A method, system, apparatus, and storage medium for testing a power supply chip

Abstract

The application discloses a kind of methods, systems, devices and storage medium for testing power supply chip, it is related to chip testing field.In the method, first operating state data of power supply chip is predicted according to fault type and fault parameter by preset generative adversarial network;Target fault is generated according to fault type and fault parameter, second operating state data of power supply chip is collected, and second operating state data is compared with first operating state data;When there is second target data in second operating state data and the target difference of corresponding first target data in first operating state data is greater than threshold value, determine preset recovery strategy according to second target data;When power supply chip recovers to preset state, third operating state data of power supply chip is collected, and the test result of power supply chip is determined according to the comparison result of third operating state data and preset recovery strategy.The technical scheme provided in the application is implemented, and the effects of improving the efficiency and accuracy of test are achieved.

Description

Technical Field

[0001] This application relates to the technical field of chip testing, specifically to a method, system, device, and storage medium for testing power chips. Background Technology

[0002] With the rapid development of electronic technology, power supply chips, as core components of various electronic devices, are of paramount importance in terms of performance and stability. To ensure the reliability of power supply chips in practical applications, comprehensive testing is an indispensable step. Traditional testing methods mainly rely on manually setting faults and observing the chip's response; however, this method is inefficient and prone to overlooking potential problems.

[0003] Currently, although some automated testing solutions have been proposed, they typically can only detect pre-defined fault types and are often ineffective against unknown or complex fault modes. Furthermore, these solutions also have limitations in fault generation, data comparison, and result determination, failing to meet increasingly stringent testing requirements.

[0004] Therefore, how to efficiently and accurately test the performance and stability of power chips has become an urgent technical problem to be solved. Summary of the Invention

[0005] This application provides a method, system, device, and storage medium for testing power supply chips, which improves the accuracy and efficiency of testing.

[0006] The first aspect of this application provides a method for testing a power supply chip, applied to a chip testing platform, the method comprising:

[0007] The first operating status data of the power chip is collected, and the fault type and fault parameters are predicted based on the first operating status data through a preset generative adversarial network. The first operating status data includes operating temperature, load, output voltage and output current.

[0008] A target fault is generated based on the fault type and fault parameters. The second operating status data of the power chip is collected, and the second operating status data is compared with the first operating status data.

[0009] When the target difference between the second target data and the corresponding first target data in the first operating status data is greater than a threshold, a preset recovery strategy is determined based on the second target data.

[0010] When the power chip recovers to the preset state, the third operating state data of the power chip is collected, and the test result of the power chip is determined based on the comparison result of the third operating state data and the preset recovery strategy.

[0011] By employing the above technical solution, a pre-defined Generative Adversarial Network (GAN) can accurately predict potential fault types and parameters based on the first operating state data of the power chip. This method not only improves the efficiency of fault detection but also provides early warning before faults occur, enabling timely repair and prevention. A target fault is generated based on the predicted fault type and parameters, and the second operating state data of the power chip is collected and compared with the first operating state data. This fault simulation method realistically reflects the operation of the power chip under fault conditions, providing real data support for subsequent fault recovery and testing. When anomalies are found in the second operating state data, a pre-defined recovery strategy is determined based on the abnormal data. This recovery strategy based on actual fault data is more targeted and practical, and can more effectively repair power chip faults. After the power chip recovers to the pre-defined state, third operating state data is collected and compared with the pre-defined recovery strategy. This comparison result reflects the effectiveness of the recovery strategy and the performance status of the power chip after repair, thereby improving the reliability and accuracy of the test results. The entire testing process is automated and intelligent, reducing human intervention and errors, and improving testing efficiency. This method allows for comprehensive and systematic testing of power chips, enabling the timely detection and repair of potential problems, thereby enhancing the stability and reliability of the entire system.

[0012] Optionally, the step of predicting the fault type and fault parameters based on the first operating state data using a preset generative adversarial network includes:

[0013] Acquire historical operating status data, which includes normal status data and fault status data, and assign labels to the fault status data to represent fault type and fault parameters;

[0014] Construct a generator and a discriminator, and iteratively train the discriminator and the generator using the historical running state data until the preset function meets the requirements to construct the preset generative adversarial network;

[0015] The first operating status data is input into the preset generative adversarial network to predict fault status data.

[0016] By employing the aforementioned technical solution, historical operating state data containing both normal and fault state data is acquired. Fault state data is then labeled to represent specific fault types and parameters, providing rich samples for the training of Generative Adversarial Networks (GANs). This enables GANs to learn the mapping relationship from normal to fault states, thereby achieving high-precision fault prediction. By iteratively training the generator and discriminator until a preset function (such as a loss function) meets the requirements, the GAN model's strong generalization ability can be ensured. This means the model can not only handle fault conditions in the training data but also make relatively accurate predictions of unseen fault types and parameters. Once the GAN model is trained, real-time collected first-state data of the power chip can be quickly input into the model to predict potential fault state data. This strong real-time fault detection capability is crucial for ensuring the stable operation of the power chip. By predicting fault types and parameters using GANs, fault simulation and recovery strategy testing can be performed before actual faults occur, thus avoiding downtime and maintenance costs caused by actual faults. This significantly improves testing efficiency and reduces testing costs. Because GANs can predict specific fault types and parameters, targeted recovery strategies can be formulated based on the prediction results. This customized recovery strategy better meets actual needs and improves the efficiency and success rate of fault recovery. By timely identifying and predicting potential fault types and parameters, corresponding preventative or corrective measures can be taken before faults occur, thereby enhancing the stability and reliability of the entire system. This is especially important for critical systems that require long-term stable operation.

[0017] Optionally, the step of iteratively training the discriminator and the generator using the historical running state data until a preset function meets the requirements to construct the preset generative adversarial network includes:

[0018] The parameters of the generator are fixed, and the discriminator is trained to distinguish between real running state data and generated running state data. The parameters of the discriminator are then updated through backpropagation and gradient descent.

[0019] The parameters of the discriminator are fixed, the generator is trained to construct and generate runtime data, and the parameters of the generator are updated through backpropagation and gradient descent.

[0020] The discriminator and the generator are trained alternately until a preset number of iterations or performance convergence is reached.

