Screw locking detection method and device

Abstract

The invention relates to the technical field of automatic machining, and particularly provides a screw locking detection method and device. The screw locking detection method comprises the steps of obtaining a locking parameter set in a screw locking process, performing feature extraction on the locking parameter set to obtain locking features corresponding to each locking stage included in the screw locking process, and determining a locking detection result corresponding to the screw locking process according to the locking features of each locking stage. In the embodiment of the invention, the screw locking effect can be detected in real time in the screw locking process of the automatic screw locking machine, the coverage of the whole screw locking process is realized, and compared with a manual mode, the detection efficiency and reliability are improved, the detection cost is reduced, and the detection efficiency is improved. And the locking detection result is obtained through the divided locking characteristics of the plurality of locking stages, so that the detection accuracy of the screw locking effect is improved.

Description

Technical Field

[0001] The present invention relates to the field of automated processing technology, and in particular to a screw locking detection method and device. Background Art

[0002] With the development of automation technology, automatic screw locking machines are now widely used in automated production lines to replace manual screw locking, which can greatly improve production efficiency. However, traditional automatic screw locking machines can only perform the screw locking process according to pre-set programs, and the effect of screw locking cannot be effectively detected, resulting in poor product yield and connection reliability. Summary of the invention

[0003] In order to detect the screw locking effect and effectively improve the operating reliability of the automatic screw locking machine, the embodiments of the present disclosure provide a screw locking detection method and device, an automatic screw locking machine, and a storage medium.

[0004] In a first aspect, an embodiment of the present disclosure provides a screw locking detection method, which is applied to an automatic screw locking machine, and the method comprises:

[0005] Acquire a set of screw locking parameters in a screw locking process, wherein the screw locking process includes: an end actuator of the automatic screw locking machine starts to drive the screw to rotate until the end actuator is separated from the screw;

[0006] Performing feature extraction on the locking parameter set to obtain locking features corresponding to each locking stage included in the screw locking process; the screw locking process includes: at least one of the locking stages;

[0007] The locking detection result corresponding to the screw locking process is determined according to the locking characteristics of each locking stage.

[0008] In some embodiments, the step of obtaining a set of screw locking parameters during the screw locking process includes:

[0009] During the screw locking process, the locking parameters of the end actuator at each acquisition moment are collected at a preset sampling frequency to obtain the locking parameter set during the screw locking process.

[0010] In some implementations, the extracting features of the locking parameter set to obtain locking features corresponding to each locking stage included in the locking process includes:

[0011] According to the preset execution parameters of the end actuator, the locking parameter set of the screw locking process is divided into stages to obtain a locking parameter subset corresponding to each locking stage; the execution parameters include the rotation angle or rotation duration of the end actuator;

[0012] For each locking stage, the locking feature corresponding to the locking stage is determined according to the locking parameters included in the locking parameter subset of the locking stage.

[0013] In some embodiments, the locking parameters include the rotation angle of the end actuator, the torque output by the end actuator, and the rotation speed of the end actuator during the screw locking process; the determining the locking feature corresponding to the locking stage according to each locking parameter included in the locking parameter subset of the locking stage includes:

[0014] The torque characteristic corresponding to the locking stage is obtained by calculating the average value of the torque at each acquisition moment in the locking parameter subset;

[0015] The average value is calculated according to the rotation speed at each acquisition moment in the locking parameter subset to obtain the rotation speed characteristic corresponding to the locking stage;

[0016] Determine the angle value corresponding to the maximum torque in the locking stage according to the locking parameter subset;

[0017] The locking feature of the locking stage is obtained according to the torque feature, the rotation speed feature and the angle value corresponding to the maximum torque.

[0018] In some embodiments, determining the locking detection result corresponding to the screw locking process according to the locking features of each locking stage includes:

[0019] The locking features of each locking stage are input into a pre-trained locking detection network to obtain the locking detection result output by the locking detection network.

[0020] In some embodiments, the process of training the lock detection network includes:

[0021] Acquire a sample data set, the sample data set comprising a plurality of sample data, each sample data comprising a locking parameter set and a result label in a screw locking process, the locking parameter set comprising a locking parameter subset in each locking stage;

[0022] Feature extraction is performed based on the locking parameter set of each sample data to obtain multiple categories of candidate features corresponding to each locking stage;

[0023] Inputting multiple categories of candidate features of each sample data into the lock-attachment detection network to be trained, and obtaining the detection result output by the lock-attachment detection network and the confidence of each candidate feature;

[0024] The network parameters of the lock detection network are adjusted based on the difference between the detection result and the result label, and the candidate features of the next iteration are screened based on the confidence of each candidate feature until the convergence condition is met, thereby obtaining a trained lock detection network.

[0025] In some embodiments, the method described in the present disclosure further includes:

[0026] In response to the locking detection result being a locking abnormality, an alarm message is generated, and / or the locking detection result is displayed on a display device of the automatic screw locking machine.

[0027] In a second aspect, the present disclosure provides a screw locking detection device, which is applied to an automatic screw locking machine, and the device includes:

[0028] The parameter acquisition module is configured to acquire a set of screw locking parameters during a screw locking process, wherein the screw locking process includes: a process in which an end actuator of the automatic screw locking machine starts to drive the screw to rotate until the end actuator is separated from the screw;

[0029] A feature extraction module is configured to perform feature extraction on the locking parameter set to obtain locking features corresponding to each locking stage included in the screw locking process; the screw locking process includes: at least one of the locking stages;

[0030] The locking detection module is configured to determine a locking detection result corresponding to the screw locking process according to the locking characteristics of each locking stage.

[0031] In some implementations, the parameter acquisition module is configured to:

[0032] During the screw locking process, the locking parameters of the end actuator at each acquisition moment are collected at a preset sampling frequency to obtain the locking parameter set during the screw locking process.

[0033] In some embodiments, the feature extraction module is configured to:

[0034] According to the preset execution parameters of the end actuator, the locking parameter set of the screw locking process is divided into stages to obtain a locking parameter subset corresponding to each locking stage; the execution parameters include the rotation angle or rotation duration of the end actuator;

[0035] For each locking stage, the locking feature corresponding to the locking stage is determined according to the locking parameters included in the locking parameter subset of the locking stage.

[0036] In some embodiments, the locking parameters include the rotation angle of the end effector during the screw locking process, the torque output by the end effector, and the rotation speed of the end effector; the feature extraction module is configured as follows:

[0037] The torque characteristic corresponding to the locking stage is obtained by calculating the average value of the torque at each acquisition moment in the locking parameter subset;

[0038] The average value is calculated according to the rotation speed at each acquisition moment in the locking parameter subset to obtain the rotation speed characteristic corresponding to the locking stage;

[0039] Determine the angle value corresponding to the maximum torque in the locking stage according to the locking parameter subset;

[0040] The locking feature of the locking stage is obtained according to the torque feature, the rotation speed feature and the angle value corresponding to the maximum torque.

