An unattended aging test system and method based on electronic triggering
By simulating human triggering through biomimetic execution and perception units, and combining multi-source data fusion and digital twin prediction, the problems of unrealistic simulation, low accuracy, and reliance on manual operation and maintenance in existing electronic product aging tests are solved, achieving efficient and accurate unattended aging tests.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN TIANBANGDA TECH CO LTD
- Filing Date
- 2026-02-06
- Publication Date
- 2026-06-26
AI Technical Summary
Existing electronic product aging test systems lack simulation of subtle human operation characteristics, resulting in discrepancies between test results and actual user experience. They also fail to respond to environmental changes in real time, and the drift in material properties of actuators affects accuracy. Furthermore, system maintenance relies on manual inspections, and test data is not effectively transformed into optimization knowledge.
It employs biomimetic execution and perception units to simulate artificial triggering, combined with real-time intelligent diagnosis and digital twin prediction through multi-source data fusion, and achieves system self-adaptation and continuous optimization through cloud-edge collaboration, possessing high-fidelity triggering, real-time diagnosis and prediction capabilities.
It improves the accuracy and efficiency of testing, reduces the false fault rate, supports long-term unattended operation, and generates knowledge that can drive product optimization.
Smart Images
Figure CN122283264A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to an unattended aging test system and method based on electronic triggering, belonging to the field of electronic product reliability testing technology. Background Technology
[0002] In the research and development and production of electronic products, physical buttons, touch switches, and other triggering components need to undergo hundreds of thousands or even millions of aging tests to assess their lifespan and reliability. Traditional automated testing equipment often uses actuators such as electromagnets or cylinders, combined with simple continuity detection. Its testing actions are simplistic and lack simulation of the subtle characteristics of human operation, resulting in significant discrepancies between test results and actual user experience, leading to poor equivalence. Furthermore, such systems typically rely on preset, fixed scripts and cannot intelligently adjust based on real-time product responses, environmental changes, or the equipment's own state.
[0003] Existing testing systems generally suffer from the following pain points: First, the testing process is a black box, only able to determine whether the product will eventually fail, but unable to make detailed diagnoses and predictions of the quality of the testing actions themselves or the early degradation of product performance; Second, when facing complex testing environments such as high and low temperatures, the material properties of the actuators will drift, affecting the testing accuracy, and there is a lack of effective online compensation mechanisms; Third, system operation and maintenance heavily rely on manual inspections, resulting in delayed fault detection and handling, making it difficult to achieve large-scale, long-term continuous unattended operation; Fourth, massive amounts of test data are simply recorded and fail to be effectively transformed into knowledge that can drive product optimization and testing strategy iteration.
[0004] Therefore, there is an urgent need for an unattended aging test system that can highly simulate real operation, has real-time intelligent diagnosis and prediction capabilities, and can adapt to changes in environment and task, so as to improve the accuracy, efficiency and value of testing. Summary of the Invention
[0005] To address the aforementioned technical problems, this invention provides an unattended aging test system and method based on electronic triggering, which can achieve high-fidelity bionic triggering, real-time intelligent diagnosis through multi-source data fusion, life prediction based on digital twins, and continuous evolution through cloud-edge collaboration, thereby ensuring high precision, high reliability, and high-value output of the test under unattended conditions.
[0006] The technical solution adopted by this invention to solve its technical problem is:
[0007] An unattended aging test system based on electronic triggering, comprising:
[0008] The biomimetic execution and sensing unit is used to perform simulated human-induced triggering actions on the electronic product under test, and simultaneously collect triggering process data and product response data;
[0009] The edge intelligent diagnostic unit is communicatively connected to the bionic execution and perception unit, and is used to fuse the collected data and the pre-stored historical knowledge base data, and output the system failure probability and product failure probability in real time through weighted calculation.
[0010] The digital twin prediction and optimization unit is used to build a digital twin of a product based on test data and to use a time-series deep learning model to predict remaining life and make maintenance decisions.
[0011] The cloud-edge collaborative management platform is used to upload and aggregate test data, train and optimize cloud models, and wirelessly distribute updated models and knowledge to the edge.
