Automatic dimming mask adjustment method based on application scenarios

Abstract

The present invention discloses an automatic dimming mask adjustment method based on application scenarios, and relates to the field of intelligent adjustment of welder masks. The method includes: identifying the importance of scenes of application scene sets based on the automatic dimming mask for groups, establishing similar clustering results, establishing a calibration scene dimming adjustment strategy, activating a matching acquisition mode with an initial data acquisition result, calling a corresponding calibration scene dimming adjustment strategy, and establishing an adjustment dimming adjustment strategy to perform automatic dimming mask adjustment. The method solves the technical problem that traditional masks cannot automatically dim according to changes in light intensity during welding, resulting in the need to frequently replace masks or perform manual adjustments in environments with large light changes, resulting in complex and difficult welding operations. The method realizes automatic adjustment of the mask shading level, does not require frequent manual operations by welders, responds quickly according to real-time changes in light intensity, and improves the adaptability and flexibility of welding operations.

Description

Technical Field

[0001] The present application relates to the field of intelligent adjustment of welder masks, and in particular to an automatic dimming mask adjustment method based on application scenarios. Background Art

[0002] With the development of welder masks, especially in the field of welding, its application scenario is mainly to protect the welder's eyes from the strong light and harmful radiation generated during the welding process. The mask needs to adjust the shading level of the mask according to the changes in light intensity, so as to ensure the welder's clear vision during the welding process and protect the eyes at the same time.

[0003] However, in the prior art, different welding environments will produce light of different intensities, and traditional masks are difficult to adapt to various complex welding environments. They cannot automatically adjust the light intensity according to the changes in light intensity during the welding process, resulting in the need to frequently replace the mask or perform manual adjustments in environments with large light changes, resulting in complex and difficult welding operations. Technical problems. Summary of the invention

[0004] The present application provides an automatic dimming mask adjustment method based on application scenarios to solve the problem that different welding environments will produce light of different intensities, and traditional masks are difficult to adapt to various complex welding environments and cannot automatically dim according to changes in light intensity during welding, resulting in frequent replacement of masks or manual adjustment in environments with large light changes, and the technical problem of complex and difficult welding operations. The application realizes automatic adjustment of the mask shading level without the need for welders to perform frequent manual operations, and quickly responds to real-time changes in light intensity, thereby improving the adaptability and flexibility of welding operations.

[0005] The present application provides an automatic dimming mask adjustment method based on application scenarios, comprising: obtaining an application scenario set of the automatic dimming mask, and marking the scene importance of the application scenario set based on the group oriented by the automatic dimming mask;

[0006] Performing scene similarity clustering on the application scene set to establish similarity clustering results, wherein the scene similarity clustering is constructed by scene similarity and scene importance identifier, the scene similarity is used for aggregation determination, and the scene importance identifier is used to constrain the cluster center position;

[0007] Obtaining cluster centers of similar clustering results, performing scene adaptation fitting based on the cluster centers, and establishing a calibration scene dimming adjustment strategy;

[0008] According to the similar clustering result, a photosensor is configured in the automatic dimming mask, and when the automatic dimming mode is activated, initial data collection is performed through the universal photosensor, and a matching collection mode is activated with the initial data collection result;

[0009] Control multiple photosensors to collect collaborative environment data in an activated matching collection mode, perform similarity matching of cluster centers based on the collaborative environment data collection results, and call the corresponding calibration scene dimming adjustment strategy;

[0010] Establishing deviation environment data based on the collaborative environment data collection result and the matched cluster center, inputting the calibrated scene dimming adjustment strategy and the deviation environment data into an adaptive fine-tuning network, and establishing an adjusted dimming adjustment strategy;

[0011] Automatic dimming mask adjustment is performed using the adjusted dimming adjustment strategy.

[0012] In a possible implementation, the application scenario set is clustered by scenario similarity, and a similar clustering result is established, and the following processing is performed: scenario analysis is performed on the application scenario set, and the number of pre-clusters is configured;

[0013] Randomly distributing cluster centers based on the pre-clustering quantity, taking the randomly distributed cluster centers as search centers, and performing cluster search;

[0014] When any round of cluster search is finished, the location of the cluster center is updated through the scene importance mark, and an iterative search is completed;

[0015] Continue iterative search, and when all application scenarios are clustered, establish similar clustering results.

[0016] In a possible implementation, the iterative search is continued and the following processing is performed: when the cluster centers of any distribution are merged, a new cluster center instruction is generated;

[0017] When the continuous iterative search fails to update the existing clusters, a new cluster center instruction is generated;

[0018] Adding a cluster center to the unclustered application scenario set using the newly added cluster center instruction;

[0019] Continue iterative search based on the newly added cluster centers.

[0020] In a possible implementation, the automatic dimming mask adjustment is performed by adjusting the dimming adjustment strategy, and the following processing is performed: obtaining additional input data, wherein the additional input data includes user input data and weather input data;

[0021] Using the additional input data to perform adaptive update prediction of the application scenario;

[0022] The prediction result is adaptively updated to generate a timing fine-tuning parameter, the dimming adjustment strategy is updated in timing by the timing fine-tuning parameter, and the automatic dimming mask adjustment is performed with the timing update result.