[0021] By employing the above technical solution, during discriminator training, the generator's parameters are fixed, allowing the discriminator to learn how to distinguish between real historical running state data and fake data generated by the generator. This process continuously updates the discriminator's parameters through backpropagation and gradient descent, enabling it to more accurately identify real and generated data, thereby improving the discriminator's discriminative ability. During generator training, the discriminator's parameters are fixed, allowing the generator to learn how to construct running state data that is as similar as possible to real data. Similarly, this process also continuously updates the generator's parameters through backpropagation and gradient descent, enabling it to generate fake data that is increasingly close to real data, thereby enhancing the generator's generative ability. By alternately training the discriminator and generator, the GAN model can continuously engage in a game and optimization between the two. Improvements in the discriminator force the generator to generate more realistic data, while the more realistic data generated by the generator further prompts the discriminator to improve its discriminative ability. This mutually reinforcing training process helps improve the overall performance of the GAN model. The iterative training process continues until the preset number of iterations is reached or the model performance converges. This ensures that the GAN model achieves the expected results after sufficient training, thus constructing a stable and reliable pre-defined GAN model. Through ample iterative training, the GAN model can maintain stable performance on various operational state data and also has a certain generalization ability to unseen data. This enables the GAN model to predict fault types and parameters more stably and reliably in practical applications.

[0022] Optionally, the step of predicting the fault type and fault parameters based on the first operating state data using a preset generative adversarial network includes:

[0023] A noise signal is generated based on the first operating state data using a preset noise model, and the noise signal is added to the input terminal of the power chip.

[0024] The third operating state data of the power chip is collected, and the fault type and fault parameters are predicted based on the third operating state data through the preset generative adversarial network.

[0025] By employing the above technical solution, a noise signal is generated using a preset noise model and added to the input terminal of the power chip, simulating the noise interference that the power chip may encounter in the actual operating environment. This simulation method is closer to reality, making subsequent fault prediction more accurate and reliable. Inputting the noisy third-state data into a preset GAN allows the GAN to learn the impact of noise on the power chip's operating state and predict possible fault types and parameters. Since the GAN is trained on a large amount of historical data, its prediction results have high accuracy. By adding a noise signal to the power chip's input terminal, the performance of the power chip under different noise conditions can be tested. This comprehensive testing method helps to discover potential problems of the power chip in different scenarios, improving the comprehensiveness and effectiveness of the test. Compared to traditional physical testing methods, this method can simulate and predict faults in a virtual environment, eliminating the need to actually build a test environment and prepare test samples. This significantly reduces testing costs and time, and improves testing efficiency. Based on the fault type and fault parameters predicted by the GAN, targeted recovery strategies can be formulated. This customized recovery strategy can better meet actual needs and improve the efficiency and success rate of fault recovery. By timely detection and prediction of potential fault types and parameters, corresponding measures can be taken to prevent or repair faults before they occur, thereby enhancing the stability and reliability of the entire system.

[0026] Optionally, generating a noise signal based on the first operating state data using a preset noise model includes:

[0027] The load data in the first operating state data is mapped to the corresponding noise parameters through the preset noise model, and a noise signal is generated by combining the noise parameters with a random number generator. The noise parameters include the amplitude, frequency and phase of the noise.

[0028] By adopting the above technical solution, mapping the load data of the power supply chip to noise parameters ensures that the generated noise signal is closely related to the actual operating state of the power supply chip. This method generates noise signals that more closely resemble the noise conditions that may occur in the actual operating environment, thus improving the realism of the noise simulation. Since the noise signal is generated based on the actual load data of the power supply chip, when adding the noise signal to the input of the power supply chip for fault simulation, it can more accurately simulate the performance of the power supply chip under different load conditions. This helps improve the accuracy of fault simulation and makes the prediction results more reliable. By adjusting noise parameters (such as amplitude, frequency, and phase), the characteristics of the generated noise signal can be flexibly controlled. This flexibility allows for the customization of noise signals according to different test requirements to simulate different types of noise interference, thereby more comprehensively evaluating the performance of the power supply chip. Noise signals can be generated in a virtual environment using preset noise models and random number generators, eliminating the need to actually build a test environment and prepare test samples. This significantly reduces testing costs and time, and improves testing efficiency. Because noise signals can be generated based on the load data of the power supply chip, they can be customized according to specific test requirements. This customized testing method can better meet actual needs and improve the relevance and effectiveness of testing. By simulating different types of noise interference and predicting potential fault types and parameters, appropriate measures can be taken to prevent or repair faults before they occur. This helps enhance the stability and reliability of the entire system, and improves its operating efficiency and lifespan.

[0029] Optionally, determining the preset recovery strategy based on the second target data includes:

[0030] The similarity measurement method is used to calculate the similarity between the second target data and the data in the fault mode library to match the fault type, and the fault parameters are determined according to the fault type.

[0031] The matching value of the recovery strategy in the preset strategy database is calculated based on the fault type and the fault parameters. The matching value is obtained by weighting the recovery time, recovery success rate and percentage of the recovery strategy to reach the original state after recovery. The first recovery strategy is selected as the preset recovery strategy. The first recovery strategy and the second recovery strategy are any two recovery strategies in the preset strategy database. The matching value of the first recovery strategy is greater than the matching value of the second recovery strategy.

[0032] By employing the above technical solution and using a similarity measurement method to calculate the similarity between the second target data and the data in the fault mode library, the corresponding fault type can be matched more accurately. This method, based on historical data and pattern recognition technology, can effectively improve the accuracy of fault identification and reduce the possibility of misjudgment and missed judgment. After matching the fault type, specific fault parameters are determined according to the fault type. Accurate determination of fault parameters is crucial for subsequent fault recovery and repair, helping engineers and technicians to more accurately understand the nature and cause of the fault, thereby formulating more effective recovery strategies. Based on the fault type and fault parameters, the matching value of each recovery strategy is calculated from the preset strategy database. The calculation of the matching value considers multiple factors such as the recovery time, recovery success rate, and percentage of the original state after recovery, and is obtained through a weighted average. This method comprehensively considers multiple factors to select the most suitable recovery strategy for the current fault. Selecting the recovery strategy with the highest matching value as the preset recovery strategy ensures high recovery efficiency and success rate during the recovery process. This not only reduces the time and cost required for recovery but also reduces the risks and losses caused by recovery failure. The recovery strategies in the preset strategy database can be customized and adjusted according to actual needs. When encountering new fault types or parameters, new recovery strategies can be added or existing strategies modified to adapt to the new situation. This flexibility allows recovery strategies to continuously adapt to the ever-changing system environment and business needs. By timely and accurately identifying fault types and parameters and selecting appropriate recovery strategies for repair, the impact of faults on system stability and reliability can be minimized. This helps ensure the system can continue to operate stably, meeting business needs and user expectations.

[0033] Optionally, determining the test result of the power chip based on the comparison result of the third operating state data and the preset recovery strategy includes:

[0034] Calculate the first difference between the first recovery time of the preset recovery strategy and the second recovery time in the third operating state data, and calculate the second difference between the first percentage of the preset recovery strategy that reaches the original state after recovery and the second percentage of the third operating state data that reaches the original state after recovery. Then, perform a weighted summation of the first difference and the second difference to obtain a comparison value.

[0035] When the comparison value is less than the threshold, the power chip is determined to be in a normal state;

[0036] When the comparison value is greater than or equal to the threshold, the power chip is determined to be in an abnormal state.

