A method for monitoring the operating status of marble-like aluminum single plate stamping equipment

Abstract

The present application relates to the field of data processing technology, and in particular to a method for monitoring the operating status of a marble-like aluminum single plate stamping device, comprising: inputting the aluminum plate thickness and the real-time control parameter sequence into a classification model to obtain an abnormal probability; if the abnormal probability is greater than a preset value, calculating the gradient value of the abnormal probability at each moment to determine the adjustment moment; clustering the historical control parameter sequence to obtain candidate parameters, replacing the control parameters at the adjustment moment with the candidate parameters with the maximum priority value to update the real-time control parameter sequence until the optimal control parameter sequence is obtained; if the abnormal probability of the optimal control parameter sequence is greater than the preset value, the operating status is abnormal, otherwise, the operating status is normal. Through the technical solution of the present application, it is possible to avoid misdetection of the operating status due to inaccurate control parameters and accurately obtain the status monitoring results.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method for monitoring the operating status of marble-like aluminum single plate stamping equipment. Background Art

[0002] Stamping equipment is a kind of equipment that uses mechanical principles to achieve cold bending and forming, and is an indispensable equipment in modern industrial production. In order to ensure that the stamping equipment can produce stamping products of qualified quality, it is necessary to analyze the operating status of the stamping equipment and perform timely inspection and maintenance of the stamping equipment.

[0003] Imitation marble aluminum veneer is an aluminum plate material with marble texture on the surface. The thickness of the aluminum plate includes 1.5mm, 2.0mm, 2.5mm, 3.0mm and other options.

[0004] Usually, when the quality of stamped products is unqualified, the operating status of the stamping equipment is deemed abnormal. However, inaccurate control parameters of the stamping equipment will also lead to unqualified quality of stamped products, resulting in false detection of the operating status of the stamping equipment and inability to obtain accurate status monitoring results. Summary of the invention

[0005] In order to solve the technical problem of inaccurate operating status monitoring results of marble-like aluminum veneer stamping equipment, the present application provides an operating status monitoring method for marble-like aluminum veneer stamping equipment, which can avoid misdetection of operating status due to inaccurate control parameters and accurately obtain the status monitoring results of the stamping equipment.

[0006] In a first aspect of the present application, a method for monitoring the operating status of a marble-like aluminum single plate stamping device is provided, the monitoring method comprising: inputting the thickness of the aluminum plate and the real-time control parameter sequence in the stamping process into a classification model to obtain an abnormal probability; in response to the abnormal probability being greater than a preset value, calculating the gradient value of the abnormal probability at each moment, and taking the moment of the maximum gradient value as the adjustment moment; performing distance clustering on the historical control parameter sequence according to the control parameters at other moments other than the adjustment moment, taking the cluster cluster to which the real-time control parameter sequence belongs as the target cluster, taking the control parameters of each historical control parameter sequence in the target cluster at the adjustment moment as candidate parameters, and calculating the priority of each candidate parameter; replacing the control parameters at the adjustment moment in the real-time control parameter sequence with the candidate parameters with the maximum priority value to update the real-time control parameter sequence; updating the real-time control parameter sequence multiple times until the abnormal probability of the updated real-time control parameter sequence is not greater than the preset value or the number of updates is greater than the preset number, and obtaining the optimal control parameter sequence; in response to the abnormal probability of the optimal control parameter sequence being greater than the preset value, the operating status is abnormal, otherwise, the operating status is normal, and the stamping device is controlled according to the optimal control parameter sequence.

