a large deviation between the actual torque value and the nominal value, affecting the assembly accuracy.

[0003] The traditional way of managing electric screwdriver bits mainly relies on manual identification and manual calibration, which is not only inefficient but also prone to misjudgment and mismatching. Although some manufacturers use barcodes or RFID tags for identification, such solutions still have problems such as easy damage to the identification and inability to update data in real time, which cannot meet the precise control requirements of intelligent manufacturing for the assembly process. Summary of the invention

[0004] The present application provides a method, device, equipment and storage medium for intelligent calibration and identification of the electric screwdriver head of an intelligent electric screwdriver, thereby solving the problem of inaccurate torque caused by mixed use of electric screwdriver heads and avoiding the problem of calibration failure caused by inconsistent parameters.

[0005] In a first aspect, the present application provides a method for intelligent calibration and identification of an electric screwdriver head of an intelligent electric screwdriver, the method comprising: The torque coefficient, speed characteristics, temperature coefficient and stress distribution data of the electric screwdriver are collected and feature extracted to obtain a feature data set; Performing standardization processing on the characteristic data set to generate a device identification code, and writing the device identification code and the characteristic data set into a storage chip of the electric screwdriver head; Constructing a projection neural network, and inputting the feature data set into the projection neural network for back propagation training to obtain an intelligent calibration recognition model; When the electric screwdriver head is connected to a host, the device identification code and the feature data set in the storage chip are transmitted to the host; Inputting the real-time working parameters of the electric screwdriver head and the characteristic data set into the intelligent calibration recognition model to perform parameter recognition and Kalman filter calibration to obtain filtered calibration parameters; The deviation value between the filtered calibration parameter and the initial calibration parameter is determined, and when the deviation value exceeds a preset threshold, the filtered calibration parameter is written into the storage chip, and the calibration coefficient of the host is updated.

[0006] A second aspect of the present application provides an intelligent calibration and identification device for an electric screwdriver head of an intelligent electric screwdriver, the intelligent calibration and identification device for an electric screwdriver head of an intelligent electric screwdriver comprising: The acquisition module is used to collect and extract the torque coefficient, speed characteristics, temperature coefficient and stress distribution data of the electric screwdriver bit to obtain a feature data set; A processing module, used for performing standardization processing on the characteristic data set, generating a device identification code, and writing the device identification code and the characteristic data set into a storage chip of the electric screwdriver head; A training module, used for constructing a projection neural network, and inputting the feature data set into the projection neural network for back propagation training to obtain an intelligent calibration recognition model; A transmission module, used for transmitting the device identification code and the feature data set in the storage chip to the host when the electric screwdriver head is connected to the host; An identification module, used for inputting the real-time working parameters of the electric screwdriver bit and the characteristic data set into the intelligent calibration identification model to perform parameter identification and Kalman filter calibration to obtain filtered calibration parameters; The updating module is used to determine the deviation value between the filtered calibration parameter and the initial calibration parameter, and when the deviation value exceeds a preset threshold, write the filtered calibration parameter into the storage chip and update the calibration coefficient of the host.

[0007] The third aspect of the present application provides an electronic device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory so that the electronic device executes the above-mentioned intelligent calibration and identification method for the electric screwdriver head of the intelligent electric screwdriver.

[0008] A fourth aspect of the present application provides a computer-readable storage medium, in which instructions are stored. When the computer-readable storage medium is run on a computer, the computer executes the above-mentioned intelligent calibration and identification method for the electric screwdriver head of an intelligent electric screwdriver.

[0009] Compared with the prior art, the present application has the following beneficial effects: by integrating a storage chip and a sensor array in the electric screwdriver head, combined with digital signal processing technology, the accurate collection and data storage of the physical characteristics of the electric screwdriver head are realized, and the problem of inaccurate torque caused by the mixed use of the electric screwdriver head is solved. The projection neural network is combined with the improved BP algorithm for parameter training. By introducing an adaptive weight matrix and a dynamic activation function, the recognition accuracy of the model for the characteristics of the electric screwdriver head is improved, and the problem of insufficient accuracy of the traditional recognition method is overcome. A parameter calibration mechanism is designed based on the Kalman filter algorithm. Through the iterative optimization of state prediction and observation update, the real-time dynamic calibration of the parameters of the electric screwdriver head is realized, and the problem that the static calibration method cannot adapt to the dynamic change of parameters is solved. A parameter automatic update mechanism based on threshold judgment is designed. Through deviation quantization analysis and data encoding transmission, the synchronous update of the parameters of the electric screwdriver head and the calibration coefficient of the host is realized, avoiding the problem of calibration failure caused by inconsistent parameters. The overall solution adopts a modular design, and the interfaces between the functional modules are standardized, which is convenient for system maintenance and upgrade. At the same time, the reliability of data transmission and storage is guaranteed through multiple verification mechanisms. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0011] The structures, proportions, sizes, etc. illustrated in the drawings of this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportion or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical contents disclosed by the present invention.

[0012] Figure 1 It is a flowchart of an intelligent calibration and identification method for an electric screwdriver head of an intelligent electric screwdriver provided by an embodiment of the present invention; Figure 2 It is a schematic block diagram of the structure of an intelligent calibration and identification device for an electric screwdriver head of an intelligent electric screwdriver provided by an embodiment of the present invention; Figure 3 It is a schematic block diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0013] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0014] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0015] It should also be understood that the terms used in this application specification are only for the purpose of describing specific embodiments and are not intended to limit the application. As used in this application specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0016] It should be further understood that the term "and / or" used in the specification and appended claims of this application refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations. Figure 1 In the embodiment of the present application, an embodiment of the intelligent calibration and identification method of the electric screwdriver head of the intelligent electric screwdriver includes:

[0017] Step 100, collecting and extracting features of the torque coefficient, speed characteristics, temperature coefficient and stress distribution data of the electric screwdriver bit to obtain a feature data set; It is understandable that the execution subject of the present application can be the intelligent calibration and identification device for the electric screwdriver head of the intelligent electric screwdriver, or can be a terminal or a server, which is not limited here. The embodiment of the present application is described by taking the server as the execution subject as an example.

[0018] Specifically, the physical state information of the electric screwdriver is obtained through a high-precision sensor array. The sensors include torque sensors for measuring torque, Hall sensors or photoelectric encoders for monitoring speed, thermistors or thermocouple sensors for detecting temperature changes, and strain gauges or MEMS sensors for analyzing stress distribution. All sensors work together to monitor the key parameters of the electric screwdriver in real time during operation and output corresponding analog signal data. These analog signal data reflect the dynamic characteristics of the electric screwdriver, including torque response, speed change, temperature rise, and stress state, forming an analog signal data set. The analog signal is converted by high-precision analog-to-digital conversion (ADC) technology. A 16-bit analog-to-digital converter (ADC) is used to digitally convert the analog signal data set, so that data such as torque, speed, temperature and stress distribution can be stored and processed in the form of high-precision discrete numerical values. After analog-to-digital conversion, an initial digital signal set is obtained, which contains data points from each sensor and is stored in the form of time series data to form a digital signal trajectory in the time domain. The sampling frequency of the initial digital signal set is adjusted according to the working speed of the electric screwdriver to ensure that the sampled data can accurately reflect the dynamic characteristics of the electric screwdriver. For example, when the speed of the electric screwdriver is high, the sampling rate is increased to capture the high-speed changing data, while when the speed of the electric screwdriver is low, the sampling rate is reduced to avoid data redundancy and unnecessary computational burden. The adjustment of the sampling frequency is based on an adaptive sampling algorithm. By real-time monitoring of the speed of the electric screwdriver, the sampling rate of the ADC is dynamically adjusted to generate a sampled digital signal set with appropriate time resolution. The sampled digital signal set is subjected to noise suppression by a digital filter to eliminate high-frequency noise and low-frequency drift and improve the signal-to-noise ratio. Digital filtering methods include low-pass filtering, band-pass filtering and Kalman filtering, among which Kalman filtering can dynamically estimate and correct the signal based on the statistical characteristics of the signal to achieve the best signal filtering effect. After filtering, a filtered signal set is obtained. The filtered signal set is subjected to multi-scale decomposition of wavelet transform to analyze the changing characteristics of the signal at different time scales. Wavelet transform is a signal processing tool that decomposes a signal into sub-signals of different frequency bands to extract information at different levels. In the multi-scale decomposition process of wavelet transform, the signal is decomposed into multiple components of different frequencies, including low-frequency components (reflecting the overall trend) and high-frequency components (reflecting transient changes). In this way, the torque change, speed fluctuation, temperature drift and stress response of the electric screwdriver are analyzed at different time scales. Feature vectors are extracted from multi-scale feature components to convert high-dimensional signal data into low-dimensional key feature parameters. The feature extraction process includes calculating statistical features such as feature mean, variance, kurtosis, and kurtosis, and combining dimensionality reduction methods such as principal component analysis to extract the optimal feature vector that can characterize the state of the electric screwdriver. After feature extraction, a feature data set is formed.

