LED lamp detection method, device, equipment and medium

Abstract

The invention discloses an LED lamp detection method, device and equipment and a medium, and relates to the field of LED lamps, the method comprises the following steps: obtaining model information of a to-be-detected lamp, and calling a corresponding preset model template for the to-be-detected lamp from a preset database according to the model information; the preset model template comprises a preset deployment detection model, a preset front-end processing model deployed at the front end of the preset deployment detection model, a preset rear-end processing model deployed at the rear end of the preset deployment detection model and a processing logic relationship among the models; deploying each model according to a preset model template, and determining input information of the to-be-detected lamp according to the preset model template; and processing the input information by using a preset model template to obtain a lamp detection result of the to-be-detected lamp. The detection efficiency can be improved while the detection precision is improved, the model needing to be deployed can be flexibly replaced according to the preset model template, and the detection requirements of lamps of different models are met.

Description

Technical Field

[0001] This invention relates to the field of LED lighting fixtures, and more specifically to a method, apparatus, equipment, and medium for testing LED lighting fixtures. Background Technology

[0002] LED lighting fixtures are highly efficient and energy-saving lighting devices widely used in homes, offices, commercial spaces, and industrial settings. During the manufacturing process of LED lighting fixtures, circuit components need to be identified and tested to ensure the performance and quality of the fixtures. The lighting module and driver board are crucial components of LED lighting fixtures; the lighting module generates the light source, while the driver board provides power and control signals.

[0003] Specifically, the manufacturing process of LED lamps requires identifying the hardware status of the lamp modules and driver boards to ensure they are functioning correctly. Simultaneously, it is necessary to test the input current and load pulse width modulation (PWM) characteristics of the LED lamps to ensure current consistency, response speed, and stability under specific voltages, thereby facilitating product traceability.

[0004] The current testing method for LED lamps first identifies the model and hardware status of the lamp module and driver board, and then uses various types of test circuits to detect the corresponding operating characteristics, such as the input current and load PWM, as mentioned above. The final LED lamp test result is determined by combining the identification results and the test results. However, this testing method is inefficient and cannot meet the needs of large-scale production and testing.

[0005] Therefore, how to provide a more efficient testing method for LED lighting fixtures that can meet actual production needs is an important issue that the industry urgently needs to address. Summary of the Invention

[0006] In view of this, embodiments of the present invention provide a method, apparatus, equipment and medium for testing LED lamps, thereby solving the problem that LED lamp testing schemes are inefficient and cannot meet the needs of large-scale production and testing.

[0007] According to a first aspect, embodiments of the present invention provide a method for detecting LED lamps, the method comprising: The model information of the lamp to be tested is obtained, and the corresponding preset model template is retrieved from the preset database according to the model information. The preset model template includes a preset deployment detection model, a preset front-end processing model deployed in front of the preset deployment detection model, a preset back-end processing model deployed in back of the preset deployment detection model, and the processing logic relationship between each model. In addition, the preset model template contains at least one preset deployment detection model, each preset deployment detection model has a preset front-end processing model and a preset back-end processing model, and each preset front-end processing model has corresponding preset model requirements. Deploy each model according to the preset model template, and determine the input information of the lamp to be tested according to the preset model template; The system acquires input information, processes it using a preset model template, and obtains the lamp detection results for the lamps to be tested.

[0008] In conjunction with the first aspect, in the first embodiment of the first aspect, the step of deploying each model according to a preset model template and determining the input information of the lamp to be tested according to the preset model template specifically includes: The retrieved preset model template is parsed to obtain the execution order and data transfer relationship between the various models in the preset model template; Deploy each model according to the execution order and data transfer relationships; Identify all external input call interfaces for all models, and based on all input call interfaces and the preset model requirements of the preset front-end processing models that call the input call interfaces, determine the input information of the lamp to be tested.

[0009] In conjunction with the first implementation method of the first aspect, in the second implementation method of the first aspect, the deployment of each model according to the execution order and data transmission relationship specifically includes: The original code of the model is determined, and the original code is converted according to the hardware information of the hardware platform to obtain the converted code; Generate models on hardware platforms using conversion code; Debug the model using historical data.

[0010] In conjunction with the first aspect, in the third embodiment of the first aspect, the step of obtaining the model information of the lamp to be tested and retrieving the corresponding preset model template from the preset database according to the model information specifically includes: Acquire image data containing model information of the lamp to be tested; Text extraction and recognition are performed on image data to obtain model information; Based on the model information, the corresponding preset model template is retrieved from the preset database for the lamp to be tested.

[0011] In conjunction with the third implementation of the first aspect, in the fourth implementation of the first aspect, a trained semantic information recognition model is used to extract and recognize text from image data, wherein the semantic recognition model is trained on the feature extraction layer using a self-supervised training method.