[0007] The thickness of the aluminum plate and the real-time control parameter sequence in the stamping process are input into the classification model to obtain the abnormal probability, which can characterize the stamping quality; when the abnormal probability is greater than the preset value, it means that the stamping quality is not up to standard at this time, and the real-time control parameter sequence needs to be updated; the gradient value of the abnormal probability at each moment is calculated, and the moment of the maximum gradient value is taken as the adjustment moment, and the control parameters at the adjustment moment are adjusted, which can effectively improve the abnormal probability; the historical control parameter sequence is distance clustered according to the control parameters at other moments except the adjustment moment, and the candidate parameters of the control parameters are determined in the cluster cluster to which the real-time control parameter sequence belongs, and the candidate parameters are all from the historical control parameter sequence. The rationality of the candidate parameters is guaranteed; the control parameters at the adjustment time in the real-time control parameter sequence are replaced with the candidate parameters with the maximum priority value to realize the update of the real-time control parameter sequence, until the abnormal probability of the updated real-time control parameter sequence is not greater than the preset value or the number of updates is greater than the preset number, and the optimal control parameter sequence is obtained; if the abnormal probability of the optimal control parameter sequence is greater than the preset value, it means that the stamping equipment cannot obtain stamping products of qualified quality and the operating state is abnormal; otherwise, the operating state is normal, and the stamping equipment is controlled according to the optimal control parameter sequence. In this way, it can avoid misdetection of the operating state due to inaccurate control parameters and accurately obtain the state monitoring results.

[0008] Preferably, the classification model includes a feature extraction sub-model and a classification sub-model, wherein the feature extraction sub-model is a recurrent neural network, which is used to extract the timing characteristics of the real-time control parameter sequence; the timing characteristics and the thickness of the aluminum plate are spliced ​​and input into the classification sub-model to obtain the abnormality probability.

[0009] An end-to-end classification model is given, which can integrate the real-time control parameter sequence and aluminum plate thickness, output the abnormal probability of stamping products, predict the stamping quality, and provide a model basis for the subsequent operation status monitoring.

[0010] Preferably, the training method of the classification model includes: taking the aluminum plate thickness and historical control parameter sequence in the historical stamping process as training samples, and obtaining the quality labels of the training samples; inputting the training samples into the classification model to obtain the classification results, and back-propagating the classification model based on the cross entropy loss between the classification results and the quality labels to complete the training of the classification model.

[0011] Preferably, calculating the gradient value of the abnormal probability at each moment includes: performing back propagation on the abnormal probability to obtain the gradient value of each control parameter at each moment, and taking the average gradient value of each control parameter as the gradient value at the corresponding moment.

[0012] By back-propagating the abnormal probability, the gradient value at each moment in the real-time control parameter sequence can be calculated. The larger the gradient value, the more effective the control parameter at that moment can be in changing the value of the abnormal probability.

[0013] Preferably, the method of taking the cluster to which the real-time control parameter sequence belongs as the target cluster includes: for any two control parameter sequences, calculating the Euclidean distance of the control parameters at other times except the adjustment time, and defining the sum of the Euclidean distances at all other times as the cluster distance of the control parameter sequence; performing distance clustering on the historical control parameter sequence according to the cluster distance of the control parameter sequence to obtain multiple clusters and cluster centers of each cluster; calculating the cluster distance between the real-time control parameter sequence and each cluster center, and taking the cluster with the minimum cluster distance as the target cluster; and taking the control parameters of each historical control parameter sequence in the target cluster at the adjustment time as candidate parameters.

[0014] Since the candidate parameters are all selected from the historical control parameter sequence, the rationality of the candidate parameters can be guaranteed and the wear of the stamping equipment caused by unreasonable candidate parameters can be reduced; in other words, the candidate parameters include all reasonable values ​​of the control parameters at the adjustment moment in the real-time control parameter sequence.

[0015] Preferably, the control parameter sequence and control parameter sequence The clustering distance for:

[0016] ; and The control parameter sequences are and control parameter sequence Middle time The control parameters, To control the number of moments in the parameter sequence, To adjust the time, for and The Euclidean distance of .

[0017] Preferably, the priority of the candidate parameters is :

[0018] , are candidate parameters, is the candidate parameter in the target cluster The number of occurrences of is the number of historical control parameter sequences in the target cluster, For the control parameters at the adjustment moment in the real-time control parameter sequence, is the gradient value at the adjustment moment.