[0041] In some embodiments, the lock detection module is configured to:

[0042] The locking features of each locking stage are input into a pre-trained locking detection network to obtain the locking detection result output by the locking detection network.

[0043] In some embodiments, the lock detection module is configured to:

[0044] Acquire a sample data set, the sample data set comprising a plurality of sample data, each sample data comprising a locking parameter set and a result label in a screw locking process, the locking parameter set comprising a locking parameter subset in each locking stage;

[0045] Feature extraction is performed based on the locking parameter set of each sample data to obtain multiple categories of candidate features corresponding to each locking stage;

[0046] Inputting multiple categories of candidate features of each sample data into the lock-attachment detection network to be trained, and obtaining the detection result output by the lock-attachment detection network and the confidence of each candidate feature;

[0047] The network parameters of the lock detection network are adjusted based on the difference between the detection result and the result label, and the candidate features of the next iteration are screened based on the confidence of each candidate feature until the convergence condition is met, thereby obtaining a trained lock detection network.

[0048] In some embodiments, the lock detection module is configured to:

[0049] In response to the locking detection result being a locking abnormality, an alarm message is generated, and / or the locking detection result is displayed on a display device of the automatic screw locking machine.

[0050] In a third aspect, the present disclosure provides an automatic screw locking machine, comprising:

[0051] Processor; and

[0052] A memory storing computer instructions, wherein the computer instructions are used to enable a processor to execute the method according to any embodiment of the first aspect.

[0053] In a fourth aspect, an embodiment of the present disclosure provides a storage medium storing computer instructions, wherein the computer instructions are used to enable a computer to execute the method according to any embodiment of the first aspect.

[0054] The screw locking detection method of the embodiment of the present disclosure includes obtaining a locking parameter set in the screw locking process, extracting features of the locking parameter set to obtain locking features corresponding to each locking stage included in the screw locking process, and determining the locking detection result corresponding to the screw locking process according to the locking features of each locking stage. In the embodiment of the present disclosure, the screw locking effect can be detected in real time during the screw locking process of the automatic screw locking machine to achieve coverage of the entire screw locking process, which improves the detection efficiency and reliability and reduces the detection cost compared to the manual method, and obtains the locking detection result through the locking features of the multiple locking stages, thereby improving the detection accuracy of the screw locking effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the specific embodiments of the present disclosure or the technical solutions in the prior art, the drawings required for use in the specific embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present disclosure. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0056] Figure 1 It is a structural block diagram of an automatic screw locking machine according to some embodiments of the present disclosure.

[0057] Figure 2 is a torque curve diagram of a screw locking process in an exemplary embodiment of the present disclosure.

[0058] Figure 3 It is a flow chart of a screw locking detection method according to some embodiments of the present disclosure.

[0059] Figure 4 It is a flow chart of a screw locking detection method according to some embodiments of the present disclosure.

[0060] Figure 5 It is a flow chart of a screw locking detection method according to some embodiments of the present disclosure.

[0061] Figure 6 It is a structural block diagram of a screw locking detection device in some embodiments of the present disclosure.

[0062] Figure 7 It is a structural block diagram of an automatic screw locking machine according to some embodiments of the present disclosure. DETAILED DESCRIPTION

[0063] The technical solution of the present disclosure will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described implementation is a part of the implementation of the present disclosure, rather than all the implementations. Based on the implementation in the present disclosure, all other implementations obtained by ordinary technicians in the field without creative work are within the scope of protection of the present disclosure. In addition, the technical features involved in the different implementations of the present disclosure described below can be combined with each other as long as they do not conflict with each other.

[0064] 3C electronic products refer to the three categories of electronic products: computers, communications, and consumer electronics. Nowadays, with the development of automation technology, automatic screw locking machines are widely used on 3C production lines to replace manual screw locking.

[0065] Taking smart phones as an example, the various components of mobile phones are often connected with screws. Therefore, the production of mobile phones involves a large number of screw locking operations. The automatic screw locking machine can automatically complete the screw picking, positioning, tightening and other operations, which greatly improves the production efficiency compared to manual work.

[0066] Traditional automatic screw locking machines can only lock screws according to pre-set programs with preset torque, speed, rotation angle, etc., and cannot effectively detect the actual locking effect. However, in the actual operation process, due to interference from external factors, the screw locking process is prone to problems such as screw tilt and thread slippage. For example, if the screw holes of the connector are not aligned, the automatic screw locking machine is prone to screw tilt after unloading. At this time, after the screw is locked, it may cause the screw to protrude from the screw hole, affecting subsequent assembly operations and even causing product scrapping.

[0067] The traditional method of testing the screw locking effect is manual sampling, that is, manually randomly selecting a part of the products to test whether the screw locking effect meets the standard. This method cannot cover all products, the detection reliability is very poor, and it is more dependent on manual experience, and the detection cost and efficiency are not high.

[0068] Based on the defects mentioned in the above-mentioned related technologies, the embodiments of the present disclosure provide a screw locking detection method and device, an automatic screw locking machine, and a storage medium, which aim to achieve real-time detection and feedback of the locking effect during the screw locking process, realize automated screw locking effect detection, and achieve automated full detection of screw locking, thereby improving detection efficiency and reducing detection costs.

[0069] Figure 1 The functional module diagram of the automatic screw locking machine in some embodiments of the present disclosure is shown below. Figure 1 The structure of the automatic screw locking machine according to the embodiment of the present disclosure is described.

[0070] exist Figure 1 In the example, the automatic screw locking machine disclosed in the present invention mainly includes three parts, namely a data acquisition and processing module 1, a locking result determination module 2 and an alarm notification module 3.

[0071] The data acquisition and processing module 1 is mainly used to acquire relevant locking parameters in the screw locking process and perform some preprocessing on the relevant locking parameters, such as filtering processing, denoising processing, etc.

[0072] See also Figure 1 As shown, the data acquisition and processing module 1 includes an end actuator 13, which refers to the relevant mechanical and electrical structure used by the automatic screw locking machine to perform material picking, positioning and tightening, which may include software and hardware structures related to the use of electric screwdriver bits, industrial cameras, laser ranging, etc. For example, during the screw locking process, the end actuator 13 can take out the screws from the material bin, then put the screws into the screw holes of the parts, and then automatically tighten the screws to complete a screw locking process. For the specific structure and control principle of the end actuator 13, those skilled in the art can undoubtedly understand and implement it with reference to the relevant technology, and this disclosure will not go into details.