[0023] In a possible implementation, the adaptive update prediction result is used to generate a timing fine-tuning parameter, and the timing of adjusting the dimming adjustment strategy is updated by the timing fine-tuning parameter, and the following processing is performed: data steady-state analysis is performed on the additional input data to establish a timing steady-state factor;

[0024] Inputting the timing steady-state factor into a detection verification model to establish a self-check cycle;

[0025] Activate the photosensor through the self-check cycle to perform update data collection;

[0026] The timing fine-tuning parameters are compensated by updating the data collection results, and the timing of adjusting the dimming adjustment strategy is updated by using the compensated timing fine-tuning parameters.

[0027] In a possible implementation, the activated matching collection mode is used to control multiple photosensors to collect collaborative environment data, similarity matching of cluster centers is performed according to the collaborative environment data collection results, and the following processing is performed: photosensitive collection data of multiple photosensors is obtained, and the maximum cluster mean of the photosensitive collection data is obtained;

[0028] Perform similar trust comparison of the photosensitive acquisition data using the maximum cluster mean to generate a first trust result;

[0029] Performing a location-linked trust evaluation on the light-sensing collection data to generate a second trust result;

[0030] The first trust result and the second trust result are used to perform data screening of the photosensitive acquisition data, and the data screening result is used as the collaborative environment data collection result.

[0031] In a possible implementation, the following processing is performed: after performing the automatic dimming mask adjustment, obtaining feedback data from a user, wherein the user has a user account set;

[0032] Performing adaptive adjustment of the automatic dimming mask based on the feedback data, and establishing a user preference feature with the adaptive adjustment result;

[0033] The user preference characteristics are stored in the user account, and the automatic dimming mask adjustment optimization of subsequent users is performed based on the user account.

[0034] In a possible implementation, the following processing is performed: setting a warning photosensor for continuous recording;

[0035] Continuously recording photosensitivity data using the early warning photosensitivity sensor;

[0036] If the light sensitivity meets the first threshold, the alternative automatic dimming mask adjustment solution is directly called and an early warning is generated;

[0037] If the light sensitivity intensity satisfies the second threshold value and does not satisfy the first threshold value, performing light sensitivity continuous recording;

[0038] If the cumulative result of continuous recording meets the third threshold, the alternative automatic dimming mask adjustment solution is directly called and an early warning is generated.

[0039] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0040] The automatic dimming mask adjustment method based on application scenarios provided by the present application solves the problem that different welding environments will produce light of different intensities, and traditional masks are difficult to adapt to various complex welding environments and cannot automatically dim according to changes in light intensity during welding, resulting in the need to frequently replace masks or perform manual adjustments in environments with large light changes, and the technical problem of complex and difficult welding operations. The automatic adjustment of the mask shading level is realized without the need for welders to perform frequent manual operations, and a rapid response is made according to real-time changes in light intensity, thereby improving the adaptability and flexibility of welding operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solution of the embodiment of the present invention, the accompanying drawings of the embodiment of the present invention will be briefly introduced below. A flow chart is used in the present application to illustrate the operations performed by the method according to the embodiment of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps can be processed in reverse order or simultaneously as needed. At the same time, other operations can also be added to these processes, or one or more operations can be removed from these processes.

[0042] Figure 1 A schematic flow chart of an automatic dimming mask adjustment method based on an application scenario provided in an embodiment of the present application;

[0043] Figure 2 A schematic diagram of the process of establishing similar clustering results for the automatic dimming mask adjustment method based on application scenarios provided in an embodiment of the present application. DETAILED DESCRIPTION

[0044] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below.

[0045] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of this application.

[0046] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict, and the terms "first\second" involved are merely to distinguish similar objects and do not represent a specific ordering of objects. The terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products, or devices. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those generally understood by technicians in the technical field of this application. The terms used herein are for the purpose of describing the embodiments of the present application only.

[0047] The present application embodiment provides an automatic dimming mask adjustment method based on application scenarios, such as Figure 1 As shown, the method includes:

[0048] Step A100, obtaining an application scenario set of an automatic darkening mask, and marking the scene importance of the application scenario set based on the target group of the automatic darkening mask;