[0012] In conjunction with the first aspect, in the fifth embodiment of the first aspect, the method further includes: Based on the lighting fixture test results, the source of the fault is traced to locate the source of the lighting fixture fault and the cause of the fault.

[0013] In conjunction with the fifth embodiment of the first aspect, in the sixth embodiment of the first aspect, the step of tracing the source of the fault based on the lamp test results, locating the source of the lamp fault and the cause of the lamp fault specifically includes: Extract the causes of lamp malfunctions from the lamp test results; A pre-defined backend processing model that determines the abnormality in output data based on the cause of the lighting fixture malfunction; Determine the preset deployment detection model corresponding to the preset backend processing model, and determine the source of the lamp failure based on the preset deployment detection model.

[0014] According to a second aspect, embodiments of the present invention also provide a testing device for LED lamps, the device comprising: The template calling model is used to obtain the model information of the lamp to be tested, and retrieve the corresponding preset model template from the preset database according to the model information. The preset model template includes a preset deployment detection model, a preset front-end processing model deployed in front of the preset deployment detection model, a preset back-end processing model deployed in back of the preset deployment detection model, and the processing logic relationship between each model. In addition, the preset model template contains at least one preset deployment detection model, each preset deployment detection model has a preset front-end processing model and a preset back-end processing model, and each preset front-end processing model has corresponding preset model requirements. The model deployment module is used to deploy each model according to the preset model template and determine the input information of the lamp to be tested according to the preset model template. The lighting fixture detection module is used to acquire input information, process the input information using a preset model template, and obtain the lighting fixture detection results of the lighting fixture to be detected.

[0015] According to a third aspect, embodiments of the present invention also provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the detection method for LED lamps as described above.

[0016] According to a fourth aspect, embodiments of the present invention also provide a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the detection method for LED lamps as described above.

[0017] The present invention discloses a method, apparatus, equipment, and medium for testing LED lamps. The method involves acquiring the model information of the lamp to be tested, retrieving a corresponding preset model template from a preset database based on the model information, and using the preset model template as the basis for lamp testing. The preset model template includes a preset deployment testing model, a preset front-end processing model deployed at the front end of the preset deployment testing model, a preset back-end processing model deployed at the back end of the preset deployment testing model, and the processing logic relationships between the various models. The method then deploys each model according to the preset model template, determines the input information of the lamp to be tested based on the preset model template, processes the input information using the preset model template, and obtains the lamp testing result for the lamp to be tested. By setting up corresponding preset front-end processing models and preset back-end processing models for the front-end and back-end of the preset deployment detection model data processing, the three models—preset deployment detection model data, preset front-end processing model, and preset back-end processing model—are decoupled, resulting in low coupling between models and facilitating model debugging. When a model has a problem, it is not necessary to adjust the entire model, improving the debugging efficiency of the model on the hardware platform. At the same time, it significantly reduces the deployment cost and complexity of the model. Applying each model to the detection of LED lamps improves both detection accuracy and efficiency. When users need to detect different models of LED lamps, they can flexibly change the models to be deployed according to the preset model templates to meet the detection needs of different models of LED lamps. Attached Figure Description

[0018] The features and advantages of the invention will be more clearly understood by referring to the accompanying drawings, which are schematic and should not be construed as limiting the invention in any way. In the drawings: Figure 1 A flowchart illustrating the testing method for LED lamps provided by the present invention is shown. Figure 2 The diagram shows the structural schematics of each model in the preset model template of the LED lighting fixture testing method provided by the present invention. Figure 3 A schematic diagram of the structure of the LED lamp detection device provided by the present invention is shown; Figure 4 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] LED lighting fixtures are highly efficient and energy-saving lighting devices widely used in homes, offices, commercial spaces, and industrial settings. During the manufacturing process of LED lighting fixtures, circuit components need to be identified and tested to ensure the performance and quality of the fixtures. The lighting module and driver board are crucial components of LED lighting fixtures; the lighting module generates the light source, while the driver board provides power and control signals.

[0021] Specifically, the manufacturing process of LED lighting fixtures requires identifying the hardware status of the lighting modules and driver boards to ensure they are functioning correctly. Simultaneously, it is necessary to test the input current, load PWM, and other characteristics of the LED lighting fixtures to ensure current consistency, response speed, and stability under specific voltages, thereby facilitating product traceability.

[0022] The current testing method for LED lamps first identifies the model and hardware status of the lamp module and driver board, and then uses various types of test circuits to detect the corresponding operating characteristics, such as the input current and load PWM, as mentioned above. The final LED lamp test result is determined by combining the identification results and the test results. However, this testing method is inefficient and cannot meet the needs of large-scale production and testing.

[0023] In conclusion, how to provide a more efficient testing method for LED lighting fixtures that can meet actual production needs is an important issue that the industry urgently needs to address.