[0019] Step 200: Standardize the feature data set to generate a device identification code, and write the device identification code and the feature data set into a storage chip of the electric screwdriver; Specifically, the feature data set is subjected to zero mean standardization. By removing the mean of the data, the data has a uniform scale in the feature dimension, avoiding the deviation caused by different dimensions of the data. In this process, the mean and standard deviation of each feature data set are calculated, and then the mean is subtracted from each data point and divided by the standard deviation to obtain the standardized feature data. Based on the standardized feature data, a 128-bit original identification code is generated by a hash algorithm. The generation of the original identification code is performed by encrypting and encoding the standardized feature data. Since the standardized data has certain regularity and consistency, the directly generated identification code can efficiently represent the unique characteristics of the device. As the unique identifier of the device, the 128-bit original identification code can correspond one-to-one with the working characteristics of the electric screwdriver head, and can be used as the identity credential of the device in the subsequent calibration and identification process. The original identification code is encrypted by the SHA-256 algorithm. SHA-256 is a cryptographic hash algorithm that maps the input data to a 256-bit fixed-length hash value with extremely high security and collision resistance. The original identification code is 128 bits, but after SHA-256 encryption, the obtained device identification code will have a more complex structure and can effectively prevent the risks caused by data leakage or tampering. The encrypted device identification code not only enhances the security of the system, but also effectively avoids the possibility of different devices generating the same identification code, ensuring that each electric screwdriver head has a unique identification. The standardized characteristic data is converted into a format suitable for storage and transmission, using a binary format or a fixed-length structure, so that it can be efficiently read and parsed during storage and subsequent processing. After the format conversion, the storage format data is obtained. The storage format data is processed in blocks and a data check bit is generated. The data is split into multiple smaller data blocks to adapt to the read and write limitations of the storage chip and improve the efficiency of data processing. Each data block calculates a check bit according to a preset check algorithm (such as CRC32, MD5, etc.) as a verification tool for data integrity. When reading data, the check bit is used to verify whether the data has an error during transmission or storage, thereby ensuring the reliability of the data. During the data storage process, the device identification code and the standardized characteristic data are combined into a complete data frame. The data frame not only contains the device identification code and characteristic data, but also includes an additional frame check code to further verify the integrity and validity of the entire data frame. The calculation method of the frame check code is similar to that of the data check bit. A simple algorithm (such as XOR operation) is used to calculate the entire data frame to ensure that any possible errors can be found in time when reading. The verified data is written into the EEPROM storage area in the memory chip of the electric screwdriver according to the preset storage address.EEPROM (Electrically Erasable Programmable Read-Only Memory) has a small storage capacity and low read and write speed, but has good data persistence and low power consumption, and is suitable for storing device identification information and related feature data. Since EEPROM can retain data after power failure, it is suitable for storing the unique identification code and feature data of the electric screwdriver, so that the device status can be quickly read and restored each time the device is started.

[0020] Step 300: construct a projection neural network, and input the feature data set into the projection neural network for back propagation training to obtain an intelligent calibration recognition model; It should be noted that the network structure of the projection neural network is designed according to the dimension of the feature data set. The design of the projection neural network involves the construction of multiple levels, among which the most basic network structure includes the input layer, the projection layer, the hidden layer and the output layer. The input layer receives the standardized feature data set, the projection layer maps these feature data into a new feature space, the hidden layer performs nonlinear transformation on the mapped features to extract deeper features, and the output layer converts these features into the final prediction results. According to the dimension of the input data, each layer of the network dynamically adjusts its number of nodes and connection methods to ensure that the network structure can effectively capture the complex patterns and laws in the input data. Through this step, an initial projection neural network structure is constructed. Based on the initial network structure, the weight matrix of the projection layer is adaptively adjusted and optimized through initialization technology. The weight matrix determines how the input data is mapped to the new feature space in the neural network. Reasonable weight initialization can help the neural network converge faster and reduce instability during training. Weight initialization methods include uniform distribution initialization, Gaussian distribution initialization, and Xavier initialization. These methods provide reasonable initial values ​​for the weight matrix according to the number of layers and activation function characteristics of the network, thereby improving training efficiency. In this process, the weights of the projection layer are adaptively updated according to the characteristics of the training data set so as to better reflect the intrinsic structure of the feature data set and obtain the optimized network structure. The feature data set is input into the optimized network structure, and the feature space is mapped through the weight matrix of the projection layer. The input data is transformed from the original feature space to a new projection feature space through the weight matrix, so as to better handle the nonlinear relationship in high-dimensional data. The projection features are used as the input for subsequent hidden layer processing, and more complex and deep feature information is extracted by performing nonlinear transformation on them. The nonlinear activation function of the hidden layer (such as ReLU, Sigmoid or Tanh, etc.) enables the network to model the nonlinear relationship in the data and improve the expression ability of the model. L2 regularization constraints are imposed on the hidden layer features. L2 regularization is to constrain the size of network parameters by adding a penalty term to the loss function, thereby preventing the model from being too dependent on certain features during the training process. By imposing regularization constraints, overfitting is effectively avoided and the generalization ability of the model on unknown data is improved. The regularized hidden layer features are input into the output layer for calculation. At the output layer, the network generates a predicted value and compares it with the actual expected output, calculating the error between the network output and the expected output. This error, as a measure of network performance, reflects the gap between the current network parameters and the actual situation. Based on the training error, the neural network calculates the gradient of each weight and bias through the back propagation algorithm (BP algorithm), and uses this gradient information to update the network parameters, thereby gradually reducing the error and enabling the network to be optimized in a more accurate direction.In order to improve the training efficiency and accelerate the convergence of the network, the learning rate is dynamically adjusted during the training process. The learning rate determines the size of the step size each time the parameters are updated. A too large learning rate will lead to unstable training or even divergence; while a too small learning rate will make the training process too slow and increase the computational burden. By monitoring the changing trend of the training error, the learning rate is dynamically adjusted to better balance the training speed and stability. For example, when the error decreases rapidly, the learning rate is appropriately reduced to avoid excessive updates; and when the error changes tend to be flat, the learning rate is appropriately increased to accelerate the convergence process. Dynamic adjustment of the learning rate helps the neural network to always maintain the best optimization effect throughout the training process. The optimized learning rate is substituted into the improved back propagation algorithm, and the various parameters of the network are updated to obtain an optimized intelligent calibration recognition model. During the training process, the weights and biases of the neural network are continuously updated according to the error and gradient information, so that the network can more accurately predict the output results after inputting the feature data. After multiple rounds of training and optimization, the network will eventually converge to a stable state, which can realize accurate calibration and recognition of the electric screwdriver head in various working environments, and finally obtain an intelligent calibration recognition model.

[0021] Step 400: When the electric screwdriver is connected to the host, the device identification code and feature data set in the storage chip are transmitted to the host; Specifically, when the electric screwdriver is connected to the host, a standard wired or wireless communication interface is used, such as industrial bus protocols such as I²C, SPI, UART or CAN, to ensure stable data transmission in different application scenarios. When the electric screwdriver is inserted into the interface of the host or a connection is established wirelessly, the host sends a handshake signal to the electric screwdriver. The function of this signal is to establish a data link and confirm whether the storage chip is readable. If the handshake is successful, the electric screwdriver will respond to the request of the host, enter the data transmission mode, and activate the reading circuit of the storage chip at the same time to extract the stored data according to the request of the host. In the process of data extraction, the storage chip of the electric screwdriver uses EEPROM or Flash memory, which has the characteristics of not losing data when power is off and supports random read and write operations. The data in the storage chip is stored in a structured format, in which the device identification code is used as the unique device identity information for the host to match and authenticate the electric screwdriver, and the feature data set contains the torque characteristics, speed characteristics, temperature compensation parameters, stress distribution information and other key parameters of the electric screwdriver, which are used by the host to perform intelligent calibration. When the host requests to read data, the memory chip accesses the corresponding data block through the address instruction, and reads the data into the data buffer in sequence according to the set byte format in preparation for transmission. Since the device identification code is stored in SHA-256 encryption, the electric screwdriver head will first decode the data during the reading process to ensure that the data transmitted to the host is correct and recognizable. When the memory chip completes data extraction, the data will be transmitted according to the specified communication protocol. For example, in the I²C communication mode, the electric screwdriver head acts as a slave device and the host acts as a master device. After the host sends a read request, the electric screwdriver head will send the device identification code and feature data set in batches to the host's data receiving buffer in the order of the storage address. In serial communication modes such as UART or SPI, data is sent bit by bit in a streaming manner. The host parses the data stream at the receiving end and reorganizes the data according to the predetermined data packet format. In this process, in order to ensure the integrity and accuracy of the data, each data packet contains information such as the frame header, data body, and check code. The frame header is used to indicate the starting position of the data, and the check code is used for error detection. It is usually calculated using methods such as CRC32 or Hamming coding to ensure that the data is not damaged or tampered with during transmission. When the host receives the complete data frame, it performs a data integrity check and confirms whether an error occurs during the data transmission process by comparing the check code. If an error is found, the host will send a retransmission request to the electric screwdriver head, requesting the retransmission of the damaged data block, so as to ensure that the data finally received by the host is consistent with the original data in the storage chip. After the data verification is completed, the host parses the device identification code and compares it with the existing device information in the system to verify whether the electric screwdriver head is a legitimate device.The verification of the device identification code is matched with the database on the host side or the cloud server. Once the verification is passed, the host stores the characteristic data set of the electric screwdriver head in local storage and updates the calibration parameters to ensure that subsequent operations can be optimized based on the latest device characteristics. After the data transmission and analysis are completed, the host will send a confirmation signal to the electric screwdriver head, indicating that the device identification code and characteristic data set have been successfully received and recorded in the system log so as to track the use of the electric screwdriver head in subsequent operations. If the host finds mismatches or error information during the data analysis process, such as the device identification code is not registered in the system, or the characteristic data set format is incorrect, the host will send an abnormality report to the electric screwdriver head and decide whether to continue the operation based on the security policy. When all data transmission, analysis and calibration parameter updates are completed, the host adjusts its operating parameters according to the characteristic data set of the electric screwdriver head, including torque setting, speed adjustment, temperature compensation, etc., to ensure that the electric screwdriver head can maintain optimal performance in the current environment.