[0012] Preferably, the bionic execution and sensing unit includes a miniature linear motor, a silicone bionic head mounted on its output end, a force sensor, a displacement sensor, and a temperature sensor.
[0013] Preferably, the system further includes an environment adaptation module, which is used to compensate the output force value and trigger time parameter of the biomimetic execution and sensing unit in real time based on the ambient temperature fed back by the temperature sensor and a pre-stored dynamic parameter library of thermodynamic models.
[0014] Preferably, the edge intelligent diagnostic unit is configured to perform weighted fusion calculation based on trigger curve similarity, product response matching degree and historical success rate to obtain the system failure probability and product failure probability respectively; and is equipped with anti-false judgment logic, which only determines the product failure when the trigger execution is normal, the product does not respond and the historical success rate of the corresponding batch is extremely high.
[0015] Preferably, the method for the digital twin prediction and optimization unit to predict remaining life includes: firstly, clustering product degradation patterns based on historical data, then training dedicated prediction models for different degradation pattern categories, and calling the corresponding dedicated model according to the current data pattern during real-time prediction.
[0016] Preferably, the cloud-edge collaborative management platform includes an edge gateway deployed at the test site. The edge gateway is used to perform differential privacy processing on the original test data, extract the de-identified feature values, and then upload them to the cloud.
[0017] Preferably, it also includes a self-recovery strategy execution module, which is used to automatically execute the corresponding recovery strategy based on the risk level determined by the diagnostic results or prediction warning. The recovery strategy includes parameter tuning, task rescheduling, or channel switching.
[0018] Preferably, the system uses an event tracing mechanism to record key state changes and a command query responsibility separation architecture to handle data read and write operations.
[0019] An unattended aging test method based on electronic triggering, applied to the aforementioned system, includes the following steps:
[0020] Execute biomimetic trigger actions and simultaneously collect data from multiple sources;
[0021] By integrating real-time data with historical knowledge for weighted diagnosis, a fault probability is generated.
[0022] Dynamically adjust test parameters or paths based on diagnostic results;
[0023] The digital twin is updated based on test data, and its remaining lifespan is predicted.
[0024] Continuous iterative updates of models and knowledge are achieved through cloud-edge collaboration.
[0025] Preferably, the dynamic adjustment test includes: automatically extending or strengthening the test of products diagnosed as high-risk or with performance drift; the method also includes implementing a matching self-recovery strategy based on the risk level.
[0026] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0027] By simulating the touch of a human hand through a miniature linear motor and a silicone bionic head, and combining multi-sensor closed-loop and environmental adaptive compensation, the trigger equivalence deviation can be controlled at a low level within a wide temperature range, greatly improving the authenticity and accuracy of the test.
[0028] By innovatively integrating data from three sources—the trigger end, the product end, and the historical knowledge base—and through weighted calculation and anti-false positive logic, it can clearly distinguish between system faults and product faults, significantly reducing the fault false positive rate from over 30% in traditional methods.
[0029] By leveraging digital twin technology and time series models such as LSTM, we can achieve probabilistic prediction of the remaining product lifespan and continuously iterate and optimize the diagnostic and predictive capabilities of the entire testing system through cloud-edge collaborative OTA updates.
[0030] It has multi-layered self-recovery strategies and distributed node management capabilities, which can automatically respond to equipment anomalies, realize task migration, and support panoramic monitoring, ensuring stable operation of large-scale, long-term continuous testing and reducing reliance on human resources.
[0031] Through differential privacy-preserving data aggregation and analysis, global knowledge such as supplier quality profiles and common fault warnings can be generated, providing direct data support for R&D improvement and supply chain management. Attached Figure Description
[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.
[0033] Figure 1 This is a schematic diagram of the overall architecture of the unattended aging test system provided by the present invention;
[0034] Figure 2 A flowchart of the intelligent diagnosis and adaptive testing process provided by this invention;
[0035] Figure 3 This is a schematic diagram of the prediction and self-recovery closed loop based on digital twins provided by the present invention;
[0036] Figure 4 This is a schematic diagram of cloud-edge collaborative data flow and model update provided by the present invention. Detailed Implementation
[0037] 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, and 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.