[0024] To address the aforementioned problems, this embodiment provides a method for detecting LED lighting fixtures. This method can be used in electronic devices, including but not limited to computers, mobile terminals, etc. Figure 1 This is a flowchart illustrating a method for testing LED lamps according to an embodiment of the present invention, as shown below. Figure 1 As shown, the method may include the following steps: S10. Obtain the model information of the lamp to be tested, and retrieve the corresponding preset model template from the preset database according to the model information. In this embodiment, the preset model template includes a preset deployment detection model, a preset front-end processing model deployed in front of the preset deployment detection model, a preset back-end processing model deployed in back of the preset deployment detection model, and the processing logic relationship between each model. Moreover, the preset model template includes at least one of the above-mentioned preset deployment detection models, and each preset deployment detection model has one of the above-mentioned preset front-end processing models and one of the above-mentioned preset back-end processing models. That is, each preset deployment detection model has a one-to-one corresponding preset front-end processing model and a preset back-end processing model.

[0025] Please see Figure 2 The pre-established database stores several preset deployment detection models, preset front-end processing models, and preset back-end processing models. When storing these models in the database, a mapping relationship can be established between the preset deployment detection model and its corresponding preset front-end processing model, preset back-end processing model, processing logic relationship, and LED lighting fixture model information. This mapping relationship is then stored in the database. Thus, after obtaining the model information of the lighting fixture to be tested, the corresponding preset model template can be retrieved from the database based on the stored mapping relationship.

[0026] It should be noted that the preset model templates for different LED lighting models are usually different. Specifically, at least one piece of information in the preset model template may differ, namely, at least one piece of information in the preset deployment detection model, preset front-end processing model, and preset back-end processing model. In some special cases, the preset model templates for different LED lighting models may be completely identical.

[0027] In particular, each model can also be obtained by encapsulating multiple pre-trained sub-models with different parameters. Each sub-model is equivalent to a network layer of the model and serves as the core operator in the network layer.

[0028] S20. Deploy each model according to the preset model template, and determine the input information of the lamp to be tested according to the preset model template.

[0029] After the preset model template is determined, since the preset model template contains the processing logic relationship, the execution order of all models in the preset model template can be determined according to the processing logic relationship when processing the detection task of the lamp to be tested.

[0030] For example, the preset model template contains three preset deployment detection models: semantic information recognition model A1, current feature detection model B1, load feature detection model C1, and integrated processing model D1. Semantic information recognition model A1 has its corresponding preset front-end processing model A2 and preset back-end processing model A3; current feature detection model B1 has its corresponding preset front-end processing model B2 and preset back-end processing model B3; load feature detection model C1 has its corresponding preset front-end processing model C2 and preset back-end processing model C3; and integrated processing model D1 has its corresponding preset front-end processing model D2 and preset back-end processing model D3. These three preset deployment detection models can be executed in parallel. Integrated processing model D1 will transmit the corresponding data to the corresponding preset back-end processing model after semantic information recognition model A1, current feature detection model B1, and load feature detection model C1 have all been processed. The preset front-end processing model of integrated processing model D1 will then perform corresponding data processing after receiving the above data. Meanwhile, the data transmission relationship between the various models in the preset model template can be obtained according to the execution order. For example, the backend of the preset backend processing model A3, preset backend processing model B3 and preset backend processing model C3 are all preset frontend processing model D2. The preset frontend processing model D2 is responsible for receiving the output data of the above three models, and after performing some business logic processing according to the requirements, it is transmitted to the comprehensive processing model D1 for processing.

[0031] It should be noted that whether each preset front-end processing model needs to perform business logic processing is determined by the preset deployment and detection model of the back-end.

[0032] Based on the execution order and data transmission relationship between the various models, it can be determined whether the backend of each model is connected to other models or outputs to the user, and whether the frontend of each model is connected to other models or external input call interfaces.

[0033] For example, the preset front-end processing model of semantic information recognition model A1 receives external image data and performs preprocessing on the image data, such as deduplication, noise reduction, and correction. The pre-trained model parameters of the preset front-end processing model determine the specific data processing method and the degree of data processing (i.e., the data processing standard). Therefore, when performing the detection task of the lamp to be detected, the preset front-end processing model of semantic information recognition model A1 needs to call an external input interface. The input interface is responsible for transmitting image data that meets the requirements of the preset model to the preset front-end processing model; the preset front-end processing model of current feature detection model B1... It receives detection data from external current testing circuits and constant current detection circuits after testing the lamp under test. The preset front-end processing model converts the detection data in analog signal form into input data in digital signal form. Similarly, the model parameters of the preset front-end processing model, which have been trained, determine the specific data processing method and the degree of data processing. Therefore, when the preset front-end processing model of current feature detection model B1 performs the detection task of the lamp under test, it also needs to call the external input call interface. The input call interface is responsible for transmitting detection data that meets the requirements of the preset model to the preset front-end processing model, such as filtering out detection data with incorrect timestamp information or missing data.