[0037] By employing the above technical solution, the difference between the recovery time and the percentage of the original state reached after recovery in the third operating state data, calculated using the preset recovery strategy, can accurately quantify the difference between the actual and expected recovery effects. This quantification method makes the test results more objective and comparable. Using the recovery time and the percentage of the original state reached after recovery as evaluation indicators, and obtaining a comparison value by weighted summation, multiple factors can be comprehensively considered to evaluate the performance of the power chip. This comprehensive approach improves the accuracy of the test and reduces the possibility of misjudgment due to a single indicator. Based on the comparison value and the preset threshold, it is possible to quickly determine whether the power chip is in a normal or abnormal state. This rapid judgment method helps to promptly identify and resolve potential problems, improving the efficiency of fault handling. The threshold setting can be customized according to actual needs and system characteristics. By adjusting the threshold, the sensitivity and accuracy of the test can be flexibly controlled to adapt to the testing requirements in different scenarios. By comparing the preset recovery strategy with the actual operating state data, performance anomalies or potential problems of the power chip can be detected in a timely manner. This helps to take preventive measures before a fault occurs, improving the stability and reliability of the system. When the power chip is determined to be in an abnormal state, targeted repair work can be carried out based on the comparison results and the guidance of the preset recovery strategy. This can reduce repair time and cost and improve repair efficiency.

[0038] A second aspect of this application provides a system for testing power supply chips, including a prediction module, a comparison module, a recovery module, and an execution module, wherein:

[0039] The prediction module is configured to collect first operating status data of the power chip and predict fault type and fault parameters based on the first operating status data through a preset generative adversarial network. The first operating status data includes operating temperature, load, output voltage and output current.

[0040] The comparison module is configured to generate a target fault based on the fault type and fault parameters, collect the second operating status data of the power chip, and compare the second operating status data with the first operating status data.

[0041] The recovery module is configured to determine a preset recovery strategy based on the second target data when the target difference between the second target data in the second running status data and the corresponding first target data in the first running status data is greater than a threshold.

[0042] The execution module is configured to collect third operating state data of the power chip when the power chip recovers to a preset state, and determine the test result of the power chip based on the comparison result of the third operating state data and the preset recovery strategy.

[0043] A third aspect of this application provides an electronic device including a processor, a memory, a user interface, and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any of the foregoing.

[0044] A fourth aspect of this application provides a computer-readable storage medium storing instructions that, when executed, perform the method described in any of the preceding descriptions.

[0045] In summary, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0046] 1. By using a pre-defined Generative Adversarial Network (GAN) to predict the first operating state data of the power chip, potential fault types and parameters can be predicted with relatively high accuracy. Based on the predicted fault types and parameters, a target fault is generated, and the second operating state data of the power chip is collected to simulate realistic fault scenarios. By comparing the simulated data with the original data, the performance of the power chip under fault conditions can be evaluated more accurately.

[0047] 2. When the target difference between the second operating state data and the first operating state data exceeds a threshold, performance changes in the power supply chip can be quickly identified. This highly sensitive fault detection method helps to promptly identify potential problems and improves the reliability of testing.

[0048] 3. Determining a preset recovery strategy based on the second target data ensures that the most suitable recovery strategy is selected when the power chip fails. This method comprehensively considers factors such as recovery time, recovery success rate, and the percentage of the system returning to its original state after recovery, ensuring the efficiency and accuracy of the recovery process.

[0049] 4. After the power chip returns to the preset state, data from the third operating state is collected, and the test results are determined based on the comparison between the third operating state data and the preset recovery strategy. This method comprehensively considers multiple indicators, ensuring the objectivity and reliability of the test results. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating the method for testing a power supply chip disclosed in an embodiment of this application;

[0051] Figure 2 This is a schematic diagram of a system for testing power supply chips disclosed in an embodiment of this application;

[0052] Figure 3This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of this application.

[0053] Explanation of reference numerals in the attached figures: 201, prediction module; 202, comparison module; 203, recovery module; 204, execution module; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. Detailed Implementation

[0054] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0055] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0056] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0057] This embodiment discloses a method for testing power supply chips. Figure 1 This is a flowchart illustrating a method for testing a power supply chip disclosed in an embodiment of this application, as shown below. Figure 1 As shown, it includes the following steps:

[0058] S110. Collect the first operating status data of the power chip, and predict the fault type and fault parameters based on the first operating status data through a preset generative adversarial network. The first operating status data includes operating temperature, load, output voltage and output current.

[0059] Power supply chips generate various status data during operation, reflecting their current working condition. The initial operating status data of a power supply chip typically includes, but is not limited to, the following key parameters:

[0060] Operating temperature: Power chips generate heat during operation, and temperature is an important indicator for evaluating their operating status and heat dissipation performance. Excessively high temperatures may lead to decreased chip performance or even damage.

[0061] Load: The load reflects the current or voltage capacity currently carried by the power supply chip. Under different loads, the operating state and efficiency of the power supply chip may vary.

[0062] Output voltage: One of the main functions of a power supply chip is to provide a stable output voltage. The accuracy and stability of the output voltage directly affect the performance of the entire system.

[0063] Output current: Output current reflects the power supply chip's ability to power external devices. Similar to output voltage, the accuracy and stability of output current are also important indicators for evaluating the performance of a power supply chip.

[0064] To acquire this operational status data, specialized testing equipment or sensors are typically used to monitor the power chip's operating status in real time and record the data. After acquiring the initial operational status data of the power chip, a pre-defined Generative Adversarial Network (GAN) can be used to predict potential fault types and parameters. A GAN is a deep learning model consisting of a generator and a discriminator. The generator's role is to generate new data based on the input data. For example, it can generate fault types and parameters based on the initial operational status data. The generator learns from a large amount of historical data to simulate the distribution of real data and attempts to generate new data that is as similar as possible to the real data. The discriminator's role is to determine whether the input data is real or generated by the generator. During training, the discriminator continuously improves its discriminative ability to better distinguish between real and generated data. In the scenario of power chip testing, the initial operational status data can be used as input to the GAN, and the generator can predict potential fault types and parameters. Specifically, the generator attempts to simulate features similar to real faults based on the operational status data, while the discriminator evaluates the accuracy of these predictions.

[0065] Optionally, the step of predicting the fault type and fault parameters based on the first operating state data using a preset generative adversarial network includes:

[0066] Acquire historical operating status data, which includes normal status data and fault status data, and assign labels to the fault status data to represent fault type and fault parameters;

[0067] Construct a generator and a discriminator, and iteratively train the discriminator and the generator using the historical running state data until the preset function meets the requirements to construct the preset generative adversarial network;

[0068] The first operating status data is input into the preset generative adversarial network to predict fault status data.