[0049] The application scenarios of automatic darkening masks are mainly concentrated in the fields where eyes need to be protected from strong light or harmful radiation. The application scenario set of automatic darkening masks can include welding scenarios, metal cutting and grinding scenarios, glass processing scenarios, laboratory work scenarios, etc., and the target groups of automatic darkening masks can be welders in welding scenarios. It means that the strong arc light generated in the welding process is harmful to the eyes of welders. The automatic darkening mask can adjust the shading level in real time to protect the welder's eyes, improve the welding quality and efficiency, and mark the scene importance of welding scenes as high importance; metal processing workers in metal cutting and grinding scenarios mean that sparks and strong light will be generated in the metal cutting and grinding process, and the automatic darkening mask can effectively protect Protect workers' eyes from splashes or strong light, and the scene importance of metal cutting and grinding scenes is marked as medium importance. Glass processing workers in glass processing scenes refer to strong light reflections generated during glass processing, especially during glass cutting and edging. The automatic darkening mask can reduce the damage to the eyes of glass processing workers caused by such reflections, and the scene importance of glass processing scenes is marked as medium importance. Scientific researchers in laboratory work scenes refer to strong light or harmful rays generated in certain laboratory work, such as using laser equipment or conducting specific chemical reactions. The automatic darkening mask can be used as an auxiliary protection tool for scientific researchers, and the scene importance of laboratory work scenes is marked as low importance.

[0050] The importance identification is based on the degree of damage to the eyes caused by light in each scenario, the operating frequency, and the practicality and necessity of the automatic dimming mask in the scenario. Among them, a high-importance scene indicates that the mask is extremely practical and necessary in the scenario, a medium-importance scene indicates that the mask is somewhat practical and necessary in the scenario, and a low-importance scene indicates that the practicality and necessity of the mask in the scenario are relatively low.

[0051] Execute step A200 to perform scene similarity clustering on the application scene set and establish similarity clustering results, wherein the scene similarity clustering is constructed by scene similarity and scene importance identifier, the scene similarity is used for aggregation determination, and the scene importance identifier is used to constrain the cluster center position;

[0052] In one possible implementation, Figure 2 As shown, step A200 further includes step A210, performing scenario analysis on the application scenario set and configuring the number of pre-clusters;

[0053] To perform scene similarity clustering on the application scenario set, it is first necessary to define a measurement method for scene similarity and construct a clustering algorithm in combination with a scene importance identifier, and scene similarity is used to determine scene aggregation, and the scene importance identifier is used to constrain the location of the cluster center. Furthermore, by comparing the similarities of different scenes in key features such as changes in light intensity, the degree of light damage to the eyes, and the complexity of the working environment, the similarity of each application scenario in the application scenario set is evaluated, and then the importance of each application scenario is analyzed based on factors such as business objectives, actual needs, and potential impacts. More attention should be paid to scenes with higher importance during the clustering process, and the number of pre-clusters is set according to the scene analysis results. The number of pre-clusters refers to the number that can cover most application scenarios while avoiding clustering that is too detailed or coarse.

[0054] Executing step A220, randomly distributing cluster centers based on the pre-clustering quantity, and performing cluster search using the randomly distributed cluster centers as search centers;

[0055] After configuring the number of pre-clusters, you can use randomly distributed cluster centers as the search starting point to execute the cluster search algorithm. First, based on the number of pre-clusters, assuming that the number of pre-clusters is K, randomly select K sample points in the data set as the initial cluster centers, traverse each sample point in the data set, calculate their Euclidean distances to all cluster centers, assign each sample point to the cluster corresponding to the cluster center closest to it, update the center of each cluster, and the new cluster center is the average of all sample points in the cluster. Stop the current cluster search when the cluster center no longer changes significantly.

[0056] Execute step A230, when any round of cluster search is completed, the location of the cluster center is updated through the scene importance identifier, and an iterative search is completed;

[0057] When any round of cluster search is completed, the location of the cluster center is updated based on the scene importance identifier, which means that the weighted weight is calculated based on the scene importance identifier, that is, for each sample point, a weight is assigned to it according to the importance identifier of the application scene to which it belongs. Sample points in scenes with higher importance will receive greater weights, and weighted sample points are used to calculate new cluster centers. Specifically, the contribution of each sample point to the cluster center to which it belongs will be adjusted according to its importance weight, thereby completing an iterative search for clustering.

[0058] Execute step A240, continue iterative search, and when all application scenario sets are clustered, establish similar clustering results. In a possible implementation, step A240 further includes step A241, when arbitrarily distributed cluster centers are merged, generate a new cluster center instruction;

[0059] When the merging of cluster centers of arbitrary distribution occurs during the iterative search process, it is considered that two or more clusters become very close or overlap during the execution of the algorithm, resulting in their center points being too close. In order to avoid excessive merging or optimize the cluster structure, it is necessary to generate a new cluster center instruction, which means that after each iterative search, the distance matrix between all cluster centers is calculated, and a threshold is set according to the specific application scenario and data set to determine whether the cluster centers are too close. The distance matrix is ​​further traversed to find cluster center pairs with a distance less than the threshold. If there are cluster center pairs with a distance less than the threshold, the cluster center pairs with a distance less than the threshold are considered to be merged. When the cluster center merger is detected, a new cluster center instruction needs to be generated. A representative sample point (such as the point with the highest density) in the merged cluster can be selected as the new cluster center through the new cluster center instruction.