[0034] In this embodiment, each preset front-end processing model has its corresponding preset model requirements. The preset model requirements of different preset front-end processing models are not the same. The input data of the preset front-end processing model can be filtered by each preset model requirements.

[0035] For example, the preset front-end processing model of the integrated processing model D1 receives output data from preset back-end processing models A3, B3, and C3. After confirming that it has received the corresponding output data from these three models, it aggregates the output data and transmits it to the back-end integrated processing model D1. Therefore, when the preset front-end processing model of the semantic information recognition model A1 performs the detection task of the lamp to be detected, it does not need to interface with various external interfaces, but rather interfaces with the output layer or output end of other models.

[0036] S30. Obtain input information, process the input information using a preset model template, and obtain the lamp detection result of the lamp to be detected.

[0037] After the deployment of each model is completed, the deployed models can be debugged and updated according to actual needs. By setting up corresponding preset front-end processing models and preset back-end processing models for the front-end and back-end of the preset deployment detection model data processing, the three models with mapping relationship are decoupled. The low coupling between models facilitates model debugging. When there is a problem with the model, it is not necessary to adjust the entire model, which improves the debugging efficiency of the model on the hardware platform.

[0038] After the model is debugged, the input information is input into the preset front-end processing model, which is in the initial execution order as indicated by the execution order, for corresponding data processing. Then, according to the data transmission relationship and execution order, the corresponding data processing is performed at each step. The lamp detection result is output by the preset back-end processing model, which is in the final execution order as indicated by the execution order.

[0039] For testing multiple batches of the same type of LED lighting fixtures, multiple LED lighting fixtures can be tested simultaneously through the deployment of preset model templates and data processing, improving the testing efficiency. When testing different types of LED lighting fixtures, the models already deployed to the hardware platform can be replaced according to the newly retrieved preset model template. For example, when testing LED lighting fixtures of type A, preset model template A is retrieved from the preset database, and each model in preset model template A is deployed and tested. When testing LED lighting fixtures of type B, preset model template B is retrieved from the preset database. It can be understood that the next step is to deploy and test each model in preset model template B. By deploying the new model to the hardware platform, a testing solution for LED lighting fixtures of type B is formed.

[0040] When there are common models between different preset model templates, such as preset model templates A and B, it is only necessary to replace the models that are different between the two preset model templates and modify the interface call relationship, which significantly improves the deployment cost and complexity of the model used for LED lighting detection.

[0041] Unlike existing detection methods, this embodiment applies more efficient neural network models (such as models using computer vision technology to identify the model and hardware status of lighting modules and driver boards) and data analysis models (such as various circuit parameter analysis models to analyze the operating characteristics of LED lights, and comprehensive analysis models to summarize various data and make comprehensive judgments) to the detection of LED lights. This improves both detection accuracy and efficiency. When users need to detect different models of LED lights, they can flexibly change the models to be deployed according to preset model templates, meeting the detection needs of different models of LED lights.

[0042] The LED lighting fixture testing method of the present invention obtains the model information of the lighting fixture to be tested, retrieves the corresponding preset model template for the lighting fixture to be tested from a preset database according to the model information, the preset model template is the basis for lighting fixture testing, the preset model template includes a preset deployment testing model, a preset front-end processing model deployed in front of the preset deployment testing model, a preset back-end processing model deployed in back of the preset deployment testing model, and the processing logic relationship between each model, then deploys each model according to the preset model template, determines the input information of the lighting fixture to be tested according to the preset model template, processes the input information using the preset model template, and obtains the lighting fixture testing result of the lighting fixture to be tested. By setting up corresponding preset front-end processing models and preset back-end processing models for the front-end and back-end of the preset deployment detection model data processing, the three models—preset deployment detection model data, preset front-end processing model, and preset back-end processing model—are decoupled, resulting in low coupling between models and facilitating model debugging. When a model has a problem, it is not necessary to adjust the entire model, improving the debugging efficiency of the model on the hardware platform. At the same time, it significantly reduces the deployment cost and complexity of the model. Applying each model to the detection of LED lamps improves both detection accuracy and efficiency. When users need to detect different models of LED lamps, they can flexibly change the models to be deployed according to the preset model templates to meet the detection needs of different models of LED lamps.

[0043] In this embodiment, the method may further include the following steps: S40. Based on the lamp test results, trace the source of the fault, locate the source of the lamp fault and the cause of the lamp fault.

[0044] The final output of the lighting test results to the user will indicate whether the lighting fixture is faulty and the cause of the fault. When the lighting test results indicate that the lighting fixture under test is faulty, the LED lighting fixture will first be tagged according to the lighting test results, and the cause of the lighting fixture fault will be traced to locate the source of the fault.