[0069] A large amount of operational status data is collected from the historical operating records of the power supply chip. This historical data includes normal state data and fault state data. Normal state data is generated by the power supply chip under normal operating conditions, while fault state data is generated when the power supply chip malfunctions. For fault state data, corresponding labels need to be assigned to represent the fault type and fault parameters. For example, labels can include fault types such as overheating fault, overload fault, and abnormal output voltage, as well as specific fault parameter values ​​such as temperature threshold and load current. A generative adversarial network (GAN) generator and discriminator are constructed. The generator's task is to generate fake fault data similar to real fault state data, while the discriminator's task is to distinguish between real fault data and the fake fault data generated by the generator. The generator and discriminator are typically implemented using deep learning network models, such as convolutional neural networks (CNNs) or recurrent neural networks (RNNs). These models can be trained using backpropagation and gradient descent algorithms. After iterative training, the generator and discriminator will reach a relatively stable state, at which point the pre-defined generative adversarial network can be constructed. This network has learned to extract fault-related features from operational status data and can predict possible fault types and parameters based on these features. The initial operational status data of the power chip is input into a pre-defined generative adversarial network, which outputs predicted fault status data. This data includes the predicted fault type and fault parameter values.

[0070] By acquiring a large amount of historical operating status data, including normal and fault state data, and assigning labels to the fault state data, rich sample data is provided for GAN training. This enables GAN to learn the mapping relationship from operating status data to fault types and fault parameters. The generator and discriminator of GAN are continuously optimized through iterative training until the preset function meets the requirements, thereby constructing a high-performance preset generative adversarial network. This network model can quickly and accurately predict potential fault types and fault parameters when receiving new operating status data. Due to the powerful learning and generalization capabilities of the GAN model, it can predict fault types based on the operating status data of the power chip, generate corresponding faults based on the predicted fault types, and inject them into the chip testing platform for further testing of the power chip. Traditional power chip testing methods usually require long-term testing and verification in actual operating environments to discover potential faults. However, using the GAN model for fault prediction can analyze and predict a large amount of operating status data in a short time, thereby reducing testing costs and cycles. The GAN model can handle various types of fault data and operating status data and learn the complex relationships between them. This allows testing to cover more fault scenarios and operating states, improving the comprehensiveness and reliability of the tests. GAN models have strong versatility and scalability, and can be adapted to the testing needs of different power supply chip models and specifications by adjusting the network structure and parameters. This makes the testing method more flexible and efficient.

[0071] Optionally, the step of iteratively training the discriminator and the generator using the historical running state data until a preset function meets the requirements to construct the preset generative adversarial network includes:

[0072] The parameters of the generator are fixed, and the discriminator is trained to distinguish between real running state data and generated running state data. The parameters of the discriminator are then updated through backpropagation and gradient descent.

[0073] The parameters of the discriminator are fixed, the generator is trained to construct and generate runtime data, and the parameters of the generator are updated through backpropagation and gradient descent.

[0074] The discriminator and the generator are trained alternately until a preset number of iterations or performance convergence is reached.

[0075] Historical operating status data of the power chip is collected, including normal and fault status data. For fault status data, a unique label is assigned to each fault type and fault parameter. This data will be used as samples to train the GAN model. The network parameters of the generator and discriminator are initialized. The generator is responsible for generating fake data similar to the real operating status data (i.e., generating operating status data), while the discriminator is responsible for distinguishing between real and fake data. When training the discriminator, the generator parameters are fixed, and a batch of real operating status data and a batch of fake data generated by the generator are input into the discriminator. The discriminator's goal is to correctly distinguish between these two sets of data, that is, to give high scores to real data and low scores to fake data. The error of the discriminator in distinguishing between real and fake data is calculated using the backpropagation algorithm, and the discriminator's parameters are updated using the gradient descent algorithm to reduce this error. This process is repeated multiple times until the discriminator can distinguish between real and fake data well. The discriminator parameters are fixed, and only the generator is trained. The generator's goal is to generate fake data that can deceive the discriminator, that is, to make the discriminator mistake fake data for real data. Using a new batch of real-world runtime data as a reference, the generator attempts to generate fake data similar to this data. This fake data is then fed into the discriminator, and the generator's error is calculated based on the discriminator's output. Similarly, the generator's parameters are updated using backpropagation and gradient descent algorithms to reduce this error. The training of the discriminator and generator is performed alternately. In each iteration, the discriminator is trained first, followed by the generator. This process is repeated multiple times until a preset number of iterations is reached or the model performance converges. Specifically, convergence can be determined by observing changes in the loss function. If the loss function remains stable or gradually decreases after multiple iterations, the model can be considered to have converged. Furthermore, other evaluation metrics (such as accuracy and recall) can be used to monitor the model's performance, and training can be stopped when the requirements are met. After the above iterative training process, a well-performing pre-defined generative adversarial network (GAN) is obtained. This network can predict potential fault types and fault parameters based on the runtime data of the power chip. When new runtime data is input into the GAN, it can quickly and accurately generate corresponding fault prediction results.

[0076] By alternating the training of the discriminator and generator, the GAN model can continuously optimize its performance. The discriminator learns to more accurately distinguish between real and generated operating state data, while the generator learns to generate more realistic fake data, thus deceiving the discriminator. This adversarial training process helps the model continuously approach the optimal solution. After training, the GAN model can predict potential fault types and fault parameters based on new operating state data. Because the model has learned the distribution and characteristics of real fault data during training, it can more accurately identify fault-related patterns, improving the accuracy of fault prediction. By training with a large amount of diverse historical operating state data, the GAN model can learn fault characteristics under different scenarios. This makes the model more robust in practical applications and able to adapt to changes in different working conditions and external environments. The GAN model has a fast prediction speed, capable of monitoring the operating status of power chips in real time and quickly providing fault prediction results. This rapid response capability helps to promptly identify potential fault risks and take corresponding preventive measures to avoid the impact of faults on system stability and reliability. The GAN model can automatically learn and adapt to changes in data during training. When new fault types or fault parameters emerge, the model can automatically update its internal parameters to adapt to the new data distribution. This adaptive capability gives GAN models stronger generalization ability, enabling them to cope with various complex real-world application scenarios.

[0077] Optionally, the step of predicting the fault type and fault parameters based on the first operating state data using a preset generative adversarial network includes:

[0078] A noise signal is generated based on the first operating state data using a preset noise model, and the noise signal is added to the input terminal of the power chip.

[0079] The third operating state data of the power chip is collected, and the fault type and fault parameters are predicted based on the third operating state data through the preset generative adversarial network.