[0022] Step 500: input the real-time working parameters and characteristic data set of the electric screwdriver bit into the intelligent calibration recognition model to perform parameter recognition and Kalman filter calibration to obtain filtered calibration parameters; Specifically, the real-time working parameters of the electric screwdriver are preprocessed and feature extracted to obtain real-time feature data. The real-time working parameters include torque, speed, temperature, pressure and other information of the electric screwdriver. Through data cleaning, denoising and normalization and other processing steps, redundant data is removed and converted into a format more suitable for model input. By inputting the real-time feature data and the feature data set into the intelligent calibration recognition model, the model performs parameter recognition based on the input data to obtain the initial recognition parameters. Using the trained neural network model or other machine learning model, the calibration parameters of the electric screwdriver under the current working state are identified according to the current input data. These initial recognition parameters provide a preliminary working parameter estimate of the electric screwdriver, but since the parameters will change over time and be affected by a variety of external factors in actual work, they are further optimized and calibrated. In order to improve the recognition accuracy, the state equation of the Kalman filter is constructed based on the initial recognition parameters. Kalman filtering is a recursive optimal estimation algorithm suitable for dynamic systems with high noise and uncertainty. In Kalman filtering, a state equation is constructed, which describes the dynamic evolution process of the system. The core of the state equation is to predict the next state of the system through the state transfer matrix, where the state transfer matrix reflects the change law of the system from the previous moment to the current moment. At the same time, the control matrix is ​​used to associate the external control input with the system state, while the process noise reflects the inherent uncertainty of the model itself. The setting of these parameters depends on the physical characteristics of the electric screwdriver and the working environment, and is obtained through experiments or system modeling. Through these parameters, the filter predicts a preliminary estimate of the current state, that is, the state prediction value. The observation equation is constructed based on the state prediction value. The observation equation associates the state variables with the actual observation values ​​so that the state prediction value can be adjusted by the actual measurement value. The observation matrix describes the linear relationship in the measurement process, while the observation noise represents the error caused by changes in the measurement equipment and the environment. The input of the observation equation is the real-time data obtained by the sensor, and the output is the predicted observation value of the system. By comparing the state prediction value with the observation prediction value, the current error is calculated. In order to optimize the estimation accuracy, the Kalman filter algorithm calculates the prior error covariance matrix, which is a quantification of the uncertainty of the current estimation of the system. The error covariance matrix reflects the uncertainty of the predicted state and is calculated based on the previous estimate and process noise. Combined with the observed predicted values, the Kalman filter calculates the Kalman gain matrix, which is used to balance the weights between the state predicted values ​​and the observed values, and determines the contribution ratio of the observed values ​​and the predicted values ​​when updating the state estimate. By minimizing the square of the prediction error, the Kalman gain can ensure the optimality of the state update and achieve an accurate estimate of the system state. The state variables are updated based on the optimal gain matrix to obtain the initial state estimate. The state estimate is the optimal estimate of the current system state, taking into account the weighted results of the predicted information and the actual observed information.By calculating the posterior error covariance matrix and performing iterative optimization, the filter can gradually improve the accuracy of state estimation. Through multiple iterations, the Kalman filter continuously adjusts the estimated value to obtain the optimal state estimate. The optimal state estimate is used as the calibration parameter after filtering, and the real-time control parameters of the electric screwdriver are updated through the feedback mechanism. The calibration parameters are adjusted in real time through the control system of the electric screwdriver to achieve precise control and calibration, ensuring that the electric screwdriver always maintains the best performance in various working environments. Through Kalman filtering, the influence of external disturbances and noise on the working parameters of the electric screwdriver can be effectively removed, the accuracy and stability of calibration can be improved, and the use effect and reliability of the electric screwdriver can be optimized.

[0023] The state transfer matrix is ​​corrected and calculated according to the optimal gain matrix to improve the accuracy of state prediction. The state transfer matrix is ​​weighted updated by the optimal gain matrix to minimize the prediction error of state transfer and obtain the corrected state transfer matrix. The state variables are updated based on the corrected state transfer matrix to calculate the initial state estimate. The update of state variables depends on the state transfer equation. By inputting the state estimate of the previous moment and combining it with the corrected state transfer matrix, a preliminary estimate of the state at the current moment is obtained. The residual calculation of the initial state estimate and the current observation is performed to quantify the deviation between the state estimate and the actual observation. The core of the residual calculation is to compare the error between the state prediction value and the observation value, and to correct it according to the size and direction of the error. By calculating the state update residual, the degree of deviation of the state estimate from the true value is identified, and the initial state estimate is corrected accordingly. The correction calculation process is based on methods such as Kalman filtering or extended Kalman filtering. By combining the prior state estimate and the observation data, the state estimate is adjusted in a way that minimizes the error to obtain a more accurate corrected state estimate. The a posteriori error covariance matrix is ​​calculated based on the corrected state estimate to evaluate the reliability of the state estimate. The a posteriori error covariance matrix describes the statistical characteristics of the estimation error and reflects the uncertainty of the system in the current state. The purpose of calculating this matrix is ​​to optimize the confidence interval of the state estimate and provide a basis for subsequent feature analysis and iterative optimization. Since the dimension of the error covariance matrix is ​​high, in order to extract key information, the error covariance data is subjected to singular value decomposition. Singular value decomposition is a matrix decomposition method that maps high-dimensional data into a low-dimensional space and extracts the main eigenvalues. By extracting the main eigenvalues ​​of the error covariance matrix, the main mode of system state change is identified and the low-energy noise components are removed, thereby improving the robustness of the state estimation. The state estimate is iteratively calculated based on the eigendecomposition results to ensure that the final state estimate is stable and optimal. The iterative calculation gradually adjusts the state estimate so that it converges continuously in multiple calculation rounds. When the difference between the state estimates of two consecutive iterations is less than the preset target threshold, it means that the state estimate has stabilized and no longer changes significantly. At this time, the system reaches a convergence state. This converged state value is considered to be the optimal state estimate. When the converged state value is determined to be the optimal state estimate, it is used to update the real-time control parameters of the electric screwdriver, thereby optimizing the working accuracy and stability of the equipment.

[0024] Step 600: Determine the deviation between the filtered calibration parameter and the initial calibration parameter. When the deviation exceeds a preset threshold, write the filtered calibration parameter into the storage chip and update the calibration coefficient of the host.