[0045] More specifically, step S40 includes: S41. Extract the causes of lamp malfunctions from each item in the lamp test results. It should be noted that when the lamp test results indicate that the lamp under test has a malfunction, at least one cause of the lamp malfunction can be extracted; when the lamp test results indicate that the lamp under test does not have a malfunction, no cause of the lamp malfunction can be extracted.

[0046] S42. Determine the preset backend processing model for abnormal output data based on the cause of the lamp malfunction.

[0047] S43. Determine the preset deployment detection model corresponding to the preset backend processing model, and determine the source of lamp failure based on the preset deployment detection model.

[0048] Subsequently, since a certain lighting fixture failure may be caused by multiple factors, the preset back-end processing model is first located based on the lighting fixture detection cause, and then the specific preset deployment detection model is determined. The source of the lighting fixture failure is determined by the specific type of the preset deployment detection model.

[0049] After identifying the source of the lighting fixture failure, the fixture is retested using manual verification or other testing tools. By comparing the test results, the performance of the deployed model can be understood.

[0050] In this embodiment, step S10 specifically includes: S11. Acquire image data containing model information of the lamp to be tested. This may include taking multi-angle photos of the assembled lamp to be tested or acquiring images of the original configuration number and factory information.

[0051] S12. Extract and recognize text from the image data to obtain model information.

[0052] In this embodiment, a trained semantic information recognition model, such as the semantic information recognition model A1 described above, is used to extract and recognize text from image data.

[0053] The semantic recognition model uses a self-supervised training method to train the feature extraction layer, that is, to complete the model training using a large amount of historical image data. The feature extraction layer is the core network layer of the semantic recognition model, and its performance directly reflects the feature perception ability of the semantic recognition model. The feature extraction layer is mainly responsible for feature extraction. Therefore, by restoring the feature vectors obtained after processing the feature extraction layer and comparing the differences between the restored image and the historical image data, the feature extraction layer can be trained in a self-supervised manner.

[0054] S13. Retrieve the corresponding preset model template for the lamp to be tested from the preset database according to the model information.

[0055] In this embodiment, step S20 specifically includes: S21. Parse the retrieved preset model template to obtain the execution order and data transfer relationships between the various models in the preset model template. By parsing and processing the logical relationships, the required execution order and data transfer relationships can be obtained.

[0056] S22. Deploy each model according to the execution order and data transfer relationship.

[0057] Similarly, deploying each model to the hardware platform also requires consideration of execution order and data transfer relationships. In this embodiment, the original code of each model is also stored in a preset database, and the corresponding original code of the model is retrieved along with the preset model template.

[0058] More specifically, step S22 includes: S221. Determine the original code of the model, and convert the original code according to the hardware information of the hardware platform to obtain the converted code.

[0059] S222. Generate a model on the hardware platform using conversion code.

[0060] Based on the hardware platform's own hardware information, the original code corresponding to each mode is converted into a new format. Successful conversion yields the converted code deployed on that hardware platform, which can then be used to generate the corresponding model. If the original code conversion fails, the failed parts are replaced until all conversions are successful.

[0061] S223. Debug the model using historical data.

[0062] After generating the model, its performance is adjusted using historical data to ensure that the model's accuracy meets the requirements. Understandably, if the generated model's accuracy does not meet the requirements, the conversion code can be further modified until the model's accuracy meets the requirements.

[0063] S23. Determine the external input call interfaces for all models, and based on all input call interfaces and the preset model requirements of the preset front-end processing models that call the input call interfaces, determine the input information of the lamp to be tested.

[0064] The input data required by each model comes partly from the output data of other models, and partly from the transmission of external input call interfaces. By receiving the transmitted data through all the input call interfaces, and then using the preset model requirements corresponding to the preset front-end processing models that need to call these input call interfaces to filter the transmitted data, the final required input information can be determined.

[0065] The detection device for LED lamps provided in the embodiments of the present invention will be described below. The detection device for LED lamps described below can be referred to in correspondence with the detection method for LED lamps described above.

[0066] To address the aforementioned issues, this embodiment provides an LED lighting fixture detection device designed to generate continuous frame detection boxes, significantly reducing the workload of manual annotation and improving the accuracy of frame center point and size information. Figure 3 This is a schematic diagram of the structure of the LED lamp detection method according to an embodiment of the present invention, as shown below. Figure 3 As shown, the device may include: Template calling module 10 is used to obtain the model information of the lamp to be tested and retrieve the corresponding preset model template from the preset database according to the model information. In this embodiment, the preset model template includes a preset deployment detection model, a preset front-end processing model deployed in front of the preset deployment detection model, a preset back-end processing model deployed in back of the preset deployment detection model, and the processing logic relationship between each model. Moreover, the preset model template includes at least one of the above-mentioned preset deployment detection models, and each preset deployment detection model has one of the above-mentioned preset front-end processing models and one of the above-mentioned preset back-end processing models. That is, each preset deployment detection model has a one-to-one corresponding preset front-end processing model and a preset back-end processing model.