[0019] The larger the value is, the more likely the candidate parameters are in the historical stamping process. The frequency of occurrence of The less damage to the stamping equipment; The larger the value, the better the candidate parameter The more obvious the effect of reducing the probability of abnormality, the better the candidate parameters It can effectively improve the stamping quality; it can accurately quantify the priority of candidate parameters from two aspects: the degree of damage to the stamping equipment and the effect of improving the stamping quality. The larger the priority, the less damage the candidate parameter causes to the stamping equipment, and the more obvious the effect of the candidate parameter on improving the stamping quality.

[0020] Preferably, after each update of the real-time control parameter sequence, the monitoring method further comprises: inputting the thickness of the aluminum plate and the updated real-time control parameter sequence into the classification model again to obtain the abnormal probability of the updated real-time control parameter sequence.

[0021] The technical solution of this application has the following beneficial technical effects:

[0022] The thickness of the aluminum plate and the real-time control parameter sequence in the stamping process are input into the classification model to obtain the abnormal probability, which can characterize the stamping quality; when the abnormal probability is greater than the preset value, it means that the stamping quality is not up to standard at this time, and the real-time control parameter sequence needs to be updated; the gradient value of the abnormal probability at each moment is calculated, and the moment of the maximum gradient value is taken as the adjustment moment, and the control parameters at the adjustment moment are adjusted, which can effectively improve the abnormal probability; the historical control parameter sequence is distance clustered according to the control parameters at other moments except the adjustment moment, and the candidate parameters of the control parameters are determined in the cluster cluster to which the real-time control parameter sequence belongs, and the candidate parameters are all from the historical control parameter sequence. The rationality of the candidate parameters is guaranteed; the control parameters at the adjustment time in the real-time control parameter sequence are replaced with the candidate parameters with the maximum priority value to realize the update of the real-time control parameter sequence, until the abnormal probability of the updated real-time control parameter sequence is not greater than the preset value or the number of updates is greater than the preset number, and the optimal control parameter sequence is obtained; if the abnormal probability of the optimal control parameter sequence is greater than the preset value, it means that the stamping equipment cannot obtain stamping products of qualified quality and the operating state is abnormal; otherwise, the operating state is normal, and the stamping equipment is controlled according to the optimal control parameter sequence. In this way, it can avoid misdetection of the operating state due to inaccurate control parameters and accurately obtain the state monitoring results. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] By reading the detailed description below with reference to the accompanying drawings, the above and other purposes, features and advantages of the exemplary embodiments of the present application will become easy to understand. In the accompanying drawings, several embodiments of the present application are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0024] Figure 1It is a flow chart of a method for monitoring the operating status of marble-like aluminum single plate stamping equipment according to an embodiment of the present application. DETAILED DESCRIPTION

[0025] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0026] It should be understood that when the terms "first", "second", etc. are used in the claims, specification and drawings of the present application, they are only used to distinguish different objects, rather than to describe a specific order. The terms "include" and "comprise" used in the specification and claims of the present application indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their collections.

[0027] According to the first aspect of the present application, the present application provides a method for monitoring the operating status of marble-like aluminum single plate stamping equipment. Figure 1 1 is a flow chart of a method for monitoring the operating status of a marble-like aluminum single plate stamping device according to an embodiment of the present application. Figure 1 As shown, the operation status monitoring method of the marble-like aluminum single plate stamping equipment includes steps S101 to S105, which are described in detail below.

[0028] S101, the thickness of the aluminum plate and the real-time control parameter sequence during the stamping process are input into the classification model to obtain the abnormal probability.

[0029] In one embodiment, when a stamping device is used to stamp an aluminum plate, the control parameters of the stamping device include one or more of stamping pressure, stamping depth, impact speed, and impact temperature. A complete stamping process includes multiple moments, and the values ​​of the control parameters at each moment in the stamping process are collected to obtain a real-time control parameter sequence in the stamping process. The thickness of the aluminum plate and the real-time control parameter sequence in the stamping process are input into the classification model to obtain an abnormal probability, which is used to reflect the stamping quality; the larger the abnormal probability, the worse the stamping quality.