[0080] Acquire first-level operating state data of the power supply chip under normal operating conditions. This data may include various parameters such as voltage, current, temperature, and power consumption. This data will serve as the basis for fault prediction. Generate a noise signal based on the first-level operating state data using a pre-defined noise model. This noise model can be designed based on interference factors in the actual working environment (such as electromagnetic interference, temperature fluctuations, etc.) or can be randomly generated. The generated noise signal needs to simulate the impact of these interference factors on the operating state of the power supply chip. Add the generated noise signal to the input terminal of the power supply chip. This can be achieved by adding a controllable noise source to the input line of the power supply chip. After adding the noise signal, the operating state of the power supply chip will be affected to a certain extent, thus simulating different operating conditions and environmental interference. After adding the noise signal, collect third-level operating state data of the power supply chip. This data will reflect the actual operating state of the power supply chip under the influence of the noise signal. By comparing the first-level and third-level operating state data, the impact of the noise signal on the operating state of the power supply chip can be analyzed. Input the third-level operating state data into a pre-defined generative adversarial network (GAN) for fault prediction. This GAN model has been trained on a large amount of historical operating status data and has learned to predict potential fault types and fault parameters from the operating status data.

[0081] By generating noise signals based on the first operating state data using a pre-defined noise model and adding them to the input of the power chip, various interferences and changes that the power chip may encounter in a real operating environment can be simulated. This helps test the performance and reliability of the power chip under adverse conditions, thereby identifying potential failure risks in advance. The third operating state data collected after adding the noise signal contains the actual performance of the power chip under interference. Inputting this data into a pre-defined generative adversarial network (GAN) for fault prediction allows the network to be trained to learn and adapt to fault characteristics under different conditions. This makes the model more robust in practical applications and able to more accurately predict fault types and fault parameters under various conditions. Due to the powerful learning and generalization capabilities of GANs, they can learn the mapping relationship from operating state data to fault types and fault parameters from a large amount of training data. By introducing noise signals, the network can learn a wider range of fault modes and characteristics, thereby improving the accuracy of fault prediction. By collecting and analyzing the operating state data of the power chip under different noise interferences, the performance and reliability bottlenecks of the power chip can be further understood. This provides data support for the optimized design of the power chip, helping to improve the structure, materials, and manufacturing process of the power chip, and improve its performance and reliability.

[0082] Optionally, generating a noise signal based on the first operating state data using a preset noise model includes:

[0083] The load data in the first operating state data is mapped to the corresponding noise parameters through the preset noise model, and a noise signal is generated by combining the noise parameters with a random number generator. The noise parameters include the amplitude, frequency and phase of the noise.

[0084] A preset noise model is a mathematical model that maps load data to noise parameters. This model can be based on physical principles or empirically. In this example, it is assumed that the amplitude, frequency, and phase of the noise are related to the load data. Specifically, a mapping function can be defined to map load data (e.g., load current or load power) to the amplitude, frequency, and phase of the noise. This function can be obtained by fitting experimental data or based on expert experience. Using the preset noise model, the load data in the first operating state data is mapped to the corresponding noise parameters. This typically involves substituting the load data into the mapping function to calculate the corresponding noise amplitude, frequency, and phase. After obtaining the noise parameters, a random number generator is used to generate the actual noise signal. The random number generator can generate a noise signal with these characteristics based on the specified noise amplitude, frequency, and phase. Specifically, a sine wave or a more complex waveform can be used as the basis for the noise signal, and its amplitude, frequency, and phase can be adjusted to match the previously calculated noise parameters. The random number generator ensures that the generated noise signal has a certain degree of randomness and unpredictability, thus better simulating noise interference in a real-world environment. After generating the noise signal, it is fed into the input of the power supply chip. This can be achieved by adding a noise source to the input circuit of the power chip, or by introducing noise signals during the data processing stage through software simulation. After adding the noise signal, the third operating state data of the power chip is collected. This data will reflect the actual operating state of the power chip under noise interference. The third operating state data is then input into a pre-defined generative adversarial network (GAN) for fault prediction. The GAN model will attempt to identify the fault type and fault parameters from this data.

[0085] Mapping load data directly to noise parameters allows for a more accurate simulation of the noise generated by power supply chips under varying loads in real-world operating environments. This method better reflects the behavior of power supply chips under actual operating conditions than simply applying random noise. Since the generated noise signal is based on real load data, it more accurately represents the performance of the power supply chip in a real-world working environment. This more accurate simulation improves the accuracy of Generative Adversarial Networks (GANs) in fault prediction because they can learn fault characteristics closer to the real world. By analyzing the performance of power supply chips under different load and noise conditions, potential design problems can be identified and optimized. This approach helps improve the design of power supply chips, enhancing their stability and reliability. By incorporating simulated noise signals into the training data, GAN models can learn more diverse fault characteristics. This helps the model maintain stable performance when facing various unknown noises and interferences in real-world applications, improving the model's robustness. By adjusting noise parameters (such as amplitude, frequency, and phase), different operating environments and fault scenarios can be simulated. This provides a flexible and controllable environment for the testing and verification of power supply chips, facilitating the identification and optimization of potential problems.

[0086] S120. Generate a target fault based on the fault type and fault parameters, collect the second operating status data of the power chip, and compare the second operating status data with the first operating status data.

[0087] After predicting the fault type and parameters using a pre-defined Generative Adversarial Network (GAN), a target fault can be artificially created or simulated based on the prediction results. This process typically involves specific operations or settings on the power supply chip to make it exhibit the predicted fault characteristics. For example, if the predicted fault type is voltage instability, and the fault parameters include the amplitude and frequency of voltage fluctuations, the input voltage or load of the power supply chip can be adjusted to simulate this voltage instability state. After successfully generating the target fault, second operating state data of the power supply chip needs to be collected. This data should reflect the performance and behavior of the power supply chip under fault conditions. Similar to the first operating state data, the second operating state data can include various parameters such as voltage, current, temperature, and power consumption. Comparing the second operating state data with the previously collected first operating state data is a crucial step in verifying the accuracy of fault prediction and understanding the fault performance of the power supply chip. By comparing, the differences in the operating state of the power supply chip under fault conditions and normal conditions can be analyzed, thereby verifying the accuracy of fault prediction. Specifically, various parameters under the two states can be compared, such as voltage fluctuations, current trends, and temperature ranges.

[0088] S130. When the target difference between the second target data and the corresponding first target data in the first operating state data is greater than a threshold, a preset recovery strategy is determined based on the second target data.

[0089] After acquiring the second operating state data, it is compared with the previously collected first operating state data. In particular, attention is paid to target data that significantly impacts the power chip's performance, such as specific voltage and current values. For each pair of corresponding target data (first target data and second target data), the chip testing platform calculates the difference between them. This difference reflects the degree of performance or state change of the power chip under two different states. The chip testing platform determines whether the calculated target difference exceeds a preset threshold. This threshold is set based on the power chip's normal operating range and possible fault modes, used to determine whether the power chip has experienced an anomaly or fault. If the target difference exceeds the threshold, the chip testing platform determines a preset recovery strategy based on the specific value and nature of the second target data. This recovery strategy is a predefined series of operational steps or instructions used to restore the power chip from an abnormal or faulty state to a normal state. The specific content of the recovery strategy depends on the type and severity of the power chip's fault. For example, if the fault is caused by voltage instability, the recovery strategy might include adjusting the power supply voltage or replacing the power module. If the fault is caused by overload, the recovery strategy might include reducing the load or increasing heat dissipation.