[0025] Specifically, the calibration parameters after filtering are subjected to data normalization, and data of different scales and dimensions are uniformly converted into a standardized range, and minimum-maximum normalization or Z-score normalization is used to ensure that the data will not affect the calculation results due to different numerical ranges. At the same time, the initial calibration parameters are synchronously normalized to obtain normalized calibration parameters and normalized initial parameters. The normalized calibration parameters and the normalized initial parameters are subjected to difference operation to calculate the specific deviation between the two. A parameter deviation matrix is ​​constructed, and each element of the matrix represents the numerical difference between the corresponding calibration parameters after filtering and the initial state. In order to measure the overall deviation degree, the parameter deviation matrix is ​​subjected to norm calculation to obtain the deviation quantization value. The norm calculation uses the Euclidean norm or the maximum norm to measure the overall deviation degree of the parameter vector, wherein the Euclidean norm calculates the square root value of the parameter vector, and the maximum norm takes the maximum absolute value in the parameter deviation. In this way, the deviation quantization value can intuitively reflect the overall change degree of the calibration parameters after filtering and the initial state. The deviation quantization value is compared with a preset threshold. The preset threshold is a limit value set according to the accuracy requirements of the device, historical data analysis and actual application environment, ensuring that the update is triggered only when the deviation exceeds a reasonable range, and frequent storage operations and parameter adjustments will not be caused by small random errors. If the calculated deviation quantization value is greater than the preset threshold, it indicates that the calibration parameters have changed significantly compared to the initial parameters. At this time, the update signal is triggered to ensure that the calibration parameters of the system can adapt to the current working state. To this end, an update trigger signal is generated, which is used to indicate the execution of subsequent data storage and host update processes. After receiving the update trigger signal, the filtered calibration parameters are data encoded to generate an update data packet written to the storage chip. The purpose of data encoding is to convert the calibration parameters into a structured format for storage and subsequent parsing, using a fixed format binary encoding or a structured encoding based on JSON / XML to ensure data integrity and readability. After the update data packet is generated, it is written to a specific storage area of ​​the storage chip according to the preset address mapping rules. The preset address mapping is set according to the data category, the partition structure of the storage chip, and the read optimization strategy to ensure that the calibration data can be read quickly and will not conflict with other stored data. During the data writing process, in order to ensure the success of the storage operation, a write confirmation is performed, that is, after the data is successfully stored, the storage chip returns an update confirmation signal to indicate that the write operation is complete. When the update confirmation signal is received, the update data packet stored in the chip is parsed and the latest calibration coefficients are extracted. The data parsing process involves data format decoding, error detection, and data integrity verification to ensure that the read data is consistent with the written data and avoid distortion of the calibration parameters due to storage damage or transmission errors. During the data verification process, CRC verification or hash comparison methods are used to verify the integrity of the data.After successful parsing, the calibration coefficient is transmitted to the host as an updated parameter. After receiving the updated calibration coefficient, the host performs parameter replacement and applies the new calibration parameter to the control logic of the system, so that subsequent electric screwdriver operations can be adjusted based on the latest calibration value, thereby optimizing the key performance of the equipment such as torque control, speed compensation and temperature correction.

[0026] In the embodiment of the present application, by integrating a storage chip and a sensor array in the electric screwdriver head and combining digital signal processing technology, the accurate collection and data storage of the physical characteristics of the electric screwdriver head are realized, and the problem of inaccurate torque caused by the mixed use of the electric screwdriver head is solved. The projection neural network is combined with the improved BP algorithm for parameter training. By introducing an adaptive weight matrix and a dynamic activation function, the recognition accuracy of the model for the characteristics of the electric screwdriver head is improved, and the problem of insufficient accuracy of the traditional recognition method is overcome. A parameter calibration mechanism is designed based on the Kalman filter algorithm. Through the iterative optimization of state prediction and observation update, the real-time dynamic calibration of the parameters of the electric screwdriver head is realized, and the problem that the static calibration method cannot adapt to the dynamic change of parameters is solved. A parameter automatic update mechanism based on threshold judgment is designed. Through deviation quantization analysis and data encoding transmission, the synchronous update of the parameters of the electric screwdriver head and the calibration coefficient of the host is realized, and the calibration failure problem caused by inconsistent parameters is avoided. The overall solution adopts a modular design, and the interfaces between the functional modules are standardized, which is convenient for system maintenance and upgrading. At the same time, the reliability of data transmission and storage is guaranteed through multiple verification mechanisms.

[0027] In a specific embodiment, the process of executing step 100 may specifically include the following steps: The torque coefficient, speed characteristics, temperature coefficient and stress distribution of the electric screwdriver are collected through the sensor array to obtain a simulation signal data set; Digitally converting the analog signal data set through a 16-bit analog-to-digital converter to obtain an initial digital signal set; According to the working speed of the electric screwdriver, the sampling frequency of the initial digital signal set is adjusted to obtain a sampled digital signal set, and the noise of the sampled digital signal set is suppressed by a digital filter to obtain a filtered signal set; The filtered signal set is subjected to multi-scale decomposition by wavelet transform to obtain multi-scale feature components, and feature vectors are extracted from the multi-scale feature components to obtain a feature data set.

[0028] Specifically, the sensor array collects signals in various key parts of the electric screwdriver (such as torque sensors, temperature sensors, speed sensors and stress sensors). Each sensor generates a corresponding analog signal based on the physical quantity it measures. These signals have certain physical properties. For example, the torque sensor outputs a voltage signal that is proportional to the torque applied by the electric screwdriver, and the temperature sensor outputs a voltage signal that is proportional to the change in the operating temperature of the electric screwdriver. A 16-bit analog-to-digital converter (ADC) is used to digitize these analog signals. Assume that the analog signal collected by the sensor is ,in Indicates time, while the signal After analog-to-digital conversion, it becomes a digital signal , Indicates the index of the sampling point. The ADC conversion process is to convert the continuous analog signal Sampling is performed to obtain a set of discrete digital values, which are expressed by the following formula: ; in, is the sampling period, i.e. the sampling interval. The sampling accuracy of 16-bit ADC means that each sampling result is a 16-bit number, which represents the discrete value of the analog signal at that moment. According to the sampling theorem, the sampling frequency should be at least twice the highest frequency of the signal to prevent sampling distortion. Assume that the working frequency of the electric screwdriver is , sampling frequency Must meet: ; Adjust the sampling frequency according to the working speed of the electric screwdriver. The speed of the electric screwdriver determines the frequency of signal change. If the speed is high, the frequency components of the signal are more, so a higher sampling frequency is required to capture more detailed signal features. (Unit: rpm), the sampling frequency is adjusted according to the speed For example, different sampling frequency factors are selected according to the speed. Adapt to changes in the working environment. Assume that the adjusted sampling frequency is Then we have: ; in, is the coefficient adjusted based on the speed. This coefficient is adjusted according to actual needs to ensure that sufficient signal details can be captured at high speeds. After adjusting the sampling frequency, a digital filter is used to suppress noise on the sampled signal. A low-pass filter (for example, a Butterworth filter or a Chebyshev filter) is used to smooth the sampled signal. Assume that the transfer function of the filter is , the filtered signal It is obtained by the following discrete time filtering formula: ; in, is a complex variable in discrete time domain, is the sampling signal, is the transfer function of the digital filter. Through filtering, the noise component of the signal is suppressed, and the obtained is the filtered signal set. Wavelet transform is used to perform multi-scale decomposition on the filtered signal. Wavelet transform is a signal processing tool suitable for signals with non-stationary and mutation characteristics. Through wavelet transform, the signal is decomposed into frequency bands of different scales in order to extract feature information at different scales. Assume that the filtered signal is , the result of wavelet transform is: ; in, is the mother wavelet function, and They are scale factor and translation factor respectively. By selecting appropriate wavelet basis functions (for example, Daubechies wavelet, Haar wavelet, etc.), the signal is decomposed at different scales to obtain multi-scale feature components. Each scale component corresponds to the energy and characteristics of the signal in the frequency band. These characteristics can reflect the changing trend of the signal at different scales. Feature vectors are extracted for the multi-scale feature components, and the most representative features are extracted from the multiple scale components of the signal for subsequent analysis and identification. Feature extraction methods include principal component analysis, independent component analysis, etc. Through feature extraction, the signal is mapped from a high-dimensional data space to a low-dimensional feature space for subsequent classification or regression tasks. Assume that the feature data set after feature extraction is , where each Represents a feature vector extracted from wavelet decomposition, and finally obtains a feature data set .

[0029] In a specific embodiment, the process of executing step 200 may specifically include the following steps: Performing zero-mean normalization processing on the feature data set to obtain standardized feature data, and generating a 128-bit original identification code based on the standardized feature data; The original identification code is encrypted using the SHA-256 algorithm to obtain a device identification code; Performing data format conversion on the standardized characteristic data to obtain storage format data, performing block processing on the storage format data, and generating data check bits to obtain block storage data; The device identification code and the standardized characteristic data are combined into a data frame, and a frame check code is generated to obtain the checked data, which is then written into the EEPROM storage area of ​​the storage chip according to a preset storage address.