[0067] The pre-established database stores several preset deployment detection models, preset front-end processing models, and preset back-end processing models. When storing these models in the database, a mapping relationship can be established between the preset deployment detection model and its corresponding preset front-end processing model, preset back-end processing model, processing logic relationship, and LED lighting fixture model information. This mapping relationship is then stored in the database. Thus, once the model information of the lighting fixture to be tested is obtained, the corresponding preset model template can be retrieved from the database based on the stored mapping relationship.

[0068] It should be noted that the preset model templates for different LED lighting models are usually different. Specifically, at least one piece of information in the preset model template may differ, namely, at least one piece of information in the preset deployment detection model, preset front-end processing model, and preset back-end processing model. In some special cases, the preset model templates for different LED lighting models may be completely identical.

[0069] In particular, each model can also be obtained by encapsulating multiple pre-trained sub-models with different parameters. Each sub-model is equivalent to a network layer of the model and serves as the core operator in the network layer.

[0070] The model deployment module 20 is used to deploy each model according to the preset model template and determine the input information of the lamp to be tested according to the preset model template.

[0071] After the preset model template is determined, since the preset model template contains the processing logic relationship, the execution order of all models in the preset model template can be determined according to the processing logic relationship when processing the detection task of the lamp to be tested.

[0072] For example, the preset model template contains three preset deployment detection models: semantic information recognition model A1, current feature detection model B1, load feature detection model C1, and integrated processing model D1. Semantic information recognition model A1 has its corresponding preset front-end processing model A2 and preset back-end processing model A3; current feature detection model B1 has its corresponding preset front-end processing model B2 and preset back-end processing model B3; load feature detection model C1 has its corresponding preset front-end processing model C2 and preset back-end processing model C3; and integrated processing model D1 has its corresponding preset front-end processing model D2 and preset back-end processing model D3. These three preset deployment detection models can be executed in parallel. Integrated processing model D1 will transmit the corresponding data to the corresponding preset back-end processing model after semantic information recognition model A1, current feature detection model B1, and load feature detection model C1 have all been processed. The preset front-end processing model of integrated processing model D1 will then perform corresponding data processing after receiving the above data. Meanwhile, the data transmission relationship between the various models in the preset model template can be obtained according to the execution order. For example, the backend of the preset backend processing model A3, preset backend processing model B3 and preset backend processing model C3 are all preset frontend processing model D2. The preset frontend processing model D2 is responsible for receiving the output data of the above three models, and after performing some business logic processing according to the requirements, it is transmitted to the comprehensive processing model D1 for processing.

[0073] It should be noted that whether each preset front-end processing model needs to perform business logic processing is determined by the preset deployment and detection model of the back-end.

[0074] Based on the execution order and data transmission relationship between the various models, it can be determined whether the backend of each model is connected to other models or outputs to the user, and whether the frontend of each model is connected to other models or external input call interfaces.

[0075] For example, the preset front-end processing model of semantic information recognition model A1 receives external image data and performs preprocessing on the image data, such as deduplication, noise reduction, and correction. The pre-trained model parameters of the preset front-end processing model determine the specific data processing method and the degree of data processing (i.e., the data processing standard). Therefore, when performing the detection task of the lamp to be detected, the preset front-end processing model of semantic information recognition model A1 needs to call an external input interface. The input interface is responsible for transmitting image data that meets the requirements of the preset model to the preset front-end processing model; the preset front-end processing model of current feature detection model B1... It receives detection data from external current testing circuits and constant current detection circuits after testing the lamp under test. The preset front-end processing model converts the detection data in analog signal form into input data in digital signal form. Similarly, the model parameters of the preset front-end processing model, which have been trained, determine the specific data processing method and the degree of data processing. Therefore, when the preset front-end processing model of current feature detection model B1 performs the detection task of the lamp under test, it also needs to call the external input call interface. The input call interface is responsible for transmitting detection data that meets the requirements of the preset model to the preset front-end processing model, such as filtering out detection data with incorrect timestamp information or missing data.

[0076] In this embodiment, each preset front-end processing model has its corresponding preset model requirements. The preset model requirements of different preset front-end processing models are not the same. The input data of the preset front-end processing model can be filtered by each preset model requirements.

[0077] For example, the preset front-end processing model of the integrated processing model D1 receives output data from preset back-end processing models A3, B3, and C3. After confirming that it has received the corresponding output data from these three models, it aggregates the output data and transmits it to the back-end integrated processing model D1. Therefore, when the preset front-end processing model of the semantic information recognition model A1 performs the detection task of the lamp to be detected, it does not need to interface with various external interfaces, but rather interfaces with the output layer or output end of other models.

[0078] The lighting fixture detection module 30 is used to acquire input information, process the input information using a preset model template, and obtain the lighting fixture detection result of the lighting fixture to be detected.