[0060] Execute step A242, when the continuous iterative search fails to update the existing clusters, generate a new cluster center instruction;

[0061] When the existing clusters are not updated during the continuous iterative search process, that is, the cluster centers remain stable, and no new sample points are reallocated to different clusters, the current cluster structure is considered to be relatively stable, but has not yet reached the optimal state, or there are uncovered patterns in the data set. At the same time, generating new cluster center instructions can discover potential new clusters, which means that after each iterative search, the update of the cluster clusters is checked by comparing the sample point allocation or cluster center position of the cluster clusters in the current iteration and the previous iteration. If the cluster clusters are not significantly updated in multiple consecutive iterations, for example, the change in the cluster center position is less than a certain threshold, or no new sample points are reallocated, it is considered that the continuous iterative search has not updated the existing clusters, and the current clusters are relatively stable. The new cluster center instruction is generated by the same method as above to select the new cluster center.

[0062] Execute step A243, add a new cluster center in the unclustered application scenario set with the new cluster center instruction; execute step A244, continue iterative search based on the newly added cluster center.

[0063] First, the unclustered application scenario set is identified, which refers to determining the application scenarios that are not covered by the clustering algorithm or identified as independent clusters. The unclustered application scenarios can be newly added data points, previously misclassified data points, or patterns in the data set that have not yet been mined. Furthermore, new cluster centers are selected or generated in the unclustered application scenario set according to the data density method, and the generated newly added cluster centers are added to the list of cluster centers and marked as new, unstable clusters. The updated cluster center list (including original and newly added cluster centers) is further used to continue the iterative search. In each iteration, the distances from all sample points to the cluster center are recalculated, and the sample points are redistributed to the corresponding clusters according to the minimum distance principle. The center of each cluster is updated, usually by calculating the average value of all sample points in the cluster. If new sample points are redistributed to different clusters during the iteration, or the position of the cluster center changes significantly, the iterative search continues until a certain stopping condition is met. Finally, after each iteration, check whether there are any unclustered application scenarios by checking whether there are any sample points that have not been assigned to any cluster. If all application scenarios have been clustered and the clustering results remain stable, that is, no new sample points are redistributed and the position of the cluster center no longer changes significantly, the clustering is considered to be completed, and the similarity of the clustering results between different application scenario sets is evaluated by calculating the similarity index between the clustering results, and set appropriate termination conditions, such as reaching the preset maximum number of iterations, the quality of the clustering results meets a certain threshold, or no new cluster center is generated. When the termination condition is met, the iterative search stops and the final clustering result is output. The similar clustering result can include information such as the center position of each cluster, the list of sample points included, the size and shape of the cluster, etc.

[0064] Execute step A300 to obtain cluster centers of similar clustering results, perform scene adaptation fitting based on the cluster centers, and establish a calibration scene dimming adjustment strategy;

[0065] The cluster centers of similar clustering results are extracted. The cluster centers are used to characterize the center points or representative points of the data in each cluster. Before the scene is adapted and fitted based on the cluster centers, the similarity between the cluster centers can be calculated based on the distance between the cluster centers, and the similarity between all cluster centers is calculated to form a similarity matrix. Clustering is performed based on the similarity matrix, and similar cluster centers are clustered together to find the most relevant cluster set. Adapting these cluster centers to different scenes means that the data represented by the cluster centers needs to be parsed to determine which specific scenes they correspond to, that is, the parsing of cluster center features, the application of business rules, the fusion of expert knowledge, etc. Based on the results of scene adaptation, a calibration scene dimming adjustment strategy is constructed, and a dimming adjustment strategy is formulated for each scene. The dimming adjustment strategy can include automatic adjustment of light intensity, color temperature and other parameters at specific times or conditions.

[0066] Execute step A400, configure a photosensor in the automatic dimming mask according to the similarity clustering result, perform initial data collection through the universal photosensor when the automatic dimming mode is activated, and activate the matching collection mode with the initial data collection result;

[0067] Based on the above cluster analysis, the specific environment and light intensity scene represented by each cluster center is determined according to similar clustering results, and each cluster center is associated with a specific environment or light intensity scene. On this basis, multiple photosensors are installed on the automatic dimming mask. The installed multiple photosensors can be used to measure different lighting parameters, such as brightness, color temperature, lighting direction, etc. At the same time, according to the design and use environment of the mask, the position and number of sensors are determined to ensure that all key lighting changes can be covered, and when the automatic dimming mode is activated, through the universal photosensor, the universal photosensor refers to the universal photosensor on the mask. One or more main sensors are used to collect initial data. The collected initial data may include lighting parameters of the current environment, such as brightness, color temperature, etc. The results of the initial data collection are matched with the clustering results to find the most similar cluster center, and based on the correlation between each cluster center and the specific ambient light intensity scene, the specific environment or lighting scene associated with the cluster center is determined, and the collection mode matching the scene is activated, that is, the matching collection mode. The matching collection mode may include adjusting the sensitivity, sampling frequency or data processing method of other photosensors to more accurately respond to the lighting changes in the current environment.