[0030] Among them, the classification model includes a feature extraction sub-model and a classification sub-model. The feature extraction sub-model is a recurrent neural network, which is used to extract the timing characteristics of the real-time control parameter sequence; the timing characteristics and the thickness of the aluminum plate are spliced ​​and input into the classification sub-model to obtain the abnormal probability.

[0031] The recurrent neural network is an LSTM network or a GRU network, which is not limited in this application. The classification sub-model is a binary classification network such as a logical classification or a Softmax function, which is used to judge whether the stamping quality is normal or abnormal.

[0032] It can be understood that in order for the classification model to output accurate abnormality probabilities, the classification model needs to be trained. The training method of the classification model includes: taking the aluminum plate thickness and the historical control parameter sequence in the historical stamping process as training samples, and obtaining the quality labels of the training samples, wherein the quality labels are manually labeled, and if the stamping quality of the historical stamping process is normal, the quality label is 0, and if the stamping quality of the historical stamping process is abnormal, the quality label is 1; inputting the training samples into the classification model to obtain the classification results, and using the cross entropy loss between the classification results and the quality labels to back-propagate the classification model, thereby realizing the training of the classification model.

[0033] In this way, the abnormal probability is obtained according to the trained classification model, and the abnormal probability is used to reflect the stamping quality.

[0034] S102, in response to the abnormal probability being greater than a preset value, calculating the gradient value of the abnormal probability at each moment, and taking the moment of maximum gradient value as the adjustment moment.

[0035] In one embodiment, the preset value is 0.5. When the abnormal probability is not greater than 0.5, it indicates that the stamping quality is qualified, indicating that the stamping equipment can operate normally and can obtain stamping products that meet the stamping quality standards. Therefore, the operating state of the stamping equipment is normal; conversely, when the abnormal probability is greater than 0.5, it indicates that the stamping quality is unqualified, indicating that either the operating state of the stamping equipment is abnormal and stamping products that meet the stamping quality standards cannot be obtained, or the real-time control parameter sequence in the stamping process is inaccurate. Therefore, in order to obtain an accurate operating state and ensure that stamping products that meet the stamping quality standards are obtained, it is necessary to update the real-time control parameter sequence in the stamping process. If the updated real-time control parameter sequence can obtain stamping products that meet the stamping quality standards, it means that the operating state is normal; if the updated real-time control parameter sequence still cannot obtain stamping products that meet the stamping quality standards, it means that the operating state is abnormal and the stamping equipment needs to be repaired.

[0036] Specifically, calculating the gradient value of the abnormal probability at each moment includes: performing back propagation on the abnormal probability to obtain the gradient value of each control parameter at each moment, and taking the average gradient value of each control parameter as the gradient value at the corresponding moment. The moment of the maximum gradient value is taken as the adjustment moment.

[0037] It can be understood that according to the gradient descent method, by calculating the gradient information of a function at any parameter (the gradient information includes the gradient value and the gradient direction), and updating the parameters along the direction of gradient descent (i.e., the opposite direction of the gradient direction), the value of the function can be reduced at the fastest speed; on the contrary, by updating the parameters along the direction of gradient ascent (i.e., the gradient direction), the value of the function can be increased at the fastest speed. Therefore, if the gradient value at a moment is larger, it means that the control parameter at that moment can effectively change the value of the abnormal probability, that is, updating the control parameter at that moment can quickly and effectively obtain the optimal control parameter sequence.

[0038] S103, distance clustering the historical control parameter sequence according to the control parameters at other times except the adjustment time, taking the cluster cluster to which the real-time control parameter sequence belongs as the target cluster, taking the control parameters of each historical control parameter sequence in the target cluster at the adjustment time as candidate parameters, and calculating the priority of each candidate parameter.