[0079] After the deployment of each model is completed, the deployed models can be debugged and updated according to actual needs. By setting up corresponding preset front-end processing models and preset back-end processing models for the front-end and back-end of the preset deployment detection model data processing, the three models with mapping relationship are decoupled. The low coupling between models facilitates model debugging. When there is a problem with the model, it is not necessary to adjust the entire model, which improves the debugging efficiency of the model on the hardware platform.

[0080] After the model is debugged, the input information is input into the preset front-end processing model, which is in the initial execution order as indicated by the execution order, for corresponding data processing. Then, according to the data transmission relationship and execution order, the corresponding data processing is performed at each step. The lamp detection result is output by the preset back-end processing model, which is in the final execution order as indicated by the execution order.

[0081] For testing multiple batches of the same type of LED lighting fixtures, multiple LED lighting fixtures can be tested simultaneously through the deployment of preset model templates and data processing, improving the testing efficiency. When testing different types of LED lighting fixtures, the models already deployed to the hardware platform can be replaced according to the newly retrieved preset model template. For example, when testing LED lighting fixtures of type A, preset model template A is retrieved from the preset database, and each model in preset model template A is deployed and tested. When testing LED lighting fixtures of type B, preset model template B is retrieved from the preset database. It can be understood that the next step is to deploy and test each model in preset model template B. By deploying the new model to the hardware platform, a testing solution for LED lighting fixtures of type B is formed.

[0082] When there are common models between different preset model templates, such as preset model templates A and B, it is only necessary to replace the models that are different between the two preset model templates and modify the interface call relationship, which significantly improves the deployment cost and complexity of the model used for LED lighting detection.

[0083] Unlike existing detection methods, this embodiment applies more efficient neural network models (such as models using computer vision technology to identify the model and hardware status of lighting modules and driver boards) and data analysis models (such as various circuit parameter analysis models to analyze the operating characteristics of LED lights, and comprehensive analysis models to summarize various data and make comprehensive judgments) to the detection of LED lights. This improves both detection accuracy and efficiency. When users need to detect different models of LED lights, they can flexibly change the models to be deployed according to preset model templates, meeting the detection needs of different models of LED lights.

[0084] The LED lighting fixture testing device of the present invention acquires the model information of the lighting fixture to be tested, retrieves the corresponding preset model template for the lighting fixture to be tested from a preset database according to the model information, the preset model template is the basis for lighting fixture testing, the preset model template includes a preset deployment testing model, a preset front-end processing model deployed in front of the preset deployment testing model, a preset back-end processing model deployed in back of the preset deployment testing model, and the processing logic relationship between each model, then deploys each model according to the preset model template, determines the input information of the lighting fixture to be tested according to the preset model template, processes the input information using the preset model template, and obtains the lighting fixture testing result of the lighting fixture to be tested. By setting up corresponding preset front-end processing models and preset back-end processing models for the front-end and back-end of the preset deployment detection model data processing, the three models—preset deployment detection model data, preset front-end processing model, and preset back-end processing model—are decoupled, resulting in low coupling between models and facilitating model debugging. When a model has a problem, it is not necessary to adjust the entire model, improving the debugging efficiency of the model on the hardware platform. At the same time, it significantly reduces the deployment cost and complexity of the model. Applying each model to the detection of LED lamps improves both detection accuracy and efficiency. When users need to detect different models of LED lamps, they can flexibly change the models to be deployed according to the preset model templates to meet the detection needs of different models of LED lamps.

[0085] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical commands in the memory 430 to execute a detection method for LED lamps, the method including: The model information of the lamp to be tested is obtained, and the corresponding preset model template is retrieved from the preset database according to the model information. The preset model template includes a preset deployment detection model, a preset front-end processing model deployed in front of the preset deployment detection model, a preset back-end processing model deployed in back of the preset deployment detection model, and the processing logic relationship between each model. In addition, the preset model template contains at least one preset deployment detection model, each preset deployment detection model has a preset front-end processing model and a preset back-end processing model, and each preset front-end processing model has corresponding preset model requirements. Deploy each model according to the preset model template, and determine the input information of the lamp to be tested according to the preset model template; The system acquires input information, processes it using a preset model template, and obtains the lamp detection results for the lamps to be tested.

[0086] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0087] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, wherein when the program instructions are executed by a computer, the computer is able to execute the LED lamp detection method provided by the above methods, the method comprising: The model information of the lamp to be tested is obtained, and the corresponding preset model template is retrieved from the preset database according to the model information. The preset model template includes a preset deployment detection model, a preset front-end processing model deployed in front of the preset deployment detection model, a preset back-end processing model deployed in back of the preset deployment detection model, and the processing logic relationship between each model. In addition, the preset model template contains at least one preset deployment detection model, each preset deployment detection model has a preset front-end processing model and a preset back-end processing model, and each preset front-end processing model has corresponding preset model requirements. Deploy each model according to the preset model template, and determine the input information of the lamp to be tested according to the preset model template; The system acquires input information, processes it using a preset model template, and obtains the lamp detection results for the lamps to be tested.