[0090] Optionally, determining the preset recovery strategy based on the second target data includes:

[0091] The similarity measurement method is used to calculate the similarity between the second target data and the data in the fault mode library to match the fault type, and the fault parameters are determined according to the fault type.

[0092] The matching value of the recovery strategy in the preset strategy database is calculated based on the fault type and the fault parameters. The matching value is obtained by weighting the recovery time, recovery success rate and percentage of the recovery strategy to reach the original state after recovery. The first recovery strategy is selected as the preset recovery strategy. The first recovery strategy and the second recovery strategy are any two recovery strategies in the preset strategy database. The matching value of the first recovery strategy is greater than the matching value of the second recovery strategy.

[0093] The system collects secondary operating state data from the power supply chip, paying particular attention to abnormal secondary target data. This secondary target data may include key parameters such as voltage, current, and temperature. Similarity metrics (e.g., Euclidean distance, cosine similarity) are used to calculate the similarity between the secondary target data and fault data stored in a fault mode library. The fault mode library is a pre-built database containing various known fault types and their corresponding feature data. The chip testing platform iterates through each fault type in the library, calculating the similarity between the secondary target data and the feature data of each fault type, and identifies the fault type with the highest similarity as the matching result. This process is essentially fault identification, helping the chip testing platform determine the type of fault the power supply chip is currently experiencing. After determining the fault type, the chip testing platform further determines specific fault parameters based on the characteristics of that fault type. These parameters may include the severity of the fault, the time of occurrence, and the location. Determining the fault parameters helps to more accurately describe the fault state of the power supply chip, providing a basis for subsequent recovery strategy selection. The chip testing platform then queries a pre-set strategy database, which stores various recovery strategies. Each recovery strategy targets a specific fault type and parameters, and includes key metrics such as recovery time, recovery success rate, and percentage of the system returning to its original state after recovery. The chip testing platform calculates the matching value for each recovery strategy in the preset strategy database based on the identified fault type and parameters. This matching value is obtained through a weighted calculation, where different weights are assigned to metrics such as recovery time, recovery success rate, and percentage of the system returning to its original state. The weights can be adjusted according to actual conditions to reflect the importance of different metrics in the recovery process. After calculating the matching values ​​for all recovery strategies, the chip testing platform selects the strategy with the highest matching value as the preset recovery strategy. During this process, if multiple recovery strategies have similar matching values, the chip testing platform can further filter them based on other factors (such as user preferences and historical experience).

[0094] By using a similarity metric to calculate the similarity between the second target data and the data in the fault mode library, the fault type encountered by the power chip can be accurately matched. This data-driven fault identification method is more accurate and efficient than traditional manual diagnosis, helping to quickly locate problems. After determining the fault type and fault parameters, the chip testing platform calculates the matching value of each recovery strategy in the preset strategy database and selects the recovery strategy with the highest matching value as the preset recovery strategy. This strategy matching mechanism ensures a high degree of matching between the selected strategy and the current fault situation, improving the pertinence and effectiveness of the recovery strategy. By comprehensively considering factors such as the recovery time, recovery success rate, and percentage of the original state after recovery, and by weighting these factors to obtain the matching value, the chip testing platform can select a recovery strategy that is both efficient and reliable. This helps to shorten the fault recovery time, improve the recovery success rate, and reduce downtime and losses caused by faults. Timely fault identification and accurate recovery strategy selection can quickly restore the power chip from the fault state to the normal state, thereby reducing the impact of the fault on the entire system. This helps to ensure the stable operation of the system and improve the system's reliability and availability. Because this process is automated based on data analysis and strategy matching, the chip testing platform can automatically identify faults and select recovery strategies based on real-time acquired data. This makes the chip testing platform more adaptable and flexible, able to adapt to fault conditions in different environments. Through accurate fault identification and efficient recovery strategy selection, the chip testing platform can reduce unnecessary maintenance and replacement costs. At the same time, the shortened fault recovery time also reduces downtime and production losses caused by faults, further reducing maintenance costs.

[0095] S140. When the power chip recovers to the preset state, the third operating state data of the power chip is collected, and the test result of the power chip is determined based on the comparison result of the third operating state data and the preset recovery strategy.

[0096] Once the power chip is restored to its preset state, the chip testing platform immediately begins collecting its third operating state data. This data includes key parameters of the power chip, such as voltage, current, temperature, and power consumption, as well as any specific parameters related to fault recovery. The accuracy and completeness of this data are crucial for determining subsequent test results. The preset state can be the original state before the injection of faults and noise, or a state with minor differences from the original state. For example, when the differences between the output current and output voltage and those in the original state are both within a preset range, it can be considered a restoration to the preset state. After collecting the third operating state data, the chip testing platform compares it with the previously executed preset recovery strategy. The comparison mainly includes:

[0097] Recovery parameter verification: Check whether the actual operating status parameters (third operating status data) of the power chip have reached the target value or range set in the preset recovery strategy. For example, if the preset recovery strategy requires the voltage of the power chip to be restored to a specific range, the chip test platform will check whether the actual voltage value meets this requirement;

[0098] Fault feature elimination: Compare the third operating status data to see if the characteristic data of the previous fault still exists. If the fault features have been completely eliminated, it means that the power chip has been successfully restored; if the fault features still exist, it means that the recovery strategy may not have been fully effective or there are other problems.

[0099] Based on the comparison results between the third operating state data and the preset recovery strategy, the chip testing platform determines the test results of the power chip. The test results can typically be categorized into the following types:

[0100] Recovery successful: If the third operating status data meets all the requirements of the preset recovery strategy and the fault characteristics have been completely eliminated, the chip test platform will determine that the power chip has been successfully recovered and give the corresponding test results;

[0101] Partial Recovery Successful: In some cases, although the power chip's performance has recovered somewhat, it has not fully reached the preset state or some fault characteristics still exist. In this situation, the chip testing platform may determine that the recovery was partially successful and provide corresponding suggestions or warnings.

[0102] Recovery Failure: If the third operating status data does not meet any requirements of the preset recovery strategy, or if the fault characteristics still exist or even worsen, the chip test platform will determine that the recovery has failed and give corresponding error messages or suggestions.

[0103] Depending on the test results, the chip testing platform may take different follow-up actions. For example:

[0104] For power chips that have been successfully restored, the chip testing platform can put them back into use and continue to monitor their operating status.

[0105] For power chips that have been partially recovered, the chip testing platform may try other recovery strategies or advise users to perform further checks and maintenance.

[0106] For power chips that fail to recover, the chip testing platform may mark them as faulty.