[0068] Next, step A500 is executed to control multiple photosensors to collect collaborative environment data in an activated matching collection mode, perform similarity matching of cluster centers according to the collaborative environment data collection results, and call the corresponding calibration scene dimming adjustment strategy; in a possible implementation, step A500 further includes step A510, obtaining photosensitive collection data of multiple photosensors, and obtaining the maximum cluster mean of the photosensitive collection data; and step A520 is executed to perform similarity trust comparison of the photosensitive collection data with the maximum cluster mean to generate a first trust result;

[0069] Controlling multiple photosensors for collaborative environmental data collection in an activated matching collection mode refers to controlling multiple photosensors that have been activated based on the matching collection mode, and synchronously acquiring photosensitivity collection data from multiple photosensors. The photosensitivity collection data may include parameters such as brightness, color temperature, and lighting direction. Furthermore, the mean of all data in each cluster is calculated to represent the center value of the cluster, and all cluster means are compared to find the largest cluster mean as the maximum cluster mean of the photosensitivity collection data. The maximum cluster mean is then used to perform similarity comparison of the photosensitivity collection data, which refers to using the Euclidean distance to calculate the similarity between each new photosensitivity collection data point and the maximum cluster mean. And based on the similarity threshold, determine whether the data point belongs to a set similar to the maximum cluster mean. If the similarity exceeds the threshold, the data point is considered to be credible, and a first trust result is generated. The first trust result includes data such as whether the data point is credible and similarity.

[0070] Execute step A530 to perform a location linkage trust evaluation on the light-sensing acquisition data to generate a second trust result;

[0071] First, we can ensure that multiple captured or associated location information is available through GPS and indoor positioning technologies (such as Wi-Fi positioning, Bluetooth beacons, etc.), and combine the location information with the photosensitive data to form a data set containing location information. We can further establish a correlation model between location and trust based on historical data or expert knowledge, and then calculate each new photosensitive data point in the location trust model based on its location information, evaluate the data credibility at that location, and generate a location trust score. This score can be a continuous value (such as between 0 and 1) or a discrete level (such as high, medium, and low), thereby generating a second trust result for output.

[0072] Execute step A540 to perform data screening of the photosensitive acquisition data using the first trust result and the second trust result, and use the data screening result as the collaborative environment data collection result.

[0073] Integrate the photosensitive acquisition data, including the original data, the first trust result obtained by the similarity trust comparison based on the cluster mean, and the second trust result obtained by the location linkage trust evaluation, to filter the photosensitive acquisition data, and only retain the data points that meet the filtering conditions, and use the data points obtained by the screening as the collaborative environment data collection results, and perform similarity matching of the cluster centers based on the collaborative environment data collection results, and call the corresponding calibration scene dimming adjustment strategy, which means using the collaborative environment data collection results to perform cluster analysis, that is, calculate the center of each cluster, each cluster center is used to characterize different ambient light intensity scenes, and then each cluster center is similarly matched with the pre-defined calibration scene by calculating the similarity between the cluster center and the calibration scene parameters, and at the same time, the calibration scene with the highest similarity is selected as the scene matching the cluster center, and then according to the matched calibration scene, find the dimming adjustment strategy associated with it, activate and call the corresponding dimming adjustment strategy, and the calibration scene dimming adjustment strategy of the collaborative environment data collection result can include adjusting the brightness, color temperature, lighting direction and other parameters of the lighting system to adapt to the current ambient light intensity.

[0074] Execute step A600, establish deviation environment data based on the collaborative environment data collection result and the matched cluster center, input the calibrated scene dimming adjustment strategy and the deviation environment data into an adaptive fine-tuning network, and establish an adjusted dimming adjustment strategy;

[0075] By comparing the key parameters (such as brightness, color temperature, etc.) of the collaborative environment data collection results and the matching clustering centers, the difference or deviation value between each collaborative environment data collection result and the corresponding clustering center is calculated, and the calculated deviation value is used as the deviation environment data. Furthermore, based on the multi-layer perceptron, the nonlinear characteristics are introduced to construct an adaptive fine-tuning network, and the loss function is defined to measure the difference between the adjustment dimming adjustment strategy output by the network and the actual demand. The adaptive fine-tuning network is trained using simulated data as training data. The simulated data can include the calibration scene dimming adjustment strategy, the deviation environment data and the corresponding adjustment dimming adjustment strategy label. The network parameters are then updated through the back propagation algorithm to minimize the loss function. Finally, the trained adaptive fine-tuning network is verified using the validation set to ensure the effectiveness and accuracy of the adaptive fine-tuning network in actual applications. The trained adaptive fine-tuning network is deployed in the actual application environment. According to the real-time collaborative environment data collection results and the matching clustering centers, the deviation environment data is calculated and input into the adaptive fine-tuning network to obtain the adjustment dimming adjustment strategy.

[0076] Execute step A700 to perform automatic dimming mask adjustment using the dimming adjustment strategy.