[0073] Continue to refer to Figure 1 As shown, the data acquisition and processing module 1 also includes relevant sensors required for collecting locking parameters, such as a torque sensor 11, an angle sensor 12, etc. The torque sensor 11 can collect the output torque in real time during the process of screw locking by the end actuator 13. The angle sensor 12 can collect the rotation angle of the electric screwdriver bit in real time during the process of screw locking by the end actuator 13. Of course, those skilled in the art can understand that the sensors related to collecting locking parameters are not limited to the torque sensor 11 and the angle sensor 12, and more types of sensors may be included if necessary, such as an angular velocity sensor, etc., which will not be elaborated in this disclosure.

[0074] For the first control unit 14 included in the data acquisition and processing module 1, the first control unit 14 can perform pre-processing such as filtering and noise reduction on the locking data collected by each sensor, and send the processed locking parameters to the locking result determination module 2. At the same time, the first control unit 14 can also control the terminal execution mechanism 13 to execute a predetermined locking operation process.

[0075] exist Figure 1 In the example, the locking result determination module 2 is mainly used to determine whether there is an abnormality in the current screw locking process according to the locking parameters collected by the data acquisition and processing module 1. In this example, the locking result determination module 2 includes a data preprocessing unit 21, a central processing unit 22, a display and output device 23, and a second control unit 24. The data preprocessing unit 21 is used to perform preprocessing according to the received locking parameters, and the preprocessing may include, for example, exception processing, missing value filling, normalization processing, etc. The central processing unit 22 can then execute one or more method steps in the following implementation of the present disclosure based on the preprocessed locking parameters to obtain the locking detection results, which are described below. The display and output device 23 can provide an output display for the locking detection results and / or locking parameters. For example, the display and output device 23 can be a display screen. In the case of abnormal locking detection results, the relevant locking parameter information is displayed on the display screen for staff to view.

[0076] exist Figure 1 In the example, the alarm notification module 3 is used to output corresponding alarm information when the locking detection result is abnormal. In this example, the alarm notification module 3 includes a bad alarm notification 31, a third control unit 32 and a human-computer interaction unit 33. The third control unit 32 is communicably connected with the other two modules through a data bus structure, so as to receive the locking detection result of the locking result determination module and generate corresponding alarm information. The alarm information can be, for example, sound and light alarm information, so that the bad alarm notification 31 can be a sound and light alarm, which issues an alarm according to the alarm information to prompt the staff that the locking is abnormal. The human-computer interaction unit 33 can provide an interactive interface between the staff and the automatic screw locking machine, so that when the locking detection result is abnormal, the staff can retrieve the corresponding locking parameters through the human-computer interaction unit 33 and output them on the display and output device 23.

[0077] The above is an explanation of some structures and principles of the automatic screw locking machine in some embodiments of the present disclosure. For those not fully described in detail, those skilled in the art can understand and implement them by referring to relevant technologies, and the present disclosure will not elaborate on them.

[0078] Based on the aforementioned automatic screw locking machine, the embodiment of the present disclosure provides a screw locking detection method, which can be applied to the aforementioned automatic screw locking machine, and the locking result determination module performs processing. To facilitate understanding of the principle and process of the screw locking detection method disclosed in the present disclosure, some technical terms and principles of the method disclosed in the present disclosure are explained below.

[0079] The primary screw locking process of the automatic screw locking machine refers to the process from the end actuator starting to drive the screw to rotate after the material picking and positioning processes are completed to the end actuator disengaging the screw. In this process, the end actuator needs to complete the locking of the screw according to the predetermined execution procedure.

[0080] The inventor of this case has found through research that during a complete screw locking process, the torque output by the end actuator is not uniform, but will output different torques according to the different penetration levels of the screw. In other words, a complete screw locking process can be divided into one or more locking stages according to the output torque.

[0081] For example Figure 2 The torque variation curve during the screw locking process is shown in the figure. Figure 2 The horizontal axis represents the rotation angle of the end actuator in revolutions (r), and the vertical axis represents the torque output by the end actuator in Newton-meters (Nm).

[0082] By observation Figure 2 It can be seen from the curve that in the "Stage 1" process, the screw is in the threading stage. This stage mainly involves screwing the screw in to a certain length to complete the screw fixation, so only a small and stable torque needs to be output. In the "Stage 2" process, the screw needs to be quickly screwed into the screw hole, so the output torque is larger and more volatile than in "Stage 1". In the "Stage 3" process, the screw body has basically been screwed into the screw hole, and the screw needs to be tightened further to press the nut and the component to complete the locking process. Therefore, a larger torque needs to be output to complete the screw tightening. At this stage, the torque gradually increases to the maximum.

[0083] In the embodiment of the present disclosure, Figure 2 Taking the screw locking process shown in the figure as an example, "Stage 1" is defined as the threading stage, "Stage 2" is defined as the quick screwing stage, and "Stage 3" is defined as the tightening stage. Figure 2 As can be seen from the example, the entire screw locking process includes multiple time-continuous locking stages, which in this example are the threading stage, the rapid screwing stage and the tightening stage.

[0084] Therefore, in the embodiments of the present disclosure, the characteristics of multiple locking stages can be considered to describe the entire screw locking process. For example, in some embodiments of the present disclosure, the locking parameters of the entire screw locking process can be collected, and then the locking characteristics of each locking stage can be extracted respectively. Based on the locking characteristics of each locking stage, it is judged whether the screw locking process meets the requirements. Figure 3 An example is provided to illustrate.

[0085] like Figure 3 As shown, in some embodiments, the screw locking detection method of the present disclosure example includes:

[0086] S310: Obtain a set of screw locking parameters during the screw locking process.

[0087] Combination Figure 1 and Figure 2 It can be understood that when the automatic screw locking machine is operating, for a certain screw locking process, the data acquisition and processing module 1 can collect one or more locking parameters in the screw locking process in real time. The set of locking parameters collected in chronological order is the locking parameter set described in the present disclosure.

[0088] For example, in one example, the locking parameter includes the torque output by the end actuator 13, so that during the screw locking process, the torque sensor 11 can continuously collect the torque value of the end actuator 13 at a preset sampling frequency, and the set of these torque values ​​arranged in time order is the locking parameter set. For example, the torque sensor 11 collects the torque value every 10ms. Assuming that the entire screw locking process lasts 200ms, 20 torque values ​​can be collected, and the set composed of these 20 torque values ​​is the locking parameter set.

[0089] Of course, those skilled in the art will understand that the locking parameters of the embodiments of the present disclosure are not limited to torque, but may also include any parameters related to the locking process, such as the rotation angle and angular velocity of the end actuator, and thus the type of locking parameters may also include one or more, which is not limited by the present disclosure.