[0030] Specifically, the feature data set is subjected to zero-mean normalization to eliminate the unit differences and dimension effects of different physical quantities. Zero-mean normalization ensures that all feature data have zero mean and unit variance by subtracting the mean of the data and dividing by the standard deviation, so that the data is processed on the same scale. , where each is the value of the sampling point, and the process of zero-mean standardization is expressed as: ; in, is the mean of the data, is the standard deviation of the data. Based on the standardized feature data, a 128-bit original identification code is generated. In order to uniquely identify the electric screwdriver head and establish an association with its feature data in storage, the standardized feature data is combined into a 128-bit identification code. The standardized feature data is concatenated into a long string of binary data and mapped into a 128-bit identification code through a certain algorithm. Assume that the standardized data set is , then the original identification code Through some mapping function Get, for example: ; function It is a hash function, a weighted summation or other form of conversion, and finally outputs a 128-bit binary number that uniquely identifies the electric screwdriver. In order to enhance the security of the identification code and prevent it from being tampered with or forged, the original identification code is encrypted. The encryption uses the SHA-256 algorithm, an encryption algorithm widely used in the field of information security. The SHA-256 algorithm converts input data of any length into a fixed-length 256-bit hash value to ensure the integrity and uniqueness of the data. In this step, the original identification code is encrypted. Enter the SHA-256 algorithm to get the device identification code , and its generation process is expressed as: ; Through this encryption process, the device identification code is obtained The encrypted information containing the original feature data in 256 bits has high security and can prevent illegal copying and tampering. The standardized feature data will be formatted and converted into a format suitable for storage and transmission. This step involves encoding and structuring the data so that the stored data is stored efficiently and consistently in the storage medium. For example, assuming the original standardized feature data set is Converted to storage format data , this data structure will ensure that the data is effectively read and accessed in the storage chip. Data format conversion usually involves converting floating point numbers into fixed-length binary data, and considering the alignment and compression of the data to ensure its efficient storage in the storage space. After the format conversion, the data is segmented and divided into multiple blocks for easy storage and subsequent data verification. Each data block will generate check bits, which are used to verify the integrity and accuracy of the data. Check methods include CRC check, parity check, etc. Assuming the CRC-32 check method is used, the check bit Calculated by the following formula: ; The data segmentation and verification process ensures the reliability of the data and prevents errors during storage. The device identification code and standardized feature data form a data frame. Data frame is a common format for information transmission and storage, which organizes multiple data units into a structured data unit. Assume that the device identification code is , the standardized feature data is , the check code is , data frame It is expressed as: ; After the data frame is generated, it is subjected to frame check. The frame check code is generated by applying a check algorithm (such as CRC or SHA) to the data frame. Assume that the frame check code is The generation process is as follows: ; The verified data frame is written into the EEPROM storage area of ​​the storage chip through the preset storage address. EEPROM (Electrically Erasable Programmable Read-Only Memory) is used to store data that needs to be retained after the device is powered off and has high reliability. Assume that the storage address is , data is written according to the following formula: ; The device identification code, standardized feature data, and verification information are securely stored in the EEPROM area of ​​the memory chip, ensuring the uniqueness of the device and the integrity of the feature information.

[0031] In a specific embodiment, the process of executing step 300 may specifically include the following steps: Construct a network structure of a projection neural network according to the dimension of the feature data set, the network structure includes an input layer, a projection layer, a hidden layer and an output layer, and obtain an initial network structure; An adaptive weight matrix is ​​introduced into the projection layer of the initial network structure, and the parameters of the weight matrix are initialized to obtain the optimized network structure; Input the feature data set into the optimized network structure, map the feature space through the weight matrix of the projection layer, and obtain the projection features; The projection features are nonlinearly transformed through the dynamic activation function of the hidden layer to obtain the hidden layer features, and the L2 regularization constraint is applied to the hidden layer features to obtain the regularized features; The regularized features are input into the output layer, the error between the network output and the expected output is calculated to obtain the training error, and the learning rate is dynamically adjusted based on the changing trend of the training error to obtain the optimized learning rate; The optimized learning rate is substituted into the improved BP algorithm to update the parameters of the optimized network structure to obtain an intelligent calibration recognition model.

[0032] Specifically, the basic structure of the projection neural network is constructed according to the dimension of the feature data set. The structure of the network includes an input layer, a projection layer, a hidden layer, and an output layer. The input layer receives the standardized feature data set. Assume that the feature data set is ,in represents the number of samples, Represents the feature dimension, and the input feature vector of each sample is expressed as The number of neurons in the input layer should be equal to the dimension of the feature dataset, that is , to ensure that all feature information can enter the neural network for calculation. Construct a projection layer, the main function of which is to map the input features into feature space, making it easier to separate and identify in a new high-dimensional space. In order to achieve this mapping, an adaptive weight matrix is ​​introduced ,in is the projected dimension, usually set To improve the nonlinear expression ability of data. The projection transformation is calculated by the following formula: ; in, represents the projection feature, is the bias vector of the projection layer. In the initialization stage of the projection layer, the weight matrix Initialize with a random distribution, such as uniform distribution or Xavier initialization: ; in Represents uniform distribution, ensuring that the initial weights are within a suitable range to prevent gradient vanishing or gradient exploding. After the feature data passes through the projection layer, it is transformed nonlinearly through the hidden layer to increase the network's expressive power. The hidden layer uses an activation function Perform nonlinear transformation to transform the projection features into hidden layer features : ; in, is the weight matrix of the hidden layer, is the number of neurons in the hidden layer, is the bias term. In order to enhance the robustness of the neural network, the nonlinear activation function of the hidden layer selects ReLU, Leaky ReLU or Swish, where ReLU is defined as follows: ; In the hidden layer, L2 regularization constraints are introduced to prevent overfitting and improve the generalization ability of the model. The goal of L2 regularization is to constrain the size of the weight matrix, and its regularization term is: ; in, is the regularization coefficient, which controls the strength of the L2 constraint. The hidden layer features after regularization It is expressed as: ; After the nonlinear transformation of the hidden layer, the data enters the output layer. The goal of the output layer is to calculate the predicted value of the network. and the true value Compare and calculate the training error. The calculation formula of the output layer is as follows: ; in, is the weight matrix of the output layer, is the bias of the output layer. The loss function uses mean square error, and the calculation formula is as follows: ; The goal of the loss function is to minimize the predicted value With the true value However, since the loss changes at different speeds in different training stages, the learning rate needs to be adjusted dynamically during training to accelerate convergence and improve training stability. The dynamic adjustment strategy adopts the Adam optimization algorithm, and its update rules are as follows: ; ; ; ; in, represents the current gradient, and are the first and second order momentum of the gradient, is the initial learning rate, is the momentum parameter, is a smoothing term that prevents the denominator from being zero. Through the Adam optimization algorithm, the learning rate is adjusted according to the trend of the gradient change, so that it maintains a large value in the early stage of training to accelerate convergence, and is reduced in the later stage to prevent oscillation. After obtaining the optimized learning rate, it is substituted into the improved BP back propagation algorithm to update the network parameters. The core of the BP algorithm is to update the weights based on gradient descent. The update formula is as follows: ; in, is the weight matrix of a layer of the neural network, is the gradient of the loss function with respect to the weight. Through continuous iteration of the weight update process, the network loss value will gradually decrease, and eventually the optimal intelligent calibration recognition model will be obtained.

[0033] In a specific embodiment, the process of executing step 400 may specifically include the following steps: Perform data preprocessing and feature extraction on the real-time working parameters of the electric screwdriver bit to obtain real-time feature data; The real-time feature data and the feature data set are input into the intelligent calibration recognition model, and the parameters are recognized by the intelligent calibration recognition model to obtain the initial recognition parameters; The state equation of the Kalman filter is constructed based on the initial identification parameters, and the state transfer matrix, control matrix and process noise are used as parameters of the state equation to obtain the state prediction value; The observation equation is constructed according to the state prediction value, and the observation matrix and observation noise are used as parameters of the observation equation to obtain the observation prediction value; Calculate the prior error covariance matrix for the state prediction value, and calculate the Kalman gain matrix in combination with the observed prediction value to obtain the optimal gain matrix; The state variables are updated based on the optimal gain matrix to obtain the initial state estimate, the posterior error covariance matrix is ​​calculated based on the state estimate, and iterative optimization is performed to obtain the optimal state estimate; The optimal state estimation value is used as the filtered calibration parameter to update the real-time control parameters of the electric screwdriver.

[0034] Specifically, the real-time working parameters include dynamically changing data such as torque, speed, temperature, and pressure. Data preprocessing and feature extraction are performed on the real-time working parameters of the electric screwdriver. The preprocessing steps include data cleaning, outlier removal, and missing value filling to ensure the accuracy of the input data. The real-time parameter data after data cleaning is normalized or standardized to ensure that each parameter has the same dimension and range to avoid excessive influence of certain features due to dimensional differences in the subsequent processing process. Assume that the real-time working parameters are , the characteristic data obtained after normalization is , normalized by the following formula: ; in, and They are The mean and standard deviation of each parameter are calculated to ensure that each feature data is converted to a standard normal distribution with a mean of 0 and a standard deviation of 1. The standardized real-time feature data and feature data set are input into the intelligent calibration recognition model. The model performs parameter recognition based on the input feature data and outputs the initial recognition parameters. The initial estimate of the current state of the electric screwdriver head is identified from data from multiple sources. This process is implemented through a trained neural network or other machine learning model. The input data includes real-time data sets and historical feature data sets, and the output is the initial recognition parameters inferred by the model. Assume that the output initial recognition parameters are , which is represented as the parameter vector after model inference. After obtaining the initial identification parameters, the Kalman filter is applied for state estimation. The Kalman filter is a recursive estimation algorithm that makes the best estimate of the system state based on real-time observation data and historical data. In the Kalman filter, a state equation is defined, which describes the dynamic changes of the system state. Assume that the state vector of the system is (such as the angle and speed of the electric screwdriver), the state transfer equation is expressed as: ; in, is the state transition matrix, which describes the change of system state over time; is the control input vector, which represents the influence of external control on the system; is the control matrix, defining the relationship between control inputs and states; is the process noise, which represents the uncertainty in the system model. The observation equation is constructed based on the state prediction value, which describes the relationship between the measurement data and the system state. Assume that the measurement value is , the observation equation is expressed as: ; in, is the observation matrix, which describes the linear relationship between the state vector and the observation value; is the observation noise, which represents the measurement error or uncertainty. Through the above state equation and observation equation, the Kalman filter can recursively estimate the system state. Based on the current prediction value and the system model, the prior error covariance matrix is ​​calculated : ; in, is the posterior error covariance matrix of the previous moment, is the process noise covariance matrix, which describes the uncertainty of the system model. Combining the observed and predicted values, the Kalman filter calculates the Kalman gain matrix , which determines the weight of the state update: ; in, is the observation noise covariance matrix, which represents the uncertainty of the measurement data. Kalman gain Dynamically adjust according to the reliability of system prediction and measurement data. Use Kalman gain to update the state variable to obtain the updated state estimate : ; in, It is a priori state estimate calculated based on the prediction model. is the current observation. Through this update process, the Kalman filter provides a more accurate state estimate. Calculate the posterior error covariance matrix , which is used to describe the credibility of the state estimate: ; The iterative optimization process is carried out, and the state estimate is adjusted in each filtering process until convergence. By calculating the difference between the state estimates of multiple consecutive iterations, when the difference between the state estimates is less than the preset threshold, the Kalman filter process is considered to have converged. By continuously updating the state estimate, the optimal state estimate is obtained. , the optimal state estimate is the calibration parameter of the electric screwdriver. The optimal state estimate is applied to the real-time control system of the electric screwdriver as the filtered calibration parameter, and the dynamic calibration of the electric screwdriver is realized by updating the control parameters. For example, when the angle error of the electric screwdriver is too large, the optimal state estimate obtained by the Kalman filter can accurately adjust the angle control parameters of the electric screwdriver, thereby maintaining the high-precision operation of the equipment.