[0039] In one embodiment, since the control parameters change over time, the change amount of the control parameters between adjacent moments is affected by the stamping equipment; for example, the adjustment moment is recorded as ,time The control parameters are affected by the time and time The control parameters of the machine are affected by the large changes, which will cause the wear of the stamping equipment. Therefore, in order to reduce the wear of the stamping equipment, the candidate parameters at the adjustment time are determined according to the historical control parameter sequence to ensure the rationality of the candidate parameters.

[0040] Specifically, the method of taking the cluster to which the real-time control parameter sequence belongs as the target cluster includes: for any two control parameter sequences, calculating the Euclidean distance of the control parameters at other times except the adjustment time, and defining the sum of the Euclidean distances at all other times as the cluster distance of the control parameter sequence; performing distance clustering on the historical control parameter sequence according to the cluster distance of the control parameter sequence to obtain multiple clusters and cluster centers of each cluster; calculating the cluster distance between the real-time control parameter sequence and each cluster center, and taking the cluster with the minimum cluster distance as the target cluster; and taking the control parameters of each historical control parameter sequence in the target cluster at the adjustment time as candidate parameters.

[0041] Among them, the control parameter sequence and control parameter sequence The clustering distance for: ; and The control parameter sequences are and control parameter sequence Middle time The control parameters, To control the number of moments in the parameter sequence, To adjust the time, for and The distance clustering can adopt Kmeans algorithm or Kmeans++ algorithm.

[0042] Among them, a cluster includes multiple historical control parameter sequences, and the lengths of the historical control parameter sequences are the same. At any moment in the historical control parameter sequence, the average control parameter of all historical control parameter sequences at that moment is calculated to obtain the cluster center of the cluster.

[0043] It can be understood that since the candidate parameters are all selected from the historical control parameter sequence, the rationality of the candidate parameters can be guaranteed. In other words, the candidate parameters include all reasonable values ​​of the control parameters at the adjustment moment in the real-time control parameter sequence.

[0044] In one embodiment, after obtaining the candidate parameters, it is necessary to calculate the priority of each candidate parameter, and evaluate the priority of the candidate parameter from two aspects: the degree of damage to the stamping equipment by the candidate parameter and the improvement effect of the candidate parameter on the stamping quality. Specifically, the priority of the candidate parameter is :

[0045] , are candidate parameters, is the candidate parameter in the target cluster The number of occurrences of is the number of historical control parameter sequences in the target cluster, For the control parameters at the adjustment moment in the real-time control parameter sequence, is the gradient value at the adjustment moment.

[0046] Understandably, the candidate parameters in the target cluster The more times , The larger the value is, the more likely the candidate parameters are in the historical stamping process. The frequency of occurrence of The smaller the damage to the stamping equipment, the better the candidate parameters Priority The bigger.

[0047] Use the gradient descent method to update the control parameters at the adjustment time in the real-time control parameter sequence The process is as follows: , To adjust the gradient value at the moment, is the update amplitude; update amplitude The larger the value is, the more the control parameters (i.e. ) The more obvious the effect of reducing the abnormal probability, Equal to candidate parameter The update range, The larger the value, the better the candidate parameter The more obvious the effect of reducing the probability of abnormality, the better the candidate parameters Can effectively improve the stamping quality,

[0048] In this way, the priority of candidate parameters can be accurately quantified from two aspects: the degree of damage to the stamping equipment and the effect of improving the stamping quality. The larger the priority, the less damage the candidate parameter causes to the stamping equipment, and the more obvious the effect of the candidate parameter on improving the stamping quality.

[0049] S104: Replace the control parameter at the adjustment time in the real-time control parameter sequence with the candidate parameter with the maximum priority to update the real-time control parameter sequence.