[0090] In some embodiments, after acquiring the locking parameter set, the locking parameters in the locking parameter set may be preprocessed. The preprocessing may include, for example, filtering, denoising, outlier processing, missing value filling, etc., which is not limited in the present disclosure.

[0091] S320: Extract features from the locking parameter set to obtain locking features corresponding to each locking stage included in the screw locking process.

[0092] Combined with the above Figure 2It can be seen that the entire screw locking process may include multiple locking stages, and each locking stage has unique locking parameter characteristics. Therefore, in the embodiment of the present disclosure, after the locking parameter set is obtained through the aforementioned S310, the locking parameters in the locking parameter set represent the entire screw locking process, so the locking parameters in the locking parameter set need to be divided into locking stages.

[0093] by Figure 2 Taking the example scenario as an example, the elements in the locking parameter set represent the entire screw locking process. Therefore, the locking parameter set needs to be divided into three subsets according to "stage one" to "stage three", and each subset represents the locking parameter subset of the corresponding locking stage.

[0094] In some embodiments, in combination Figure 2 As shown, the screw locking process can be divided into stages according to the rotation angle of the end actuator in advance, so as to obtain the rotation angle corresponding to each locking stage. Figure 2 In the example, the rotation angle corresponding to stage 1 is 0-380r, the rotation angle corresponding to stage 2 is 380-950r, and the rotation angle corresponding to stage 3 is 950r-1200r.

[0095] Therefore, after obtaining the locking parameter set, the locking parameter set can be divided according to the rotation angle, that is, the locking parameters corresponding to the rotation angle of 0~380r are divided into the locking parameter subset of stage one, the locking parameters corresponding to the rotation angle of 380~950r are divided into the locking parameter subset of stage two, and the locking parameters corresponding to the rotation angle of 950r~1200r are divided into the locking parameter subset of stage three.

[0096] In other embodiments, the screw locking process can also be divided into stages according to the working time of the end actuator in advance, so as to obtain the time corresponding to each locking stage. For example, in one example, the time corresponding to stage 1 is 0 to 50 ms, the time corresponding to stage 2 is 50 ms to 150 ms, and the time corresponding to stage 3 is 150 ms to 200 ms.

[0097] Therefore, after obtaining the locking parameter set, the locking parameter set can be divided according to the rotation duration, that is, the locking parameters corresponding to the rotation duration of 0 to 50ms are divided into the locking parameter subset of stage one, the locking parameters corresponding to the rotation angle of 50ms to 150ms are divided into the locking parameter subset of stage two, and the locking parameters corresponding to the rotation angle of 950r to 1200r are divided into the locking parameter subset of stage three.

[0098] Of course, those skilled in the art can understand that the manner of dividing the locking parameter set into stages is not limited to the above example, and the present disclosure does not impose any limitation on this.

[0099] Combining the above, it can be seen that for each locking stage, the corresponding locking features are also different.

[0100] Therefore, after dividing the locking parameter set into stages, feature extraction can be performed based on the locking parameter subset of each locking stage to obtain the locking features corresponding to each locking stage, and the locking features can reflect the locking characteristics of the locking stage.

[0101] For example, Figure 2 For example, the torque corresponding to each locking stage is quite different, so the torque can be used as the locking feature, that is, the torque information of each locking stage is used to represent the locking feature of the locking stage. For example, in one example, the average value of all torques in each locking stage is calculated and used as the locking feature of each locking stage.

[0102] Of course, those skilled in the art will understand that the type of locking parameters used to reflect the locking characteristics of each locking stage is not limited to torque, but may also include, for example, rotation angle, angular velocity, etc. The present disclosure does not limit the number and type of specific locking parameters included in the locking characteristics. The present disclosure describes the feature extraction process below.

[0103] S330: Determine a locking detection result corresponding to the screw locking process according to the locking features of each locking stage.

[0104] From the foregoing, it can be seen that the locking features of each locking stage can reflect the locking conditions of that locking stage. Therefore, in the embodiments of the present disclosure, after obtaining the locking features of each locking stage, it can be determined whether there are any abnormalities in the entire screw locking process based on the locking features of each locking stage, thereby obtaining the corresponding locking detection results.

[0105] In some embodiments, a locking detection network based on a deep neural network (DNN) can be used, and the locking features of each locking stage are input into the locking detection network to obtain the locking detection results predicted and output by the locking detection network. The following embodiments of the present disclosure illustrate this.

[0106] It can be understood that in the embodiment of the present disclosure, by dividing the screw locking process into multiple locking stages and extracting the locking features of each locking stage respectively, the locking features of each locking stage can more accurately reflect the locking situation of the screw locking process, thereby obtaining a more accurate locking detection result.

[0107] After the locking detection result is obtained, the locking detection result can be fed back to the alarm notification module 3, and when the locking detection result is abnormal, the corresponding alarm information is output and displayed, which is explained in the following embodiments of the present disclosure.

[0108] From the above, it can be seen that in the embodiments of the present disclosure, the screw locking effect can be detected in real time during the screw locking process of the automatic screw locking machine, thereby achieving coverage of the entire screw locking process, which improves the detection efficiency and reliability and reduces the detection cost compared to the manual method, and obtains the locking detection results through the locking features of multiple locking stages, thereby improving the detection accuracy of the screw locking effect.

[0109] In some embodiments of the present disclosure, during the screw locking process, the locking parameters in the acquired locking parameter set include the rotation angle Q of the end actuator, the torque N output by the end actuator, and the rotation speed R of the end actuator. Of course, the locking parameters may also include more parameters related to screw locking, which is not limited by the present disclosure.

[0110] In the disclosed example, an angle sensor, a torque sensor and an angular velocity sensor are provided on the end actuator of the automatic screw locking machine, so that during the entire screw locking process, the angle sensor, the torque sensor and the angular velocity sensor can periodically collect the rotation angle Q, the torque N and the rotation speed R of the end actuator according to a preset sampling frequency.

[0111] In the embodiment of the present disclosure, the time when the sensor collects data each time is defined as the collection moment. The frequency of the collection moment can be selected according to the needs of the specific scenario. For example, in one example, data can be collected every 10ms, that is, the time interval between two adjacent collection moments is 10ms.

[0112] For each acquisition time T i , the angle sensor can collect the current rotation angle Q i The torque sensor can collect the current output torque N i , the angular velocity sensor can collect the current rotation speed R i The parameter set obtained by periodically collecting the locking parameters of the entire screw locking process is the locking parameter set M. That is, in the example disclosed in the present disclosure, the locking parameter set M includes each collection time T i The corresponding locking parameters include the rotation angle Q i , torque N i and the speed R i .