[0038] Example 1
[0039] like Figure 1 As shown, this embodiment provides an unattended aging test system based on electronic triggering. The system is based on a cloud-edge collaborative digital twin platform and supports a three-in-one intelligent closed loop consisting of a physical layer, a cognitive layer, and an evolutionary layer. The physical layer is a biomimetic execution and perception unit, the cognitive layer is an edge intelligent diagnostic unit, and the evolutionary layer is a digital twin prediction and optimization unit.
[0040] Furthermore, the core of the bionic execution and sensing unit is a replaceable silicone bionic head driven by a miniature linear motor. The Shore hardness of the silicone head can be selected between A30° and A70° according to the needs of the test product to simulate the pressing sensation of different softness and hardness. When the motor executes the preset force-displacement-time curve action, the miniature force sensor and laser displacement sensor integrated on the drive shaft synchronously collect the complete force-displacement curve for each trigger. At the same time, a vibration accelerometer collects micro-vibration signals in the 5-100Hz frequency band. The system actively injects random micro-vibrations with an amplitude of 0.1 to 0.3N to simulate muscle tremors during human operation, making the trigger action more realistic.
[0041] To address the impact of high and low temperature environments on testing, the biomimetic execution and sensing unit incorporates a temperature sensor and temperature-compensated strain gauges. The environmental adaptation module continuously monitors the ambient temperature. When the temperature changes, the module does not use a fixed compensation value. Instead, it queries a pre-stored dynamic parameter library of thermodynamic models to determine the rate of change of parameters such as the elastic modulus of the silicone material and motor efficiency at the current temperature, and dynamically calculates correction coefficients for the target force curve and execution time. These coefficients are applied in real-time to the motion control algorithm to ensure that the physical effects of the mechanical excitation applied to the product remain consistent across different temperatures.
[0042] The edge intelligent diagnostic unit is deployed on the edge computing device at the testing station and is required to complete a diagnostic decision within 100 milliseconds after the triggered action. Its workflow is as follows: Figure 2 As shown.
[0043] Specifically, after each triggered action is completed, the edge intelligent diagnostic unit simultaneously receives three data streams: the force-displacement curve and vibration signal from the sensing unit; the current jump waveform and communication response delay time from the product monitoring circuit; and the historical test baseline and historical statistical success rate of the product model and batch retrieved from the local cache.
[0044] The diagnostic process employs weighted calculations based on multi-source data fusion, using a weighted decision-making algorithm. First, the collected force-displacement curve is dynamically time-warped compared to a standard curve to obtain a trigger curve similarity between 0 and 1. Second, the product response characteristics are compared with normal response fingerprints to obtain a product response matching degree. Finally, this is combined with the historical success rate of this batch.
[0045] The weighted calculation is performed as follows: 1-trigger curve similarity is considered the evidence weight for system execution anomalies, 1-product response matching degree is considered the evidence weight for product anomalies, and 1-historical success rate is considered the uncertainty weight of statistical priors. Different weight combinations are set for system failures and product failures. In one embodiment, when calculating the probability of system failure, more emphasis is placed on execution process evidence; when calculating the probability of product failure, more emphasis is placed on product response evidence. Through this weighted calculation, two independent probability values are obtained.
[0046] To prevent misjudging system instability as product failure, the edge intelligent diagnostic unit incorporates anti-misjudgment logic. It will only ultimately determine a product failure when the following three conditions are met:
[0047] The trigger curves are extremely similar;
[0048] The product is completely unresponsive;
[0049] This batch of products has a historically very high success rate.
[0050] For cases that are suspicious but do not meet this logic, the system will generate a visual review report containing a comparison chart of abnormal data and standard data, displaying the abnormal curve and the standard curve overlaid for remote experts to assist in the judgment.
[0051] The digital twin prediction and optimization unit deploys some lightweight models at the edge for real-time early warning, while the core model runs in the cloud. For example... Figure 3 As shown, its core is to build a digital twin of the product that evolves synchronously with the physical test body.