[0077] When the automatic darkening mask is in working state, the system can receive the adjustment darkening adjustment strategy output from the adaptive fine-tuning network, and convert the adjustment darkening adjustment strategy into an adjustment instruction, and operate the corresponding components of the automatic darkening welding mask for adjustment according to the parsed adjustment instruction. In a possible implementation, step A700 further includes step A710, obtaining additional input data, wherein the additional input data includes user input data and weather input data; executing step A720, using the additional input data to perform adaptive update prediction of the application scenario;

[0078] User input data can be obtained through data input through an interface, sensor, or other means, which may include user preferences (such as brightness preference, color temperature preference), the type of light intensity used, and then real-time or predicted weather data is obtained from a weather API or other reliable sources, which may include temperature, humidity, light intensity, weather conditions (such as sunny, cloudy, rainy), etc. Further, key features are extracted from the user input data, such as the type of light intensity used by the user, preference settings, etc., and then features related to light are extracted from the weather input data, such as light intensity, weather conditions, etc., and then all features are combined into a feature vector.

[0079] The prediction model is trained by deep learning based on historical lighting data. The historical lighting data is used to represent the optimal lighting settings or other related parameters under different application scenarios. When new additional input data is available, it is input into the trained prediction model, and predictions are made based on the characteristics of the input data to generate one or more prediction results. The lighting system or other related equipment is automatically adjusted based on the prediction results to adapt to the current application scenario.

[0080] Execute step A730 to generate timing fine-tuning parameters with adaptive update prediction results, update the timing of adjusting the dimming adjustment strategy with the timing fine-tuning parameters, and perform automatic dimming mask adjustment with the timing update results.

[0081] In a possible implementation, step A730 further includes step A731, performing data steady-state analysis on the additional input data to establish a timing steady-state factor; executing step A732, inputting the timing steady-state factor into a detection verification model to establish a self-check cycle;

[0082] First, a data steady-state analysis is performed on the additional input data. The steady-state of the data can be evaluated by calculating indicators such as the mean, variance, and autocorrelation coefficient of the data to identify periodic, seasonal, or trend changes in the data, thereby analyzing the changing trend and stability of the additional input data. Then, based on the results of the data steady-state analysis, the key time series features are extracted. The time series features may include periodic factors (such as daily cycles, weekly cycles, annual cycles, etc.), seasonal factors, trend factors, etc. On this basis, they are combined to obtain the time series steady-state factors, which can be used to describe the stability and periodicity of data over time.

[0083] Furthermore, the detection and verification model is trained using historical data, which characterizes the steady-state state of the data (such as steady state, non-steady state). At the same time, the established time series steady-state factor is used as input data and input into the detection and verification model. The model evaluates and predicts the steady-state of the data according to the characteristics of the time series steady-state factor. The self-test cycle is determined according to the output results of the detection and verification model, the steady-state of the data, periodic changes and application scenarios. For example, if the data changes are relatively stable and the periodicity is strong, a longer self-test cycle can be set; conversely, if the data changes are more frequent or unstable, a shorter self-test cycle needs to be set.

[0084] Execute step A733, activate the photosensor through the self-test cycle, and perform update data collection; execute step A734, compensate the timing fine-tuning parameters with the updated data collection results, and adjust the dimming adjustment strategy timing update with the compensated timing fine-tuning parameters.

[0085] Determine whether the current time period has reached the self-test cycle. When the self-test cycle is reached, the photosensor can be activated by sending a control signal to the sensor or the corresponding hardware interface. The activated photosensor captures the light intensity of the current environment, and uses the photosensor or other data acquisition tools (such as a web crawler) to perform updated data acquisition. The collected data are multiple parameters related to light intensity.

[0086] Furthermore, based on the collected data and the previously determined timing fine-tuning parameters, the required compensation amount is calculated. The timing fine-tuning parameters may include the delay time and adjustment speed of the lighting adjustment. The timing fine-tuning parameters are optimized and compensated according to the calculated compensation amount. The compensated timing fine-tuning parameters are used to update the timing of the dimming adjustment strategy, which may be operations such as adjusting the trigger time and duration of the lighting adjustment.

[0087] Executing step A800 further includes step A810, after performing the automatic dimming mask adjustment, obtaining feedback data of a user, wherein the user has a user account set; executing step A820, performing adaptive adjustment of the automatic dimming mask based on the feedback data, and establishing a user preference feature with the adaptive adjustment result;

[0088] After the automatic dimming mask is adjusted, the system can provide feedback to the user. The feedback content may include the user's satisfaction with the current lighting conditions (such as very satisfied, satisfied, average, dissatisfied, very dissatisfied), evaluation of the adjustment speed, feeling of the adjustment accuracy, etc., wherein the user sets up a user account, and based on the association between the user account and the feedback data, specific feedback is associated with a specific user. The user account contains the user's basic information and any previously collected user preferences or historical data.