[0050] In one embodiment, the control parameters at other times in the real-time control parameter sequence except the adjustment time are kept unchanged, and only the control parameters at the adjustment time are replaced with the candidate parameters with the maximum priority, so as to complete the update of the real-time control parameter sequence.

[0051] S105, updating the real-time control parameter sequence multiple times until the abnormal probability of the updated real-time control parameter sequence is no greater than a preset value or the number of updates is greater than a preset number, thereby obtaining an optimal control parameter sequence.

[0052] In one embodiment, after each update of the real-time control parameter sequence, the thickness of the aluminum plate and the updated real-time control parameter sequence are input into the classification model again to obtain the abnormal probability of the updated real-time control parameter sequence; if the abnormal probability of the updated real-time control parameter sequence is not greater than the preset value, it means that the updated real-time control parameter sequence can obtain stamping products with qualified quality. At this time, the update is stopped, and the updated real-time control parameter sequence is used as the optimal control parameter sequence; on the contrary, if the abnormal probability of the updated real-time control parameter sequence is still greater than the preset value, it means that the updated real-time control parameter sequence still cannot obtain stamping products with qualified quality, and steps S102 to S104 are repeated to iteratively update the real-time control parameter sequence until the abnormal probability of the updated real-time control parameter sequence is not greater than the preset value or the number of updates is greater than the preset number, and the last update result is used as the optimal control parameter sequence.

[0053] Among them, the preset number of times is 10.

[0054] S106, in response to the abnormal probability of the optimal control parameter sequence being greater than a preset value, the operating state is abnormal, otherwise, the stamping equipment is controlled according to the optimal control parameter sequence.

[0055] In one embodiment, if the optimal control parameter sequence is obtained because the number of updates is greater than the preset number, it means that the abnormal probability of the optimal control parameter sequence is greater than the preset value. Even if the real-time control parameter sequence is adjusted, stamping products with qualified quality cannot be obtained. Therefore, the operating state of the stamping equipment is abnormal; conversely, if the abnormal probability of the optimal control parameter sequence is not greater than the preset value, the optimal control parameter sequence is used to control the stamping equipment to ensure that the stamping equipment can obtain stamping products with qualified quality.

[0056] The above introduces the technical principles and implementation details of the operation status monitoring method of the marble-like aluminum single plate stamping equipment of the present application through specific embodiments. The thickness of the aluminum plate and the real-time control parameter sequence in the stamping process are input into the classification model to obtain the abnormal probability, which can characterize the stamping quality; when the abnormal probability is greater than the preset value, it means that the stamping quality is not up to standard at this time, and the real-time control parameter sequence needs to be updated; the gradient value of the abnormal probability at each moment is calculated, and the moment of the maximum gradient value is used as the adjustment moment, and the control parameters at the adjustment moment are adjusted, which can effectively improve the abnormal probability; the historical control parameter sequence is distance clustered according to the control parameters at other moments other than the adjustment moment, and the candidate parameters of the control parameters are determined in the cluster cluster to which the real-time control parameter sequence belongs, and the candidate parameters are all from the historical control parameter sequence. The rationality of the candidate parameters is guaranteed; the control parameters at the adjustment time in the real-time control parameter sequence are replaced with the candidate parameters with the maximum priority value to realize the update of the real-time control parameter sequence, until the abnormal probability of the updated real-time control parameter sequence is not greater than the preset value or the number of updates is greater than the preset number, and the optimal control parameter sequence is obtained; if the abnormal probability of the optimal control parameter sequence is greater than the preset value, it means that the stamping equipment cannot obtain stamping products of qualified quality and the operating state is abnormal; otherwise, the operating state is normal, and the stamping equipment is controlled according to the optimal control parameter sequence. In this way, it can avoid misdetection of the operating state due to inaccurate control parameters and accurately obtain the state monitoring results.

[0057] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0058] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the patent application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent application shall be subject to the attached claims.