[0052] In the cloud, the system extracts 12-dimensional degradation feature sequences from massive historical test data streams, such as the peak attenuation rate of the force curve, changes in rebound slope, and temporal drift of response delay. The predictive model is trained using a clustering-then-specialized strategy: first, an unsupervised clustering method is used to perform on the degradation feature sequences of all historical products, identifying several typical degradation patterns hidden in the data. Then, for each degradation pattern, corresponding sample data is collected to train a dedicated long short-term memory neural network model. Specifically, when predicting the remaining life of a newly tested product, the system first compares its real-time uploaded degradation feature sequence with each cluster center to determine its degradation pattern category, and then calls an LSTM model specifically trained for that category for prediction. The model outputs not a fixed point, but a probability distribution of remaining life; that is, the output shows a failure probability exceeding 90% after 60,000 cycles, providing a quantitative basis for preventative maintenance.
[0053] Based on the prediction results, the digital twin can conduct sandbox simulations in virtual space. In one embodiment, when it is predicted that the performance of a motor at a certain test node will degrade after 100 hours, the system will simulate the impact of different maintenance strategies on the overall test plan, such as immediate replacement, adjusting parameters to reduce operating speed, and scheduling backup nodes, thereby selecting the globally optimal decision and automatically generating maintenance work orders or optimization suggestions.
[0054] like Figure 4 As shown, the cloud-edge collaborative management platform adopts a three-layer architecture: edge nodes, edge gateways, and the cloud. Each test station acts as an edge node, performing real-time control and diagnostics. Multiple nodes are managed by a single edge gateway, which is responsible for performing differential privacy processing on sensitive data such as raw force curves uploaded by the nodes. This involves adding carefully calculated random noise to the data, making it impossible to reconstruct the original waveform from the processed feature values, thus protecting the privacy of core processes while facilitating data aggregation and utilization. The processed feature data is then uploaded to the cloud.
[0055] The cloud integrates data from various test sites worldwide to update the global knowledge base and train complex models. Every week, the cloud packages and distributes the newly trained prediction model, optimized diagnostic weight parameters, and new fault cases to all edge gateways and nodes via over-the-air (OTA) download technology, completing a capability iteration.
[0056] Furthermore, the electronically triggered unattended aging test system also incorporates a self-recovery strategy execution module, which triggers corresponding actions based on the risk level. In one embodiment, if the diagnostics detect a trend of increasing response delay in a product, the self-recovery strategy execution module may automatically extend the number of tests for that product for in-depth investigation. If the predictive model warns of an impending failure of a certain actuator, the self-recovery strategy execution module will use distributed node management to dynamically migrate the queue of products under test on that node to other healthy nodes in advance, and then safely shut down that node, achieving seamless switching and zero data loss.
[0057] To ensure reliable system operation, the underlying architecture employs event sourcing, with all critical operations persistently recorded as immutable events, supporting end-to-end traceability. Simultaneously, a command query responsibility separation architecture is adopted, separating write operations for sending control commands from read operations for querying monitoring data, ensuring they do not interfere with each other and guaranteeing system responsiveness under high concurrency.
[0058] Example 2
[0059] This embodiment provides a test method for an unattended aging test system based on electronic triggering, which is implemented according to the following steps in a complete test cycle:
[0060] Startup and parameter loading: Before the test begins, the system loads the exclusive parameter package for the current product model, including the target force curve library, normal response fingerprint, and historical statistical information.
[0061] Execution and Acquisition: The bionic execution unit executes a high-fidelity trigger once, and at the same time, the three sources of data are collected synchronously.
[0062] Fusion Diagnosis: The edge intelligent diagnostic unit completes weighted calculations within hundreds of milliseconds and outputs the failure probability of the system and product.
[0063] Adaptive decision-making and execution:
[0064] If the diagnosis is normal, the process continues.
[0065] If the system failure probability exceeds the limit, the system will trigger a three-level process in sequence: automatic retry, switching to the backup channel, and safety shutdown alarm.
[0066] If the product failure probability exceeds the standard and passes the anti-false positive logic, then mark the product as faulty, terminate the test for that product, and generate a report.
[0067] If early performance drift of a product is identified, an automatic decision is made to extend the total number of tests.
[0068] Data Upload and Prediction: After single-item or batch testing is completed, the feature data is anonymized and uploaded to the cloud to update the digital twin. The cloud model predicts the remaining lifespan probability of the online product and generates maintenance recommendations.