[0107] Optionally, determining the test result of the power chip based on the comparison result of the third operating state data and the preset recovery strategy includes:

[0108] Calculate the first difference between the first recovery time of the preset recovery strategy and the second recovery time in the third operating state data, and calculate the second difference between the first percentage of the preset recovery strategy that reaches the original state after recovery and the second percentage of the third operating state data that reaches the original state after recovery. Then, perform a weighted summation of the first difference and the second difference to obtain a comparison value.

[0109] When the comparison value is less than the threshold, the power chip is determined to be in a normal state;

[0110] When the comparison value is greater than or equal to the threshold, the power chip is determined to be in an abnormal state.

[0111] Two key parameters are extracted from the preset recovery strategy: the first recovery time and the first percentage of the original state after recovery. Simultaneously, two corresponding parameters are obtained from the third operating state data: the second recovery time and the second percentage of the original state after recovery. The chip testing platform calculates two differences:

[0112] First difference: The difference between the first recovery time of the preset recovery strategy and the second recovery time in the third running status data, representing the deviation between the actual recovery time and the expected recovery time;

[0113] The second difference: The difference between the first percentage of the recovery to the original state after the preset recovery strategy and the second percentage of the recovery to the original state after the third running state data, represents the deviation between the actual recovery effect and the expected recovery effect.

[0114] The chip testing platform performs a weighted sum of these two differences to obtain a comprehensive comparison value. The weighting can be adjusted based on actual conditions to reflect the importance of different parameters in evaluating the recovery effect. The chip testing platform then compares the calculated comparison value with a preset threshold to determine the test result of the power chip.

[0115] When the comparison value is less than the threshold, it indicates that the deviation between the actual recovery time and the expected recovery time is small, and the deviation between the actual recovery effect and the expected recovery effect is also within an acceptable range. Therefore, the chip testing platform will determine that the power chip is in a normal state.

[0116] When the comparison value is greater than or equal to the threshold, it indicates a significant deviation between the actual recovery time and the expected recovery time, or an excessive deviation between the actual recovery effect and the expected recovery effect. This may mean that the power chip has not fully recovered to the preset state, or that other problems have occurred during the recovery process. Therefore, the chip testing platform will determine that the power chip is in an abnormal state.

[0117] This embodiment also discloses a system for testing power supply chips. Figure 2This is a schematic diagram of a system for testing power supply chips disclosed in an embodiment of this application, as shown below. Figure 2 As shown, the system includes a prediction module 201, a comparison module 202, a recovery module 203, and an execution module 204, wherein:

[0118] Prediction module 201 is configured to collect first operating status data of the power chip and predict fault type and fault parameters based on the first operating status data through a preset generative adversarial network. The first operating status data includes operating temperature, load, output voltage and output current.

[0119] The comparison module 202 is configured to generate a target fault based on the fault type and fault parameters, collect the second operating status data of the power chip, and compare the second operating status data with the first operating status data.

[0120] Recovery module 203 is configured to determine a preset recovery strategy based on the second target data when the target difference between the second target data in the second running status data and the corresponding first target data in the first running status data is greater than a threshold.

[0121] The execution module 204 is configured to collect the third operating state data of the power chip when the power chip recovers to a preset state, and determine the test result of the power chip based on the comparison result of the third operating state data and the preset recovery strategy.

[0122] Optionally, the prediction module 201 is configured to:

[0123] Acquire historical operating status data, which includes normal status data and fault status data, and assign labels to the fault status data to represent fault type and fault parameters;

[0124] Construct a generator and a discriminator, and iteratively train the discriminator and the generator using the historical running state data until the preset function meets the requirements to construct the preset generative adversarial network;

[0125] The first operating status data is input into the preset generative adversarial network to predict fault status data.

[0126] Optionally, the prediction module 201 is configured to:

[0127] The parameters of the generator are fixed, and the discriminator is trained to distinguish between real running state data and generated running state data. The parameters of the discriminator are then updated through backpropagation and gradient descent.

[0128] The parameters of the discriminator are fixed, the generator is trained to construct and generate runtime data, and the parameters of the generator are updated through backpropagation and gradient descent.

[0129] The discriminator and the generator are trained alternately until a preset number of iterations or performance convergence is reached.

[0130] Optionally, the prediction module 201 is configured to:

[0131] A noise signal is generated based on the first operating state data using a preset noise model, and the noise signal is added to the input terminal of the power chip.

[0132] The third operating state data of the power chip is collected, and the fault type and fault parameters are predicted based on the third operating state data through the preset generative adversarial network.

[0133] Optionally, the prediction module 201 is configured to:

[0134] The load data in the first operating state data is mapped to the corresponding noise parameters through the preset noise model, and a noise signal is generated by combining the noise parameters with a random number generator. The noise parameters include the amplitude, frequency and phase of the noise.

[0135] Optionally, the recovery module 203 is configured to:

[0136] The similarity measurement method is used to calculate the similarity between the second target data and the data in the fault mode library to match the fault type, and the fault parameters are determined according to the fault type.

[0137] The matching value of the recovery strategy in the preset strategy database is calculated based on the fault type and the fault parameters. The matching value is obtained by weighting the recovery time, recovery success rate and percentage of the recovery strategy to reach the original state after recovery. The first recovery strategy is selected as the preset recovery strategy. The first recovery strategy and the second recovery strategy are any two recovery strategies in the preset strategy database. The matching value of the first recovery strategy is greater than the matching value of the second recovery strategy.

[0138] Optionally, the execution module 204 is configured to:

[0139] Calculate the first difference between the first recovery time of the preset recovery strategy and the second recovery time in the third operating state data, and calculate the second difference between the first percentage of the preset recovery strategy that reaches the original state after recovery and the second percentage of the third operating state data that reaches the original state after recovery. Then, perform a weighted summation of the first difference and the second difference to obtain a comparison value.

[0140] When the comparison value is less than the threshold, the power chip is determined to be in a normal state;

[0141] When the comparison value is greater than or equal to the threshold, the power chip is determined to be in an abnormal state.

[0142] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0143] This embodiment also discloses an electronic device, as shown in the reference. Figure 3 The electronic device may include: at least one processor 301, at least one communication bus 302, user interface 303, network interface 304, and at least one memory 305.

[0144] The communication bus 302 is used to enable communication between these components.

[0145] The user interface 303 may include a display screen and a camera. Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0146] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0147] The processor 301 may include one or more processing cores. The processor 301 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 305, and by calling data stored in memory 305. Optionally, the processor 301 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 301 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 301 and may be implemented as a separate chip.

[0148] The memory 305 may include random access memory (RAM) or read-only memory. Optionally, the memory 305 may include a non-transitory computer-readable storage medium. The memory 305 can be used to store instructions, programs, code, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 305 may also be at least one storage device located remotely from the aforementioned processor 301. Figure 3 As shown, the memory 305, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program for testing power supply chips.