[0088] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the aforementioned detection methods for LED lighting fixtures, the method comprising: The model information of the lamp to be tested is obtained, and the corresponding preset model template is retrieved from the preset database according to the model information. The preset model template includes a preset deployment detection model, a preset front-end processing model deployed in front of the preset deployment detection model, a preset back-end processing model deployed in back of the preset deployment detection model, and the processing logic relationship between each model. In addition, the preset model template contains at least one preset deployment detection model, each preset deployment detection model has a preset front-end processing model and a preset back-end processing model, and each preset front-end processing model has corresponding preset model requirements. Deploy each model according to the preset model template, and determine the input information of the lamp to be tested according to the preset model template; The system acquires input information, processes it using a preset model template, and obtains the lamp detection results for the lamps to be tested.

[0089] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for testing LED lighting fixtures, characterized in that, The method includes: The model information of the lamp to be tested is obtained, and the corresponding preset model template is retrieved from the preset database according to the model information. The preset model template includes a preset deployment detection model, a preset front-end processing model deployed in front of the preset deployment detection model, a preset back-end processing model deployed in back of the preset deployment detection model, and the processing logic relationship between each model. In addition, the preset model template contains at least one preset deployment detection model, each preset deployment detection model has a preset front-end processing model and a preset back-end processing model, and each preset front-end processing model has corresponding preset model requirements. Deploy each model according to the preset model template, and determine the input information of the lamp to be tested according to the preset model template; The system acquires input information, processes it using a preset model template, and obtains the lamp detection results for the lamps to be tested.

2. The testing method for LED lamps according to claim 1, characterized in that, The step of deploying each model according to a preset model template and determining the input information of the lamp to be tested according to the preset model template specifically includes: The retrieved preset model template is parsed to obtain the execution order and data transfer relationship between the various models in the preset model template; Deploy each model according to the execution order and data transfer relationships; Identify all external input call interfaces for all models, and based on all input call interfaces and the preset model requirements of the preset front-end processing models that call the input call interfaces, determine the input information of the lamp to be tested.

3. The testing method for LED lamps according to claim 2, characterized in that, The deployment of each model according to the execution order and data transmission relationship specifically includes: The original code of the model is determined, and the original code is converted according to the hardware information of the hardware platform to obtain the converted code; Generate models on hardware platforms using conversion code; Debug the model using historical data.

4. The testing method for LED lamps according to claim 1, characterized in that, The step of obtaining the model information of the lamp to be tested, and retrieving the corresponding preset model template from the preset database based on the model information, specifically includes: Acquire image data containing model information of the lamp to be tested; Text extraction and recognition are performed on image data to obtain model information; Based on the model information, the corresponding preset model template is retrieved from the preset database for the lamp to be tested.

5. The testing method for LED lamps according to claim 4, characterized in that, The trained semantic information recognition model is used to extract and recognize text from image data. The semantic recognition model is trained by a self-supervised training method for the feature extraction layer.

6. The testing method for LED lamps according to claim 1, characterized in that, The method further includes: Based on the lighting fixture test results, the source of the fault is traced to locate the source of the lighting fixture fault and the cause of the fault.

7. The testing method for LED lamps according to claim 6, characterized in that, The process of tracing the source of the fault based on the lamp test results, locating the source of the lamp fault and the cause of the lamp fault, specifically includes: Extract the causes of lamp malfunctions from the lamp test results; A pre-defined backend processing model that determines the abnormality in output data based on the cause of the lighting fixture malfunction; Determine the preset deployment detection model corresponding to the preset backend processing model, and determine the source of the lamp failure based on the preset deployment detection model.

8. A testing device for LED lamps, characterized in that, The device includes: The template calling model is used to obtain the model information of the lamp to be tested, and retrieve the corresponding preset model template from the preset database according to the model information. The preset model template includes a preset deployment detection model, a preset front-end processing model deployed in front of the preset deployment detection model, a preset back-end processing model deployed in back of the preset deployment detection model, and the processing logic relationship between each model. In addition, the preset model template contains at least one preset deployment detection model, each preset deployment detection model has a preset front-end processing model and a preset back-end processing model, and each preset front-end processing model has corresponding preset model requirements. The model deployment module is used to deploy each model according to the preset model template and determine the input information of the lamp to be tested according to the preset model template. The lighting fixture detection module is used to acquire input information, process the input information using a preset model template, and obtain the lighting fixture detection results of the lighting fixture to be detected.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the LED lighting detection method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the detection method for LED lamps as described in any one of claims 1 to 7.

Images