[0035] In this embodiment, before the real-time working parameters are input into the intelligent calibration identification model, the real-time working parameters are also included in delay prediction compensation, which specifically includes: performing Fourier transform on the real-time working parameters to obtain a frequency domain signal sequence, and performing phase analysis on the frequency domain signal sequence to obtain a phase characteristic curve; constructing an ideal transformer model based on the phase characteristic curve, inputting the frequency domain signal sequence into the ideal transformer model to obtain an initial compensation signal; performing stability analysis on the initial compensation signal, calculating the system pole distribution through the transfer function, and obtaining a frequency domain stability criterion; performing delay coefficient correction on the initial compensation signal according to the frequency domain stability criterion, introducing a delay-related normalization factor, and obtaining a frequency domain compensation signal; performing amplitude analysis on the frequency domain compensation signal, calculating the signal amplitude gain, and introducing a frequency-related normalization factor to obtain a normalized compensation signal; performing inverse Fourier transform on the normalized compensation signal to obtain a time domain compensation signal, and performing real-time parameter adjustment on the time domain compensation signal to obtain a delay compensation parameter; constructing a signal transformation matrix based on the delay compensation parameter, performing signal transformation on the real-time working parameters, and obtaining compensated real-time working parameters; The compensated real-time working parameters are synchronously aligned with the characteristic data set to obtain a delay-free working parameter sequence.

[0036] The real-time feature data and the feature data set are input into the intelligent calibration recognition model, and the parameters are identified through the intelligent calibration recognition model to obtain the initial recognition parameters, including: data splicing of the real-time feature data and the feature data set to generate a fused data matrix, and the fused data matrix is ​​input into the input layer of the intelligent calibration recognition model; dimension mapping of the fused data matrix is ​​performed through the input layer to obtain an input feature vector, and the input layer contains 32 input nodes, and each node corresponds to a feature dimension; the input feature vector is input into the projection layer for feature space transformation, and the projection layer contains 64 neurons. The features are linearly projected through an adaptive weight matrix to obtain a projection feature vector; feature extraction of the projection feature vector is performed through a hidden layer, and the hidden layer contains 128 neurons. A dynamic activation function is used to perform nonlinear transformation on the features, and an L2 regularization constraint is introduced to obtain a deep feature vector. The deep feature vector is input into a fully connected layer for feature fusion. The fully connected layer contains 256 neurons, and the Dropout mechanism is used for feature selection to obtain a fused feature vector. The fused feature vector is feature mapped through a softmax layer. The softmax layer contains 64 output nodes, and the activation probability of each node is calculated to obtain a probability distribution vector. The probability distribution vector is input into the output layer for parameter decoding. The output layer contains 16 output nodes, and each node corresponds to a calibration parameter component to obtain a decoded parameter vector. The decoded parameter vector is denormalized, and the parameter value is mapped back to the original numerical range to obtain the initial recognition parameter.

[0037] In a specific embodiment, the process of executing step 500 may specifically include the following steps: The state transfer matrix is ​​corrected and calculated according to the optimal gain matrix to obtain a corrected state transfer matrix; Based on the modified state transfer matrix, the state variables are updated to obtain the initial state estimation value; The residual of the initial state estimate and the current observation is calculated to obtain the state update residual, and the initial state estimate is corrected based on the state update residual to obtain the corrected state estimate; The posterior error covariance matrix is ​​calculated based on the corrected state estimate to obtain error covariance data, and the error covariance data is subjected to singular value decomposition to extract the main eigenvalues ​​to obtain the eigendecomposition result; The state estimation value is iteratively calculated according to the characteristic decomposition result. When the difference between the state estimation values ​​of two consecutive iterations is less than the preset threshold, the converged state value is obtained and the converged state value is output as the optimal state estimation value to complete the state update process.

[0038] Specifically, the state transfer matrix is ​​modified and calculated according to the optimal gain matrix. Describes the dynamic behavior of the system under time step changes, which is derived through system models or physical formulas. According to the optimal gain matrix Correct the state transfer matrix. The corrected state transfer matrix The calculation is done by the following formula: ; in, is the current observation value, is the observation matrix, is a state estimate based on the previous state and the prediction model, is the original state transfer matrix, is the optimal gain matrix. The modified state transfer matrix It can more accurately reflect the dynamic behavior of the system, taking into account the noise and errors that occur in actual measurements, thereby improving the accuracy of state estimation. Based on the corrected state transfer matrix, the state variables are updated to obtain a new initial state estimate. The state update is done by using the modified transition matrix The update formula for updating the state variables is expressed as: ; in, is the estimated value of the state at the previous moment, is the current control input, is the control matrix. Through this formula, combined with the new state transfer matrix and the current control input , and obtain the initial state estimate. By comparing the initial state estimate with the current observation, the state update residual is calculated to obtain the state update residual , the residual reflects the deviation between the estimated value and the actual observed value, and the calculation formula is: ; in, is the current observation value, It is the predicted observation value calculated by the current state estimate. The residual reflects the difference between the current state estimate and the actual measured data. The initial state estimate is corrected according to the state update residual. The correction formula is: ; in, is the corrected state estimate, is the Kalman gain matrix, is the residual. Kalman gain matrix It represents the weight of the residual on the state update, which determines how to correct the current state estimate based on the measurement data. If the measurement error is large, the Kalman gain will be small, indicating less reliance on the observation data. The posterior error covariance matrix is ​​calculated based on the corrected state estimate. , the error covariance matrix describes the uncertainty of the current estimate. The calculation formula is: ; in, is the error covariance matrix at the previous moment, is the observation noise covariance matrix, is the unit matrix. Perform singular value decomposition on the error covariance data, extract the main eigenvalues, and obtain the eigendecomposition result. Singular value decomposition is a matrix decomposition method that decomposes the matrix into the product of three matrices, namely the left singular matrix, the diagonal matrix, and the right singular matrix. In Kalman filtering, the error covariance matrix is Perform singular value decomposition to extract the most important eigenvalues ​​for the estimation error, thereby optimizing the state estimation. The singular value decomposition process is expressed by the following formula: ; in, and is the unit orthogonal matrix, It is a diagonal matrix containing the singular values ​​of the matrix. In practical applications, the largest singular values ​​are selected to discard unimportant information in the error covariance matrix and retain the most useful parts, thereby improving the estimation accuracy. According to the eigendecomposition results, the state estimation value is iteratively calculated. Assume that after several iterations, two consecutive state estimation values ​​are obtained. and If the difference between the two estimates is less than a preset threshold , then the state estimation is considered to have converged. The formula for convergence judgment is: ; in, represents the norm of a vector, It is the preset convergence threshold, and an appropriate value is set according to the system requirements. When the state estimate converges, the final state estimate The optimal state estimate represents the most accurate calibration parameters of the electric screwdriver at the current moment. These optimal state estimates are used to update the real-time control parameters of the electric screwdriver, so that the device can maintain optimal performance and accuracy in various working environments.