[0113] It is worth noting that in some embodiments of the present disclosure, after each sensor acquires the locking parameters, it can use corresponding denoising and filtering algorithms to pre-process the acquired data, so as to improve the data accuracy. For example, in some embodiments, the locking parameters with obvious abnormalities can be deleted, and then the missing locking parameters can be filled by filling the missing values, and the method of filling the missing values ​​can be to average multiple adjacent parameter values. For example, in other embodiments, all the acquired locking parameters can be normalized to eliminate the dimensional effect and facilitate subsequent calculations. The method of data preprocessing can undoubtedly be understood by those skilled in the art, and the present disclosure will not elaborate on it.

[0114] After determining the locking parameter set M of the entire screw locking process, feature extraction can be performed on each locking stage based on the locking parameter set M. Figure 4 Provide explanation.

[0115] like Figure 4 As shown, in some embodiments, the screw locking detection method of the present disclosure example, the process of extracting features from the locking parameter set includes:

[0116] S410: Divide the locking parameter set of the screw locking process into stages according to the preset execution parameters of the end actuator, and obtain a locking parameter subset corresponding to each locking stage.

[0117] As can be seen from the foregoing, the locking parameter set includes the locking parameters corresponding to each acquisition moment in the entire screw locking process, and in the embodiment of the present disclosure, the entire screw locking process needs to be divided into multiple locking stages.

[0118] The execution parameter of the end actuator refers to a parameter related to the end actuator for dividing the screw fastening process into stages. In the embodiment of the present disclosure, the execution parameter can be any parameter suitable for dividing the screw fastening process into stages, such as the rotation angle of the end actuator, or the rotation time of the end actuator.

[0119] For example, in some embodiments, see Figure 2 As shown, the screw fastening process can be divided into three stages according to the rotation angle of the end actuator, that is, the execution parameter of the end actuator is the rotation angle. Figure 2 In the example, it can be predetermined that the rotation angle corresponding to stage 1 is 0-380r, the rotation angle corresponding to stage 2 is 380-950r, and the rotation angle corresponding to stage 3 is 950r-1200r. Therefore, after obtaining the locking parameter set, the rotation angle Q in the locking parameter set can be used to determine the rotation angle. iThe locking stages are divided, that is, the locking parameters corresponding to the rotation angle of 0-380r are divided into the locking parameter subset of stage one, the locking parameters corresponding to the rotation angle of 380-950r are divided into the locking parameter subset of stage two, and the locking parameters corresponding to the rotation angle of 950r-1200r are divided into the locking parameter subset of stage three.

[0120] In other embodiments, the screw locking process can be divided into: Figure 2 The three stages shown, that is, the execution parameters of the end actuator are the rotation duration. For example, in one example, it can be predetermined that the time corresponding to stage one is 0-50ms, the time corresponding to stage two is 50ms-150ms, and the time corresponding to stage three is 150ms-200ms. Thus, after obtaining the locking parameter set, the locking parameter set can be divided according to the rotation duration, that is, the locking parameters corresponding to the rotation duration of 0-50ms are divided into the locking parameter subset of stage one, the locking parameters corresponding to the rotation angle of 50ms-150ms are divided into the locking parameter subset of stage two, and the locking parameters corresponding to the rotation angle of 950r-1200r are divided into the locking parameter subset of stage three.

[0121] In the embodiment of the present disclosure, by dividing the locking parameter set into stages, the locking parameter set corresponding to the entire screw locking process can be divided into multiple locking parameter subsets corresponding to each locking stage. For example, in the above example, the locking parameter set M can be divided into the following three locking parameter subsets:

[0122] The locking parameter subset M1 includes each acquisition time T in stage 1 i The locking parameters include the rotation angle Q i , torque N i and the speed R i ;

[0123] The locking parameter subset M2 includes each acquisition time T in stage 2 i The locking parameters include the rotation angle Q i , torque N i and the speed R i ;

[0124] The locking parameter subset M3 includes each acquisition time T in stage 3 i The locking parameters include the rotation angle Q i , torque N i and the speed R i .

[0125] S420: For each locking stage, determine a locking feature corresponding to the locking stage according to the locking parameters included in the locking parameter subset of the locking stage.

[0126] In the disclosed embodiment, after obtaining the locking parameter subset of each locking stage, feature extraction can be performed on each locking parameter subset respectively, so as to obtain the locking feature corresponding to each locking stage.

[0127] It can be understood that the locking feature refers to the relevant features that can reflect the characteristics of the locking stage. There are many types of locking features that can be selected, such as average torque, average speed, angle value corresponding to maximum torque, maximum torque value, maximum angle value, maximum speed value, torque standard deviation, speed standard deviation, etc. The present disclosure does not impose any restrictions on this.

[0128] It is worth noting that although there are many types of locking features that can reflect the characteristics of the locking stage, each locking feature has a different correlation with the locking stage, so the accuracy of the results obtained by selecting different locking features will also be different. Therefore, in the embodiments of the present disclosure, a locking detection network based on a random forest architecture can be used in advance to screen various locking features and select the locking feature with the best effect, which is explained in the following embodiments of the present disclosure.

[0129] For example, in some embodiments, the locking characteristics of each locking stage include the average torque N in the locking stage. ave , average speed R ave And the angle value Q corresponding to the maximum torque Nmax .

[0130] Taking the implementation method shown in the previous text as an example, for stage 1, all torques N in the locking parameter subset M1 can be calculated according to the locking parameter subset M1 corresponding to stage 1. i The average value of the torque N ave_1 , the average torque N ave_1 This is the torque characteristic corresponding to stage 1. Similarly, calculate all the speeds R in the locking parameter subset M1. i The average speed R ave_1 , the average speed R ave_1 This is the speed characteristic corresponding to stage 1. At the same time, the maximum torque N is determined according to the locking parameter subset M1. max_1 The corresponding rotation angle, get the angle value Q corresponding to the maximum torque Nmax_1 .

[0131] In the obtained stage 1, the torque characteristic N ave_1 , speed characteristic R ave_1 And the angle value Q corresponding to the maximum torque Nmax_1Then, the locking feature P1 corresponding to the first stage can be obtained, which is expressed as: P1 = {N ave_1 , R ave_1 , Q Nmax_1}.

[0132] For stage 2, all torques N in the locking parameter subset M2 can be calculated based on the locking parameter subset M2 corresponding to stage 2. i The average value of the torque N ave_2 , the average torque N ave_2 This is the torque characteristic corresponding to stage 2. Similarly, calculate all the speeds R in the locking parameter subset M2. i The average speed R ave_2 , the average speed R ave_2 This is the speed characteristic corresponding to stage 2. At the same time, the maximum torque N is determined according to the locking parameter subset M2. max_2 The corresponding rotation angle, get the angle value Q corresponding to the maximum torque Nmax_2 .