[0089] Furthermore, by calculating the average, median or other statistics of user satisfaction to obtain the user's overall evaluation of the current adjustment result, and analyzing the user's feedback on different adjustment parameters (such as adjustment speed and accuracy), the automatic dimming mask is adaptively adjusted based on the analysis results of the feedback data. For example, if the user is generally dissatisfied with the adjustment speed, the system can adjust the algorithm to speed up or slow down the adjustment speed. If the user has specific requirements for the adjustment accuracy, the system can adjust the sensor sensitivity or algorithm parameters. Finally, user preference features are established based on user feedback data and historical data. The preference features may include the user's favorite light intensity range, adjustment speed, color temperature preference, etc.

[0090] Execute step A830 to store the user preference characteristics in the user account, and use the user account to perform subsequent automatic dimming mask adjustment optimization for the user.

[0091] After analyzing the user feedback data and establishing the user preference characteristics, they are stored in a database associated with the user account. The database contains various types of preference characteristics, such as light intensity range, adjustment speed, color temperature preference, etc. Furthermore, the next time the user uses the automatic dimming mask, the user can authenticate the account through a username and password or biometric technology. When the authentication is successful, the system can access the preference characteristics associated with the user account. Once the user successfully authenticates and logs in to his or her account, the system should automatically read and load the preference characteristics associated with the user account. When adjusting the automatic dimming mask, the system should use the loaded user preference characteristics as a reference. Exemplarily, the system can automatically adjust the brightness of the mask according to the user's preferred light intensity range, or adjust the speed at which the mask responds to changes in external light according to the user's preferred adjustment speed.

[0092] Executing step A800 further includes step A840, continuously recording light-sensitive data with the warning light-sensitive sensor; executing step A850, directly calling the alternative automatic dimming mask adjustment scheme and generating a warning if the light-sensitive intensity meets the first threshold;

[0093] When the photosensor is set to the continuous working mode, the ambient light intensity data is continuously captured and recorded by the photosensor, and the captured data is stored in a local or remote database in the form of a time series. According to the actual application scenario and user needs, multiple photosensitivity thresholds are set, and the multiple photosensitivity thresholds include a first threshold, a second threshold, and a third threshold. Then, it is determined whether the photosensitivity in the collected photosensitivity data record meets the first threshold. The first threshold is a critical value of the light intensity that is too high or too low. When the photosensitivity meets the first threshold, the alternative automatic dimming mask adjustment scheme is immediately triggered. The alternative automatic dimming mask adjustment scheme may include quickly adjusting the transmittance of the mask, switching to a specific preset mode, sending a notification to remind the user, etc. When the alternative scheme is triggered, an early warning signal is generated. The early warning signal can display a warning message through the device's display screen, send an email or push a notification to the user's mobile device, and the warning content in the early warning signal should clearly point out the current problem (such as excessive or weak light) and the measures that the system has taken.

[0094] Execute step A860. If the photosensitivity meets the second threshold but does not meet the first threshold, perform continuous photosensitivity recording. Execute step A870. If the cumulative result of continuous recording meets the third threshold, directly call the alternative automatic dimming mask adjustment plan and generate an early warning.

[0095] First, a second threshold is set according to the application scenario and requirements, wherein the first threshold set above is used to trigger emergency operations or alternative plans, which may be extreme lighting conditions, and the second threshold is used to trigger continuous light sensitivity recording, which may be a non-extreme but more important lighting change that requires attention. When the light sensitivity intensity meets the second threshold but does not reach the first threshold, continuous light sensitivity recording is started, which means that the light intensity data is stored in a local or remote database in the form of a time series according to the recording frequency for real-time adjustment to capture sufficient data to analyze the changing trend of lighting conditions. When the light sensitivity intensity data no longer meets the second threshold, the system can stop continuous light sensitivity recording.

[0096] Furthermore, if the cumulative result of continuously recorded photosensitivity meets the third threshold, it means that the lighting condition is continuously in a certain specific state for a period of time. The third threshold is obtained based on the cumulative result of continuously recorded photosensitivity, that is, the average photosensitivity of the time period, the total photosensitivity increment or the duration of continuous exceeding a certain range, etc. The cumulative result of continuous recording is checked regularly or in real time to see whether it meets the third threshold condition by calculating the average photosensitivity in the time period, the accumulated total photosensitivity change, or counting the duration of continuous exceeding a certain range. If the cumulative result of continuous recording meets the third threshold condition, the system directly calls the alternative automatic dimming mask adjustment scheme, which is the same as the alternative automatic dimming mask adjustment scheme called above. The alternative scheme may include quickly adjusting the transmittance of the mask, switching to a specific lighting mode, or triggering linkage operations of other related devices, etc., and generating a warning signal while calling the alternative scheme. The warning signal can display a warning message through the device's display screen, send an email or push notification to the user's mobile device, and the warning content in the warning signal should clearly point out the current problem (such as excessive or weak light) and the measures the system has taken.