[0069] Closed-loop evolution: The cloud periodically pushes optimized intelligent models to the edge via OTA, enabling stronger diagnostic and predictive capabilities for the next test cycle. Simultaneously, the effectiveness data of self-recovery actions is fed back into the system to optimize the strategy library.
[0070] In summary, the above systems and methods have enabled a leap from fixed automation to intelligent self-adaptation in aging testing. While improving testing accuracy and efficiency, they have also deeply explored the value of test data, providing a powerful next-generation tool for the reliability assessment and quality improvement of electronic products.
[0071] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. An unattended aging test system based on electronic triggering, characterized in that, include: The biomimetic execution and sensing unit is used to perform simulated human-induced triggering actions on the electronic product under test, and simultaneously collect triggering process data and product response data; The edge intelligent diagnostic unit is communicatively connected to the bionic execution and perception unit, and is used to fuse the collected data and the pre-stored historical knowledge base data, and output the system failure probability and product failure probability in real time through weighted calculation. The digital twin prediction and optimization unit is used to build a digital twin of a product based on test data and to use a time-series deep learning model to predict remaining life and make maintenance decisions. The cloud-edge collaborative management platform is used to upload and aggregate test data, train and optimize cloud models, and wirelessly distribute updated models and knowledge to the edge.
2. The unattended aging test system based on electronic triggering according to claim 1, characterized in that, The bionic execution and sensing unit includes a miniature linear motor, a silicone bionic head mounted on its output end, a force sensor, a displacement sensor, and a temperature sensor.
3. The unattended aging test system based on electronic triggering according to claim 1, characterized in that, The system also includes an environment adaptation module, which is used to compensate the output force value and trigger time parameter of the biomimetic execution and sensing unit in real time based on the ambient temperature fed back by the temperature sensor and a pre-stored dynamic parameter library of thermodynamic models.
4. The unattended aging test system based on electronic triggering according to claim 1, characterized in that, The edge intelligent diagnostic unit is configured to perform weighted fusion calculations based on trigger curve similarity, product response matching degree, and historical success rate to obtain the system failure probability and product failure probability, respectively; and is equipped with anti-false judgment logic, which only determines the product failure when the trigger execution is normal, the product does not respond, and the historical success rate of the corresponding batch is extremely high.
5. The unattended aging test system based on electronic triggering according to claim 1, characterized in that, The method for the digital twin prediction and optimization unit to predict remaining life includes: firstly, clustering product degradation patterns based on historical data; then, training dedicated prediction models for different degradation pattern categories; and finally, calling the corresponding dedicated model based on the current data pattern during real-time prediction.
6. The unattended aging test system based on electronic triggering according to claim 1, characterized in that, The cloud-edge collaborative management platform includes an edge gateway deployed at the test site. The edge gateway is used to perform differential privacy processing on the original test data, extract the de-identified feature values, and then upload them to the cloud.
7. The unattended aging test system based on electronic triggering according to claim 1, characterized in that, It also includes a self-recovery strategy execution module, which is used to automatically execute the corresponding recovery strategy based on the risk level determined by the diagnostic results or prediction warning. The recovery strategy includes parameter tuning, task rescheduling or channel switching.
8. The unattended aging test system based on electronic triggering according to claim 1, characterized in that, The system uses an event tracing mechanism to record key state changes and a command query responsibility separation architecture to handle data read and write operations.
9. An unattended aging test method based on electronic triggering, characterized in that, The system applied to any one of claims 1-8 includes the following steps: Execute biomimetic trigger actions and simultaneously collect data from multiple sources; By integrating real-time data with historical knowledge for weighted diagnosis, a fault probability is generated. Dynamically adjust test parameters or paths based on diagnostic results; The digital twin is updated based on test data, and its remaining lifespan is predicted. Continuous iterative updates of models and knowledge are achieved through cloud-edge collaboration.
10. The unattended aging test method based on electronic triggering according to claim 9, characterized in that, The dynamic adjustment test includes automatically extending or strengthening the test of products diagnosed as high-risk or with performance drift; the method also includes implementing a matching self-recovery strategy based on the risk level.