Claims

1. A method for monitoring the operating status of a marble-like aluminum single plate stamping equipment, characterized in that: The monitoring method comprises: Input the thickness of the aluminum plate and the real-time control parameter sequence in the stamping process into the classification model to obtain the abnormal probability; in response to the abnormal probability being greater than a preset value, calculate the gradient value of the abnormal probability at each moment, and take the moment of the maximum gradient value as the adjustment moment; The classification model includes a feature extraction sub-model and a classification sub-model. The feature extraction sub-model is a recurrent neural network, which is used to extract the time series features of the real-time control parameter sequence; the time series features and the thickness of the aluminum plate are spliced ​​and input into the classification sub-model to obtain the abnormal probability; Perform distance clustering on the historical control parameter sequence according to the control parameters at other times except the adjustment time, take the cluster to which the real-time control parameter sequence belongs as the target cluster, take the control parameters of each historical control parameter sequence in the target cluster at the adjustment time as the candidate parameters, and calculate the priority of each candidate parameter; The priority of the candidate parameters is : , are candidate parameters, is the candidate parameter in the target cluster The number of occurrences of is the number of historical control parameter sequences in the target cluster, For the control parameters at the adjustment moment in the real-time control parameter sequence, is the gradient value at the adjustment moment; Replacing the control parameter at the adjustment time in the real-time control parameter sequence with the candidate parameter with the maximum priority value to update the real-time control parameter sequence; The real-time control parameter sequence is updated multiple times until the abnormal probability of the updated real-time control parameter sequence is no greater than a preset value or the number of updates is greater than a preset number, thereby obtaining an optimal control parameter sequence; In response to the abnormal probability of the optimal control parameter sequence being greater than a preset value, the operating state is abnormal, otherwise, the operating state is normal, and the stamping equipment is controlled according to the optimal control parameter sequence.

2. The method for monitoring the operating status of a marble-like aluminum single plate stamping equipment according to claim 1 is characterized in that: The training method of the classification model includes: The aluminum sheet thickness and historical control parameter sequences in the historical stamping process are used as training samples, and the quality labels of the training samples are obtained; The training samples are input into the classification model to obtain the classification results, and the classification model is back-propagated according to the cross entropy loss between the classification results and the quality labels to complete the training of the classification model.

3. The method for monitoring the operating status of a marble-like aluminum single plate stamping equipment according to claim 1, characterized in that: Calculating the gradient value of the abnormal probability at each time includes: The abnormal probability is back-propagated to obtain the gradient value of each control parameter at each moment, and the average gradient value of each control parameter is used as the gradient value at the corresponding moment.

4. The method for monitoring the operating status of a marble-like aluminum single plate stamping equipment according to claim 1, characterized in that: The taking the cluster to which the real-time control parameter sequence belongs as the target cluster comprises: For any two control parameter sequences, the Euclidean distance of the control parameters at other times except the adjustment time is calculated, and the sum of the Euclidean distances at all other times is defined as the clustering distance of the control parameter sequence; According to the clustering distance of the control parameter sequence, the historical control parameter sequence is clustered to obtain multiple clusters and the cluster centers of each cluster; the clustering distance between the real-time control parameter sequence and each cluster center is calculated, and the cluster cluster with the minimum clustering distance is taken as the target cluster; the control parameters of each historical control parameter sequence in the target cluster at the adjustment time are taken as candidate parameters.

5. The method for monitoring the operating status of a marble-like aluminum single plate stamping equipment according to claim 4 is characterized in that: Control parameter sequence and control parameter sequence The clustering distance for: ; and The control parameter sequences are and control parameter sequence Middle time The control parameters, To control the number of moments in the parameter sequence, To adjust the time, for and The Euclidean distance of .

6. The method for monitoring the operating status of a marble-like aluminum single plate stamping equipment according to claim 1, characterized in that: After each update of the real-time control parameter sequence, the monitoring method further comprises: The thickness of the aluminum plate and the updated real-time control parameter sequence are input into the classification model again to obtain the abnormal probability of the updated real-time control parameter sequence.

Images