[0039] In a specific embodiment, the process of executing step 600 may specifically include the following steps: Performing data normalization processing on the filtered calibration parameters to obtain normalized calibration parameters, and performing synchronous normalization processing on the initial calibration parameters to obtain normalized initial parameters; Performing difference calculation based on normalized calibration parameters and normalized initial parameters to obtain a parameter deviation matrix, and performing norm calculation on the parameter deviation matrix to obtain a deviation quantization value; Compare the deviation quantization value with a preset threshold value, and when the deviation quantization value is greater than the preset threshold value, obtain an update trigger signal; Data encoding is performed on the filtered calibration parameters according to the update trigger signal to obtain an update data packet, and the update data packet is written into the storage chip according to a preset address mapping to obtain an update confirmation signal; The update data packet is parsed into calibration coefficients based on the update confirmation signal, and the calibration coefficients are transmitted to the host for parameter update.

[0040] Specifically, the calibration parameters after filtering are normalized to eliminate the influence of different dimensions and numerical ranges on the calculation. Data normalization is performed by the standardization method of zero mean and unit variance. Assume that the calibration parameters after filtering are , the normalization process uses the following formula: ; in, is the mean of the original data set, is the standard deviation of the data set. Through this process, all calibration parameters are converted to a distribution with a mean of 0 and a standard deviation of 1. The initial calibration parameters are synchronously normalized to obtain normalized initial parameters, ensuring that the initial parameters and the filtered calibration parameters are on the same scale. Assume that the initial calibration parameters are , the normalization formula is the same as the calibration parameters after filtering, and we get: ; in, and are the mean and standard deviation of the initial calibration parameters respectively. The normalized initial calibration parameters It has the same scale as the filtered calibration parameters. Based on the difference between the normalized calibration parameters and the normalized initial parameters, a difference operation is performed to obtain the parameter deviation matrix. The difference matrix is ​​calculated as follows: ; The difference matrix Describes the actual deviation of each parameter. In order to quantify this difference, the norm of the parameter deviation matrix is ​​calculated to obtain the deviation quantization value. The norm calculates the deviation quantization value, the formula is: ; in, It is The difference of the parameters, Represents the overall quantitative value of the parameter deviation. Compare the deviation quantization value with the preset threshold. The preset threshold is a value set according to the performance requirements and accuracy requirements of the device. It is a value less than 1 and represents the maximum range of the deviation allowed. If the deviation quantization value is greater than the preset threshold, it means that the difference between the current calibration result and the initial value is too large and needs to be updated. This judgment is achieved through the following conditions: ; in, is the preset threshold. Greater than , the system triggers an update signal. Once the update signal is triggered, the filtered calibration parameters are then data encoded so that these updated calibration parameters can be stored in the memory chip. The data encoding process includes converting the calibration parameters into a data packet in a specific format for easy storage and transmission. The encoding process uses a certain algorithm to compress and encrypt the data packet to ensure the security and integrity of the data. The update data packet is written to the EEPROM area in the memory chip according to the preset storage address mapping. When the data is written, check whether the memory chip has enough space and whether the write address is correct. After the data is successfully stored, the system will send an update confirmation signal to ensure that the data has been stored correctly. After receiving the update confirmation signal, the update data packet is parsed into calibration coefficients, and these calibration coefficients are transmitted to the host for parameter update. After receiving the new calibration coefficients, the host applies these coefficients to the control system of the electric screwdriver head to achieve precise adjustment of the equipment. The update transmission process requires certain decoding and processing of the data to ensure that the data can be correctly transmitted and used in practical applications.

[0041] Among them, before judging the deviation value between the filtered calibration parameters and the initial calibration parameters, it also includes multi-objective optimization of the filtered calibration parameters, specifically including: establishing a work efficiency model, a torque accuracy model and an energy consumption model for the filtered calibration parameters to obtain a multi-objective optimization evaluation index; encoding the filtered calibration parameters according to the multi-objective optimization evaluation index to generate a chromosome population, each chromosome containing a work efficiency parameter, a torque accuracy parameter and an energy consumption parameter; setting a crossover probability and a mutation probability for the chromosome population, and introducing a non-dominated sorting to obtain an initial genetic algorithm parameter; based on the initial genetic algorithm parameters, calculating the fitness function of the chromosome population to obtain a population fitness value; according to the population fitness value, selecting the chromosome population using a tournament selection method to obtain a high-quality chromosome; performing an adaptive crossover operation on the high-quality chromosome, determining the crossover position by local search, and obtaining a crossover chromosome; performing an adaptive mutation operation on the crossover chromosome, dynamically adjusting the variable step length by a simulated annealing algorithm, and obtaining a mutated chromosome; performing a ranking and crowding calculation on the mutated chromosome based on a Pareto non-dominated sorting, selecting the optimal non-dominated solution, and obtaining the optimized calibration parameters.

[0042] The above describes the intelligent calibration and identification method of the electric screwdriver head of the intelligent electric screwdriver in the embodiment of the present application. The following describes the intelligent calibration and identification device 10 of the electric screwdriver head of the intelligent electric screwdriver in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of the intelligent calibration and identification device 10 for the electric screwdriver head of the intelligent electric screwdriver includes: The acquisition module 11 is used to collect and extract features of the torque coefficient, speed characteristics, temperature coefficient and stress distribution data of the electric screwdriver bit to obtain a feature data set; The processing module 12 is used to perform standardization processing on the characteristic data set, generate a device identification code, and write the device identification code and the characteristic data set into a storage chip of the electric screwdriver; The training module 13 is used to construct a projection neural network and input the feature data set into the projection neural network for back propagation training to obtain an intelligent calibration recognition model; The transmission module 14 is used to transmit the device identification code and feature data set in the storage chip to the host when the electric screwdriver is connected to the host; The identification module 15 is used to input the real-time working parameters and characteristic data set of the electric screwdriver bit into the intelligent calibration identification model to perform parameter identification and Kalman filter calibration to obtain the calibration parameters after filtering; The updating module 16 is used to determine the deviation between the filtered calibration parameters and the initial calibration parameters. When the deviation exceeds a preset threshold, the filtered calibration parameters are written into the storage chip and the calibration coefficient of the host is updated.

[0043] Through the cooperation of the above components, by integrating storage chips and sensor arrays in the electric screwdriver head and combining digital signal processing technology, the accurate collection and data storage of the physical characteristics of the electric screwdriver head are realized, and the problem of inaccurate torque caused by the mixed use of electric screwdriver heads is solved. The projection neural network combined with the improved BP algorithm is used for parameter training. By introducing adaptive weight matrix and dynamic activation function, the recognition accuracy of the model for the characteristics of the electric screwdriver head is improved, and the problem of insufficient accuracy of traditional recognition methods is overcome. A parameter calibration mechanism is designed based on the Kalman filter algorithm. Through the iterative optimization of state prediction and observation update, the real-time dynamic calibration of the parameters of the electric screwdriver head is realized, and the problem that the static calibration method cannot adapt to the dynamic change of parameters is solved. A parameter automatic update mechanism based on threshold judgment is designed. Through deviation quantification analysis and data encoding transmission, the synchronous update of the parameters of the electric screwdriver head and the calibration coefficient of the host is realized, avoiding the problem of calibration failure caused by inconsistent parameters. The overall solution adopts modular design, and the interfaces between the functional modules are standardized, which is convenient for system maintenance and upgrade. At the same time, the reliability of data transmission and storage is guaranteed through multiple verification mechanisms.

[0044] See also Figure 3 , Figure 3 This is a schematic block diagram of the structure of an electronic device 300 provided in an embodiment of the present application. The electronic device 300 includes a processor 301 and a memory 302. The processor 301 and the memory 302 are connected via a device bus 303, wherein the memory 302 may include a non-volatile storage medium and an internal memory.

[0045] The non-volatile storage medium can store a computer program. The computer program includes program instructions, and when the program instructions are executed by the processor 301, the processor 301 can execute any of the above-mentioned intelligent calibration and identification methods for the electric screwdriver head of the intelligent electric screwdriver.

[0046] The processor 301 is used to provide computing and control capabilities to support the operation of the entire electronic device 300 .

[0047] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor 301, the processor 301 can execute any of the above-mentioned intelligent calibration and identification methods for the electric screwdriver head of the intelligent electric screwdriver.

[0048] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a partial structure related to the present application scheme, and does not constitute a limitation on the electronic device 300 involved in the present application scheme. The specific electronic device 300 may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0049] It should be understood that the processor 301 may be a central processing unit (CPU), and the processor 301 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0050] It should be noted that, those skilled in the art can clearly understand that, for the sake of convenience and brevity of description, the specific working process of the electronic device 300 described above can refer to the corresponding process of the intelligent calibration and identification method for the electric screwdriver head of the aforementioned intelligent electric screwdriver, and will not be repeated here.

[0051] An embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by one or more processors, the one or more processors implement the intelligent calibration and identification method for the electric screwdriver head of an intelligent electric screwdriver as provided in an embodiment of the present application.

[0052] The computer-readable storage medium may be an internal storage unit of the electronic device 300 in the aforementioned embodiment, such as a hard disk or memory of the electronic device 300. The computer-readable storage medium may also be an external storage device of the electronic device 300, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped with the electronic device 300.