[0133] In the obtained stage 2, the torque characteristic N ave_2 , speed characteristic R ave_2 And the angle value Q corresponding to the maximum torque Nmax_2 Then, the locking feature P2 corresponding to the second stage can be obtained, which is expressed as: P2 = {N ave_2 , R ave_2 , Q Nmax_2}.

[0134] For stage three, all torques N in the locking parameter subset M3 can be calculated based on the locking parameter subset M3 corresponding to stage three. i The average value of the torque N ave_3 , the average torque N ave_3 This is the torque characteristic corresponding to stage 3. Similarly, calculate all the speeds R in the locking parameter subset M3. i The average speed R ave_3 , the average speed R ave_3 This is the speed characteristic corresponding to stage 3. At the same time, the maximum torque N is determined according to the locking parameter subset M3. max_3 The corresponding rotation angle, get the angle value Q corresponding to the maximum torque Nmax_3 .

[0135] In the obtained stage 3, the torque characteristic N ave_3 , speed characteristic R ave_3 And the angle value Q corresponding to the maximum torque Nmax_3 Then, the locking feature P3 corresponding to the third stage can be obtained, which is expressed as: P3 = {N ave_3 , R ave_3 , QNmax_3}.

[0136] Through the above process, after feature extraction of the locking parameter subsets of each locking stage, the locking features of each locking stage can be obtained, which are:

[0137] Phase 1: P1 = {N ave_1 , R ave_1 , Q Nmax_1}

[0138] Phase 2: P2 = {N ave_2 , R ave_2 , Q Nmax_2}

[0139] Phase 3: P3 = {N ave_3 , R ave_3 , Q Nmax_3}

[0140] After obtaining the locking features of each locking stage, the locking detection result can be determined according to the locking features of each locking stage. In some embodiments, a DNN-based locking detection network can be used to obtain the locking detection result predicted and output by the locking detection network by taking the locking features of each locking stage as the input of the locking detection network.

[0141] In some embodiments of the present disclosure, the lock detection network can use a neural network based on a random forest architecture. Random forest is an ensemble learning method. Its basic unit is a decision tree. Each decision tree can be regarded as a classifier. Therefore, for one input data, n decision trees will have n classification results. Then all classification voting results are integrated, and the category with the most votes is selected as the final output result. For the specific algorithm and structure of random forest, those skilled in the art can understand it by referring to the relevant technology, and the present disclosure will not elaborate on it.

[0142] In the embodiments of the present disclosure, a locking detection network can be pre-constructed and trained. After the locking features of each locking stage are obtained through the aforementioned method process, each locking feature can be used as input data of the locking detection network. Each locking feature is input into the locking detection network, and the locking detection network predicts and outputs the final locking detection result.

[0143] In some embodiments, the output of the lock detection network may be a simple binary classification result, that is, the lock detection result output by the lock detection network only includes two classification results, "normal" and "abnormal". In other embodiments, more classification results may be set in the output layer of the lock detection network, so that the lock detection result may include corresponding fault types, for example, the lock detection result may include "normal", "not locked in", "not locked", "tilted lock", "floating height", etc. In the case of abnormal lock, the fault type can be directly obtained through the output result of the lock detection network.

[0144] In some embodiments of the present disclosure, after obtaining the lock detection result output by the lock detection network, if the lock detection result is a lock abnormality, a corresponding alarm message can be generated. Figure 1 As shown, in the case of abnormal locking detection results, the alarm notification module 3 can generate corresponding alarm information, which can be sound and light alarm information, and the sound and light alarm information can be used to remind the staff of abnormal locking. Therefore, the staff can perform human-computer interaction through the human-computer interaction unit 33, and the display and output device 23 of the locking result determination module 2 can output and display the relevant information of the current screw locking process, such as the locking detection results, locking parameter set, etc., so that the staff can quickly locate and eliminate risks.

[0145] like Figure 5 As shown, in some embodiments, the screw locking detection method of the present disclosure example, the process of training the locking detection network includes:

[0146] S510: Obtain a sample data set.

[0147] It can be understood that the sample data set refers to a data set used to perform network training on the locking detection network, which includes a plurality of sample data, including sample data with normal locking and sample data with abnormal locking.

[0148] Each sample data includes a set of locking parameters and a result label during the screw locking process. The locking parameter set is the set of locking parameters collected in real time during the screw locking process by the aforementioned method. The result label refers to the true judgment result (GT, Ground Truth) of the screw locking process obtained by manual labeling. The result label can be represented by 0 or 1, for example, 0 means normal locking and 1 means abnormal locking.

[0149] S520 , extracting features according to the locking parameter set of each sample data to obtain multiple categories of candidate features corresponding to each locking stage.

[0150] In some embodiments, after constructing the sample data set, some preprocessing operations may be performed on the data in the sample data set, and the preprocessing operations may include, for example, outlier processing, missing value filling, normalization, etc. Outlier processing refers to deleting the locking parameters with obvious abnormalities in the sample data, missing value filling refers to filling the missing or deleted locking parameters in the sample data, and the filling method may be to find the average value of multiple adjacent locking parameters, and normalization refers to converting all locking parameters of the sample data into scalars between (0, 1) or (-1, 1) to eliminate the dimension effect.

[0151] As for the specific method of data preprocessing, those skilled in the art can undoubtedly understand it by referring to the relevant technology, and this disclosure will not elaborate on it.

[0152] In the embodiments of the present disclosure, after the sample data set is constructed, it is necessary to construct features for each sample data. It is worth noting that in some embodiments, in the process of constructing features, multiple categories of candidate features can be constructed, and the network characteristics of random forests can be used to select features with better effects from the multiple categories of candidate features as features in the actual prediction stage.

[0153] For example, in some implementations, taking a sample data as an example, firstly, the locking parameter set of the sample data can be divided into stages according to the aforementioned method process to obtain a locking parameter subset for each locking stage, which will not be elaborated in the present disclosure.

[0154] Then, taking a certain locking stage as an example, according to the locking parameters included in the locking parameter subset of the locking stage, multiple categories of candidate features can be pre-constructed, and each category of candidate features can be any feature related to the screw locking process. For example, in one example, the multiple categories of candidate features can be: average torque, average speed, maximum torque, maximum speed, torque standard deviation, speed standard deviation, rotation angle corresponding to maximum torque, etc. In other words, for each locking stage, the locking parameter subset of the locking stage can be used to calculate the feature values ​​containing all the above categories respectively, and obtain multiple categories of candidate features.