[0097] The embodiments of the present application solve the problem that different welding environments will produce light of different intensities, while traditional masks are difficult to adapt to various complex welding environments and cannot automatically adjust the light according to changes in light intensity during welding, resulting in the need to frequently replace masks or perform manual adjustments in environments with large light changes, and the technical problem of complex and difficult welding operations. The embodiments of the present application solve the problem that different welding environments will produce light of different intensities, while traditional masks are difficult to adapt to various complex welding environments and cannot automatically adjust the light according to changes in light intensity, resulting in the need to frequently replace masks or perform manual adjustments in environments with large light changes, and the technical problem of complex and difficult welding operations. The automatic adjustment of the mask's shading level is achieved, without the need for welders to perform frequent manual operations, and a rapid response based on real-time changes in light intensity is achieved, thereby improving the adaptability and flexibility of welding operations.

[0098] The above specific implementation manner does not constitute a limitation to the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application. In some cases, the actions or steps recorded in the present application can be performed in an order different from that in the embodiment and can still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. The automatic dimming mask adjustment method based on application scenarios is characterized in that: The method comprises: Acquire an application scenario set of the automatic darkening mask, and identify the scene importance of the application scenario set based on the target group of the automatic darkening mask; Performing scene similarity clustering on the application scene set to establish similarity clustering results, wherein the scene similarity clustering is constructed by scene similarity and scene importance identifier, the scene similarity is used for aggregation determination, and the scene importance identifier is used to constrain the cluster center position; Obtaining cluster centers of similar clustering results, performing scene adaptation fitting based on the cluster centers, and establishing a calibration scene dimming adjustment strategy; According to the similar clustering result, a photosensor is configured in the automatic dimming mask, and when the automatic dimming mode is activated, initial data collection is performed through the universal photosensor, and a matching collection mode is activated with the initial data collection result; Control multiple photosensors to collect collaborative environment data in an activated matching collection mode, perform similarity matching of cluster centers based on the collaborative environment data collection results, and call the corresponding calibration scene dimming adjustment strategy; Establishing deviation environment data based on the collaborative environment data collection result and the matched cluster center, inputting the calibrated scene dimming adjustment strategy and the deviation environment data into an adaptive fine-tuning network, and establishing an adjusted dimming adjustment strategy; Performing automatic dimming mask adjustment with the adjusted dimming adjustment strategy; Performing scenario analysis on the application scenario set and configuring the number of pre-clusters; Randomly distributing cluster centers based on the pre-clustering quantity, taking the randomly distributed cluster centers as search centers, and performing cluster search; When any round of cluster search is finished, the location of the cluster center is updated through the scene importance mark, and an iterative search is completed; Continue iterative search, and when all application scenarios are clustered, establish similar clustering results; When arbitrarily distributed cluster centers are merged, a new cluster center instruction is generated; When the continuous iterative search fails to update the existing clusters, a new cluster center instruction is generated; Adding a cluster center to the unclustered application scenario set using the newly added cluster center instruction; Continue iterative search based on the newly added cluster centers; Acquire additional input data, wherein the additional input data includes user input data and weather input data; Using the additional input data to perform adaptive update prediction of the application scenario; Generate timing fine-tuning parameters with adaptive update prediction results, update the timing of adjusting the dimming adjustment strategy with the timing fine-tuning parameters, and perform automatic dimming mask adjustment with the timing update results; Performing data steady-state analysis on the additional input data to establish a time series steady-state factor; Inputting the timing steady-state factor into a detection verification model to establish a self-check cycle; Activate the photosensor through the self-check cycle to perform update data collection; The timing fine-tuning parameters are compensated by updating the data acquisition results, and the timing of adjusting the dimming adjustment strategy is updated by using the compensated timing fine-tuning parameters; Acquire light-sensing data collected by multiple light-sensing sensors, and obtain the maximum cluster mean of the light-sensing data; Perform similar trust comparison of the photosensitive acquisition data using the maximum cluster mean to generate a first trust result; Performing a location-linked trust evaluation on the light-sensing collection data to generate a second trust result; The first trust result and the second trust result are used to perform data screening of the photosensitive acquisition data, and the data screening result is used as the collaborative environment data collection result.

2. The method according to claim 1, characterized in that The method further comprises: After performing the automatic dimming mask adjustment, obtaining feedback data from a user, wherein the user has a user account set; Performing adaptive adjustment of the automatic dimming mask based on the feedback data, and establishing a user preference feature with the adaptive adjustment result; The user preference characteristics are stored in the user account, and the automatic dimming mask adjustment optimization of subsequent users is performed based on the user account.

3. The method according to claim 1, characterized in that The method further comprises: Set up a warning light sensor for continuous recording; Continuously recording photosensitivity data using the early warning photosensitivity sensor; If the light sensitivity meets the first threshold, the alternative automatic dimming mask adjustment solution is directly called and an early warning is generated; If the light sensitivity intensity satisfies the second threshold value and does not satisfy the first threshold value, performing light sensitivity continuous recording; If the cumulative result of continuous recording meets the third threshold, the alternative automatic dimming mask adjustment solution is directly called and an early warning is generated.

Images