[0053] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0054] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable an electronic device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0055] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for intelligent calibration and identification of an electric screwdriver head of an intelligent electric screwdriver, characterized in that: The method comprises: The torque coefficient, speed characteristics, temperature coefficient and stress distribution data of the electric screwdriver are collected and feature extracted to obtain a feature data set; Performing standardization processing on the characteristic data set to generate a device identification code, and writing the device identification code and the characteristic data set into a storage chip of the electric screwdriver head; Constructing a projection neural network, and inputting the feature data set into the projection neural network for back propagation training to obtain an intelligent calibration recognition model; When the electric screwdriver head is connected to a host, the device identification code and the feature data set in the storage chip are transmitted to the host; Inputting the real-time working parameters of the electric screwdriver head and the characteristic data set into the intelligent calibration recognition model to perform parameter recognition and Kalman filter calibration to obtain filtered calibration parameters; The deviation value between the filtered calibration parameter and the initial calibration parameter is determined, and when the deviation value exceeds a preset threshold, the filtered calibration parameter is written into the storage chip, and the calibration coefficient of the host is updated.

2. The method for intelligent calibration and identification of an electric screwdriver head of an intelligent electric screwdriver according to claim 1, characterized in that: The torque coefficient, speed characteristics, temperature coefficient and stress distribution data of the electric screwdriver are collected and feature extracted to obtain a feature data set, including: The torque coefficient, speed characteristics, temperature coefficient and stress distribution of the electric screwdriver are collected through the sensor array to obtain a simulation signal data set; Performing digital conversion on the analog signal data set by a 16-bit analog-to-digital converter to obtain an initial digital signal set; According to the working speed of the electric screwdriver, the sampling frequency of the initial digital signal set is adjusted to obtain a sampled digital signal set, and the noise of the sampled digital signal set is suppressed by a digital filter to obtain a filtered signal set; The filtered signal set is subjected to multi-scale decomposition by wavelet transform to obtain multi-scale feature components, and feature vectors are extracted from the multi-scale feature components to obtain a feature data set.

3. The method for intelligent calibration and identification of an electric screwdriver head of an intelligent electric screwdriver according to claim 1, characterized in that: The step of performing standardization processing on the characteristic data set to generate a device identification code, and writing the device identification code and the characteristic data set into a storage chip of the electric screwdriver head includes: Performing zero-mean normalization processing on the feature data set to obtain standardized feature data, and generating a 128-bit original identification code based on the standardized feature data; The original identification code is encrypted using the SHA-256 algorithm to obtain a device identification code; Performing data format conversion on the standardized characteristic data to obtain storage format data, performing block processing on the storage format data, and generating data check bits to obtain block storage data; The device identification code and the standardized characteristic data are combined into a data frame, and a frame check code is generated to obtain checked data, and the checked data is written into the EEPROM storage area of ​​the storage chip according to a preset storage address.

4. The method for intelligent calibration and identification of an electric screwdriver head of an intelligent electric screwdriver according to claim 1, characterized in that: The step of constructing a projection neural network and inputting the feature data set into the projection neural network for back propagation training to obtain an intelligent calibration recognition model includes: Constructing a network structure of a projection neural network according to the dimension of the feature data set, wherein the network structure includes an input layer, a projection layer, a hidden layer, and an output layer, to obtain an initial network structure; Introducing an adaptive weight matrix into the projection layer of the initial network structure, and initializing parameters of the weight matrix to obtain an optimized network structure; Inputting the feature data set into the optimized network structure, performing feature space mapping through the weight matrix of the projection layer, and obtaining projection features; Performing a nonlinear transformation on the projection feature through the dynamic activation function of the hidden layer to obtain a hidden layer feature, and applying an L2 regularization constraint to the hidden layer feature to obtain a regularized feature; Inputting the regularized features into the output layer, calculating the error between the network output and the expected output to obtain the training error, and dynamically adjusting the learning rate based on the changing trend of the training error to obtain the optimized learning rate; The optimized learning rate is substituted into the improved BP algorithm to update the parameters of the optimized network structure to obtain an intelligent calibration recognition model.

5. The method for intelligent calibration and identification of an electric screwdriver head of an intelligent electric screwdriver according to claim 1, characterized in that: The step of inputting the real-time working parameters of the electric screwdriver head and the characteristic data set into the intelligent calibration recognition model for parameter recognition and Kalman filter calibration to obtain the filtered calibration parameters includes: Performing data preprocessing and feature extraction on the real-time working parameters of the electric screwdriver bit to obtain real-time feature data; Inputting the real-time feature data and the feature data set into the intelligent calibration recognition model, performing parameter recognition through the intelligent calibration recognition model, and obtaining initial recognition parameters; Constructing a state equation of a Kalman filter based on the initial identification parameters, taking a state transfer matrix, a control matrix and process noise as parameters of the state equation, and obtaining a state prediction value; An observation equation is constructed according to the state prediction value, and an observation matrix and an observation noise are used as parameters of the observation equation to obtain an observation prediction value; Calculating a priori error covariance matrix for the state prediction value, and calculating Kalman gain matrix in combination with the observation prediction value to obtain an optimal gain matrix; Based on the optimal gain matrix, the state variables are updated to obtain an initial state estimate, a posterior error covariance matrix is ​​calculated according to the state estimate, and iterative optimization is performed to obtain an optimal state estimate; The optimal state estimation value is used as a filtered calibration parameter to update the real-time control parameters of the electric screwdriver.

6. The method for intelligent calibration and identification of an electric screwdriver head of an intelligent electric screwdriver according to claim 5, characterized in that: The updating of the state variables based on the optimal gain matrix to obtain an initial state estimate, calculating a posterior error covariance matrix for the state estimate, and performing iterative optimization to obtain an optimal state estimate includes: Performing correction calculation on the state transfer matrix according to the optimal gain matrix to obtain a corrected state transfer matrix; Performing an update operation on the state variables based on the modified state transfer matrix to obtain an initial state estimate; Performing residual calculation on the initial state estimate and the current observation value to obtain a state update residual, and performing correction calculation on the initial state estimate according to the state update residual to obtain a corrected state estimate; Calculating a posterior error covariance matrix based on the corrected state estimate to obtain error covariance data, and performing singular value decomposition on the error covariance data to extract main eigenvalues ​​to obtain eigendecomposition results; The state estimation value is iteratively calculated according to the characteristic decomposition result. When the difference between the state estimation values ​​of two consecutive iterations is less than the target value, a converged state value is obtained, and the converged state value is output as the optimal state estimation value to complete the state update process.

7. The method for intelligent calibration and identification of an electric screwdriver head of an intelligent electric screwdriver according to claim 1, characterized in that: The determining of the deviation between the filtered calibration parameter and the initial calibration parameter, when the deviation exceeds a preset threshold, writing the filtered calibration parameter into the storage chip, and updating the calibration coefficient of the host, comprises: Performing data normalization processing on the filtered calibration parameters to obtain normalized calibration parameters, and performing synchronous normalization processing on the initial calibration parameters to obtain normalized initial parameters; Performing a difference operation based on the normalized calibration parameters and the normalized initial parameters to obtain a parameter deviation matrix, and performing a norm calculation on the parameter deviation matrix to obtain a deviation quantization value; Compare the deviation quantization value with a preset threshold, and when the deviation quantization value is greater than the preset threshold, obtain an update trigger signal; Data encoding the filtered calibration parameters according to the update trigger signal to obtain an update data packet, and writing the update data packet into the storage chip according to a preset address mapping to obtain an update confirmation signal; The update data packet is parsed into calibration coefficients based on the update confirmation signal, and the calibration coefficients are transmitted to the host for parameter update.

8. An intelligent calibration and identification device for an electric screwdriver head of an intelligent electric screwdriver, characterized in that: The device is used to perform an intelligent calibration and identification method for an electric screwdriver head of an intelligent electric screwdriver as claimed in any one of claims 1 to 7, the device comprising: The acquisition module is used to collect and extract the torque coefficient, speed characteristics, temperature coefficient and stress distribution data of the electric screwdriver bit to obtain a feature data set; A processing module, used for performing standardization processing on the characteristic data set, generating a device identification code, and writing the device identification code and the characteristic data set into a storage chip of the electric screwdriver; A training module is used to construct a projection neural network and input the feature data set into the projection neural network for back propagation training to obtain an intelligent calibration recognition model; A transmission module, used for transmitting the device identification code and the feature data set in the storage chip to the host when the electric screwdriver head is connected to the host; An identification module, used for inputting the real-time working parameters of the electric screwdriver bit and the characteristic data set into the intelligent calibration identification model to perform parameter identification and Kalman filter calibration to obtain filtered calibration parameters; The updating module is used to determine the deviation value between the filtered calibration parameter and the initial calibration parameter, and when the deviation value exceeds a preset threshold, write the filtered calibration parameter into the storage chip and update the calibration coefficient of the host.

9. An electronic device, characterized in that: The electronic device comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the electronic device to execute the intelligent calibration and identification method for the electric screwdriver head of the intelligent electric screwdriver as described in any one of claims 1-7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the intelligent calibration and identification method of the electric screwdriver head of the intelligent electric screwdriver as described in any one of claims 1-7 is implemented.

Images