[0155] S530, inputting multiple categories of candidate features of each sample data into the lock and attachment detection network to be trained, and obtaining the detection result output by the lock and attachment detection network and the confidence of each candidate feature.

[0156] In the disclosed embodiment, when the lock and attachment detection network is trained, the input of the network is multiple categories of candidate features for each sample data, and these candidate features are input into the lock and attachment detection network to be trained. For the random forest algorithm model, in addition to obtaining the detection results of the predicted output of the lock and attachment detection network in this round of iteration, the confidence corresponding to each candidate feature can also be obtained.

[0157] The detection result output by the locking detection network refers to the detection result of the screw locking process of the sample data, such as normal locking, abnormal locking, etc. The confidence of each candidate feature indicates the importance of the corresponding candidate feature. The higher the confidence, the higher the correlation between the candidate feature and the detection result, and it is an important feature; conversely, the lower the confidence, the lower the correlation between the candidate feature and the detection result, and it is an unimportant feature.

[0158] S540, adjusting the network parameters of the lock and attachment detection network based on the difference between the detection result and the result label, and screening the candidate features of the next iteration based on the confidence of each candidate feature until the convergence condition is met, thereby obtaining the trained lock and attachment detection network.

[0159] It can be understood that the detection result output by the lock-attachment detection network refers to the lock-attachment detection result corresponding to this round of iteration, which is the predicted value. The result label is the lock-attachment result manually marked in advance, which is the true value. Therefore, the difference between the detection result and the result label is the loss of this iteration. According to the difference between the two, the network parameters of the lock-attachment detection network to be trained are adjusted by back propagation, which completes an iteration process.

[0160] In the disclosed embodiments, it is also necessary to screen the candidate features for the next iteration according to the confidence of each candidate feature in the current iteration. For example, in some embodiments, the confidence of all candidate features can be ranked from high to low, and then the candidate features with the lowest ranking can be screened out. That is, when the feature is extracted in the next iteration, the candidate features with the lowest confidence will no longer be constructed. By executing the above process in a loop, the network parameters of the lock detection network can be continuously tuned in a loop, and the candidate features with lower confidence can be continuously screened until the number of retained candidate features reaches a preset number. For example, only three candidate features with higher confidence can be included. In the subsequent iterations, only these three candidate features are used for network training until the convergence conditions are met and the training of the lock detection network is completed.

[0161] In some embodiments, during the iterative training of the lock detection network, a ten-fold crossover method can be used, that is, all sample data are randomly divided into 10 parts, 9 of which are used as training samples and the remaining 1 is used as a verification sample, and then the average accuracy of the 10 detection results is used as the accuracy of this round of iteration.

[0162] In the disclosed embodiment, through the above-mentioned training process, candidate features can be screened and only candidate features with higher confidence levels are retained. Therefore, in the actual prediction stage, only candidate features with higher confidence levels can be extracted for lock detection, thereby improving the accuracy of lock detection.

[0163] Of course, it is understandable that for different screw specifications and application scenarios, the corresponding locking parameters will also be different. Therefore, in some embodiments of the present disclosure, a variety of locking detection networks can be pre-trained for different screw specifications, application scenarios, etc., so that during actual detection, the column only needs to select a suitable locking detection network according to the screw specifications, application scenarios, etc., and the present disclosure will not go into details.

[0164] From the above, it can be seen that in the embodiments of the present disclosure, the screw locking effect can be detected in real time during the screw locking process of the automatic screw locking machine, thereby achieving coverage of the entire screw locking process, which improves the detection efficiency and reliability and reduces the detection cost compared to the manual method, and obtains the locking detection results through the locking features of multiple locking stages, thereby improving the detection accuracy of the screw locking effect.

[0165] In some embodiments, the present disclosure provides a screw locking detection device, which can be applied to the aforementioned automatic screw locking machine. Figure 6 As shown, in some embodiments, the screw locking detection device of the present disclosure example includes:

[0166] The parameter acquisition module 100 is configured to acquire a set of screw locking parameters in a screw locking process, wherein the screw locking process refers to a process in which the end actuator of the automatic screw locking machine starts to drive the screw to rotate until the end actuator is separated from the screw;

[0167] The feature extraction module 200 is configured to extract features from the locking parameter set to obtain locking features corresponding to each locking stage included in the screw locking process; the screw locking process includes multiple consecutive locking stages;

[0168] The locking detection module 300 is configured to determine a locking detection result corresponding to the screw locking process according to the locking features of each locking stage.

[0169] From the above, it can be seen that in the embodiments of the present disclosure, the screw locking effect can be detected in real time during the screw locking process of the automatic screw locking machine, thereby achieving coverage of the entire screw locking process, which improves the detection efficiency and reliability and reduces the detection cost compared to the manual method, and obtains the locking detection results through the locking features of multiple locking stages, thereby improving the detection accuracy of the screw locking effect.

[0170] In some implementations, the parameter acquisition module 100 is configured to:

[0171] During the screw locking process, the locking parameters of the end actuator at each collection moment are collected at a preset sampling frequency to obtain a set of locking parameters during the screw locking process.

[0172] In some implementations, the feature extraction module 200 is configured to:

[0173] According to the preset execution parameters of the end-actuator, the locking parameter set of the screw locking process is divided into stages to obtain a locking parameter subset corresponding to each locking stage; the execution parameters include the rotation angle or rotation duration of the end-actuator;

[0174] For each locking stage, a locking feature corresponding to the locking stage is determined according to the locking parameters included in the locking parameter subset of the locking stage.

[0175] In some embodiments, the locking parameters include the rotation angle of the end effector during the screw locking process, the torque output by the end effector, and the rotation speed of the end effector; the feature extraction module 200 is configured as follows:

[0176] The torque characteristic corresponding to the locking stage is obtained by calculating the average value of the torque at each acquisition moment in the locking parameter subset;

[0177] The average value is calculated according to the rotation speed at each acquisition moment in the locking parameter subset to obtain the rotation speed characteristic corresponding to the locking stage;

[0178] Determine the angle value corresponding to the maximum torque in the locking stage according to the locking parameter subset;

[0179] The locking characteristics of the locking stage are obtained according to the torque characteristics, the rotation speed characteristics and the angle value corresponding to the maximum torque.

[0180] In some embodiments, the lock detection module 300 is configured to:

[0181] The locking features of each locking stage are input into a pre-trained locking detection network to obtain the locking detection results output by the locking detection network.