[0149] exist Figure 3In the electronic device shown, the user interface 303 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 301 can be used to call the application program stored in the memory 305 for testing the power chip. When executed by one or more processors 301, the electronic device performs one or more methods as described in the above embodiments.

[0150] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0151] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0152] In the several embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some service interfaces; indirect couplings or communication connections between apparatuses or units may be electrical or other forms.

[0153] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0154] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0155] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 305 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 of the various embodiments of this application. The aforementioned memory 305 includes various media capable of storing program code, such as a USB flash drive, external hard drive, magnetic disk, or optical disk.

[0156] The foregoing description is merely an exemplary embodiment of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of other embodiments of this disclosure upon considering the specification and the disclosure of practical truth. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A method for testing a power supply chip, characterized in that, Applied to a chip testing platform, the method includes: The first operating status data of the power chip is collected, and the fault type and fault parameters are predicted based on the first operating status data through a preset generative adversarial network. The first operating status data includes operating temperature, load, output voltage and output current. A target fault is generated based on the fault type and fault parameters. The second operating status data of the power chip is collected, and the second operating status data is compared with the first operating status data. When the target difference between the second target data and the corresponding first target data in the first operating status data is greater than a threshold, a preset recovery strategy is determined based on the second target data. The second target data is voltage, current or temperature, and the first target data is the data corresponding to the second target data. When the power chip recovers to a preset state, the third operating state data of the power chip is collected, and the test result of the power chip is determined based on the comparison result of the third operating state data and the preset recovery strategy. The step of determining the preset recovery strategy based on the second target data includes: The similarity measurement method is used to calculate the similarity between the second target data and the data in the fault mode library to match the fault type, and the fault parameters are determined according to the fault type. Based on the fault type and the fault parameters, a matching value for the recovery strategy in the preset strategy database is calculated. This matching value is obtained by weighting the recovery time, recovery success rate, and percentage of the recovery strategy reaching the original state. A first recovery strategy is selected as the preset recovery strategy. The first and second recovery strategies are any two recovery strategies from the preset strategy database, and the matching value of the first recovery strategy is greater than the matching value of the second recovery strategy. The step of determining the test result of the power chip based on the comparison result of the third operating state data and the preset recovery strategy includes: Calculate the first difference between the first recovery time of the preset recovery strategy and the second recovery time in the third operating state data, and calculate the second difference between the first percentage of the preset recovery strategy that reaches the original state after recovery and the second percentage of the third operating state data that reaches the original state after recovery. Then, perform a weighted summation of the first difference and the second difference to obtain a comparison value. When the comparison value is less than the threshold, the power chip is determined to be in a normal state; When the comparison value is greater than or equal to the threshold, the power chip is determined to be in an abnormal state.

2. The method for testing a power supply chip according to claim 1, characterized in that, The step of predicting the fault type and fault parameters based on the first operating state data using a preset generative adversarial network includes: Acquire historical operating status data, which includes normal status data and fault status data, and assign labels to the fault status data to represent fault type and fault parameters; Construct a generator and a discriminator, and iteratively train the discriminator and the generator using the historical running state data until the preset function meets the requirements to construct the preset generative adversarial network; The first operating status data is input into the preset generative adversarial network to predict fault status data.

3. The method for testing a power supply chip according to claim 2, characterized in that, The step of iteratively training the discriminator and the generator using the historical running state data until the preset function meets the requirements to construct the preset generative adversarial network includes: The parameters of the generator are fixed, and the discriminator is trained to distinguish between real running state data and generated running state data. The parameters of the discriminator are then updated through backpropagation and gradient descent. The parameters of the discriminator are fixed, the generator is trained to construct and generate runtime data, and the parameters of the generator are updated through backpropagation and gradient descent. The discriminator and the generator are trained alternately until a preset number of iterations or performance convergence is reached.

4. The method for testing a power supply chip according to claim 1, characterized in that, The step of predicting the fault type and fault parameters based on the first operating state data using a preset generative adversarial network includes: A noise signal is generated based on the first operating state data using a preset noise model, and the noise signal is added to the input terminal of the power chip. The third operating state data of the power chip is collected, and the fault type and fault parameters are predicted based on the third operating state data through the preset generative adversarial network.

5. The method for testing a power supply chip according to claim 4, characterized in that, The step of generating a noise signal based on the first operating state data using a preset noise model includes: The load data in the first operating state data is mapped to the corresponding noise parameters through the preset noise model, and a noise signal is generated by combining the noise parameters with a random number generator. The noise parameters include the amplitude, frequency and phase of the noise.

6. A system for testing power supply chips, characterized in that, It includes a prediction module, a comparison module, a recovery module, and an execution module, among which: The prediction module is configured to collect first operating status data of the power chip and predict fault type and fault parameters based on the first operating status data through a preset generative adversarial network. The first operating status data includes operating temperature, load, output voltage and output current. The comparison module is configured to generate a target fault based on the fault type and fault parameters, collect the second operating status data of the power chip, and compare the second operating status data with the first operating status data. The recovery module is configured to determine a preset recovery strategy based on the second target data when the target difference between the second target data in the second operating status data and the corresponding first target data in the first operating status data is greater than a threshold. The second target data is voltage, current or temperature, and the first target data is the data corresponding to the second target data. The execution module is configured to collect third operating state data of the power chip when the power chip recovers to a preset state, and determine the test result of the power chip based on the comparison result of the third operating state data and the preset recovery strategy. The step of determining the preset recovery strategy based on the second target data includes: The similarity measurement method is used to calculate the similarity between the second target data and the data in the fault mode library to match the fault type, and the fault parameters are determined according to the fault type. Based on the fault type and the fault parameters, a matching value for the recovery strategy in the preset strategy database is calculated. This matching value is obtained by weighting the recovery time, recovery success rate, and percentage of the recovery strategy reaching the original state. A first recovery strategy is selected as the preset recovery strategy. The first and second recovery strategies are any two recovery strategies from the preset strategy database, and the matching value of the first recovery strategy is greater than the matching value of the second recovery strategy. The step of determining the test result of the power chip based on the comparison result of the third operating state data and the preset recovery strategy includes: Calculate the first difference between the first recovery time of the preset recovery strategy and the second recovery time in the third operating state data, and calculate the second difference between the first percentage of the preset recovery strategy that reaches the original state after recovery and the second percentage of the third operating state data that reaches the original state after recovery. Then, perform a weighted summation of the first difference and the second difference to obtain a comparison value. When the comparison value is less than the threshold, the power chip is determined to be in a normal state; When the comparison value is greater than or equal to the threshold, the power chip is determined to be in an abnormal state.

7. An electronic device, characterized in that, The device includes a processor, a memory, a user interface, and a network interface. The memory is used to store instructions. The user interface and the network interface are both used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1-5.

Images