[0182] In some embodiments, the lock detection module 300 is configured to:

[0183] Acquire a sample data set, the sample data set includes a plurality of sample data, each sample data includes a locking parameter set and a result label in a screw locking process, and the locking parameter set includes a locking parameter subset in each locking stage;

[0184] Feature extraction is performed based on the locking parameter set of each sample data to obtain multiple categories of candidate features corresponding to each locking stage;

[0185] Inputting multiple categories of candidate features of each sample data into the lock-attachment detection network to be trained, and obtaining the detection result output by the lock-attachment detection network and the confidence of each candidate feature;

[0186] Based on the difference between the detection results and the result labels, the network parameters of the lock detection network are adjusted, and the candidate features for the next iteration are screened based on the confidence of each candidate feature until the convergence condition is met to obtain the trained lock detection network.

[0187] In some embodiments, the lock detection module 300 is configured to:

[0188] In response to the locking detection result being locking abnormality, an alarm message is generated, and / or the locking detection result is displayed on a display device of the automatic screw locking machine.

[0189] In some embodiments, the present disclosure provides an automatic screw locking machine, which can refer to the aforementioned Figure 1 As shown, it includes:

[0190] Processor; and

[0191] The memory stores computer instructions, where the computer instructions are used to enable the processor to execute the method of any of the above embodiments.

[0192] In some embodiments, the present disclosure provides a storage medium storing computer instructions, where the computer instructions are used to enable a computer to execute the method of any of the above embodiments.

[0193] Specifically, Figure 7 The structure diagram of the automatic screw locking machine 600 suitable for implementing the method disclosed in the present invention is shown. Figure 7 The automatic screw locking machine shown can realize the corresponding functions of the above-mentioned processor and storage medium.

[0194] like Figure 7 As shown, the automatic screw locking machine 600 includes a processor 601, which can perform various appropriate actions and processes according to the program stored in the memory 602 or the program loaded from the storage part 608 into the memory 602. In the memory 602, various programs and data required for the operation of the automatic screw locking machine 600 are also stored. The processor 601 and the memory 602 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0195] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage section 608 as needed.

[0196] In particular, according to an embodiment of the present disclosure, the above method process can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program tangibly contained on a machine-readable medium, and the computer program includes a program code for executing the above method. In such an embodiment, the computer program can be downloaded and installed from a network through the communication part 609, and / or installed from a removable medium 611.

[0197] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and a module, a program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0198] Obviously, the above embodiments are merely examples for the purpose of clear explanation, and are not intended to limit the embodiments. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. The obvious changes or modifications derived therefrom are still within the scope of protection of the present invention.

Claims

1. A screw locking detection method, characterized in that: Applied to an automatic screw locking machine, the method comprises: Acquire a set of screw locking parameters in a screw locking process, wherein the screw locking process includes: an end actuator of the automatic screw locking machine starts to drive the screw to rotate until the end actuator is separated from the screw; Performing feature extraction on the locking parameter set to obtain locking features corresponding to each locking stage included in the screw locking process; the screw locking process includes: at least one of the locking stages; The locking detection result corresponding to the screw locking process is determined according to the locking characteristics of each locking stage.

2. The method according to claim 1, characterized in that The step of obtaining a set of screw locking parameters during the screw locking process includes: During the screw locking process, the locking parameters of the end actuator at each acquisition moment are collected at a preset sampling frequency to obtain the locking parameter set during the screw locking process.

3. The method according to claim 1, characterized in that: The feature extraction of the locking parameter set to obtain the locking features corresponding to each locking stage included in the locking process includes: According to the preset execution parameters of the end actuator, the locking parameter set of the screw locking process is divided into stages to obtain a locking parameter subset corresponding to each locking stage; the execution parameters include the rotation angle or rotation duration of the end actuator; For each locking stage, the locking feature corresponding to the locking stage is determined according to the locking parameters included in the locking parameter subset of the locking stage.

4. The method according to claim 3, characterized in that The locking parameters include the rotation angle of the end actuator, the torque output by the end actuator, and the rotation speed of the end actuator during the screw locking process; the locking feature corresponding to the locking stage is determined according to each locking parameter included in the locking parameter subset of the locking stage, including: The torque characteristic corresponding to the locking stage is obtained by calculating the average value of the torque at each acquisition moment in the locking parameter subset; The average value is calculated according to the rotation speed at each acquisition moment in the locking parameter subset to obtain the rotation speed characteristic corresponding to the locking stage; Determine the angle value corresponding to the maximum torque in the locking stage according to the locking parameter subset; The locking feature of the locking stage is obtained according to the torque feature, the rotation speed feature and the angle value corresponding to the maximum torque.

5. The method according to any one of claims 1 to 4, characterized in that: The step of determining the locking detection result corresponding to the screw locking process according to the locking features of each locking stage includes: The locking features of each locking stage are input into a pre-trained locking detection network to obtain the locking detection result output by the locking detection network.

6. The method according to claim 5, characterized in that The process of training the lock detection network includes: Acquire a sample data set, the sample data set comprising a plurality of sample data, each sample data comprising a locking parameter set and a result label in a screw locking process, the locking parameter set comprising a locking parameter subset in each locking stage; Feature extraction is performed based on the locking parameter set of each sample data to obtain multiple categories of candidate features corresponding to each locking stage; Inputting multiple categories of candidate features of each sample data into the lock-attachment detection network to be trained, and obtaining the detection result output by the lock-attachment detection network and the confidence of each candidate feature; The network parameters of the lock detection network are adjusted based on the difference between the detection result and the result label, and the candidate features of the next iteration are screened based on the confidence of each candidate feature until the convergence condition is met, thereby obtaining a trained lock detection network.

7. The method according to claim 1, characterized in that Also includes: In response to the locking detection result being a locking abnormality, an alarm message is generated, and / or the locking detection result is displayed on a display device of the automatic screw locking machine.

8. A screw locking detection device, characterized in that: Applied to an automatic screw locking machine, the device comprises: The parameter acquisition module is configured to acquire a set of screw locking parameters during a screw locking process, wherein the screw locking process includes: a process in which an end actuator of the automatic screw locking machine starts to drive the screw to rotate until the end actuator is separated from the screw; A feature extraction module is configured to perform feature extraction on the locking parameter set to obtain locking features corresponding to each locking stage included in the screw locking process; the screw locking process includes: at least one of the locking stages; The locking detection module is configured to determine a locking detection result corresponding to the screw locking process according to the locking characteristics of each locking stage.

9. An automatic screw locking machine, characterized in that: include: processor; as well as A memory storing computer instructions, wherein the computer instructions are used to enable a processor to execute the method according to any one of claims 1 to 7.

10. A storage medium, characterized in that: Computer instructions are stored, and the computer instructions are used to make a computer execute the method according to any one of claims 1 to 7.

Images