Compensation and redundancy hybrid self-healing control method and device for integrated circuit power supply

Abstract

The invention discloses an integrated circuit power supply compensation and redundancy hybrid self-healing control method and device, and relates to the technical field of integrated circuits, and the method comprises the steps: collecting preset parameters of a to-be-evaluated power supply; a degradation feature extraction algorithm is adopted to analyze the preset parameters, and degradation feature vectors are generated; determining the anomaly type of the to-be-evaluated power supply according to the degradation feature vector; under the condition of transient disturbance of the to-be-evaluated power supply, calling a degradation perception compensation algorithm to determine a PWM duty ratio adjustment amount according to the degradation feature vector to generate a PWM duty ratio adjustment instruction, and triggering the super capacitor to discharge to recover the voltage of the to-be-evaluated power supply to a safety threshold value; under the condition that the voltage of the to-be-evaluated power supply is not recovered to a safety threshold value or the to-be-evaluated power supply has a permanent fault within a first preset duration, judging the fault type of the to-be-evaluated power supply according to the real-time preset parameter information and the degradation feature vector of the to-be-evaluated power supply; and matching the target standby module according to a preset fault type and a mapping rule between the standby modules for switching.

Description

Technical Field

[0001] This invention relates to the field of integrated circuit technology, and in particular to a method and apparatus for integrated circuit power supply compensation and redundancy hybrid self-healing control. Background Technology

[0002] In electronic component aging test benches, the secondary power supply serves as the core power supply unit, providing precise electrical stress to the device under test by real-time adjustment of its output voltage / current. Its reliability directly affects the effectiveness and safety of the aging test; therefore, effectively protecting the secondary power supply is a pressing issue.

[0003] Traditional secondary power supply protection mechanisms are ill-suited to the performance degradation that occurs under long-term high-temperature, full-load conditions, and lack the ability to classify and handle transient disturbances and permanent faults. Specifically: First, the use of a fixed threshold compensation strategy fails to consider degradation effects such as capacitor attenuation and power device parameter drift, leading to a significant decrease in voltage regulation accuracy towards the end of the power supply's lifespan. A fixed threshold compensation strategy typically involves presetting fixed upper and lower voltage or current limits in the control circuit, triggering protection when the output voltage exceeds the rated value by ±5%. Second, redundant switching relies on manually preset paths, failing to respond to dynamic changes in the power supply's degradation trajectory, and frequent accidental switching interrupts the aging process. Third, it fails to distinguish between microsecond-level load surges and permanent device failures, causing short-term disturbances to trigger incorrect switching and halt tests, or permanent faults to cause cascading damage due to insufficient compensation.

[0004] It is evident that existing technologies suffer from protection inaccuracies and fault misjudgments in the secondary power supply of aging test benches under degrading environments. Summary of the Invention

[0005] The purpose of this invention is to provide a method and apparatus for integrated circuit power supply compensation and redundancy hybrid self-healing control, which can solve the above-mentioned problems existing in the prior art.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a hybrid self-healing control method for integrated circuit power supply compensation and redundancy, wherein the method includes: collecting preset parameters of the power supply to be evaluated; The preset parameters are analyzed using a degradation feature extraction algorithm to generate a degradation feature vector; Based on the degradation feature vector, the anomaly type of the power supply to be evaluated is determined, wherein the anomaly type includes: transient disturbances and permanent faults; When the power supply to be evaluated experiences a transient disturbance, the degradation sensing compensation algorithm is invoked to determine the PWM duty cycle adjustment amount based on the degradation feature vector; A PWM duty cycle adjustment command is generated based on the PWM duty cycle adjustment amount, and the supercapacitor is triggered to discharge so that the power supply voltage to be evaluated is restored to a safe threshold. If the voltage of the power supply to be evaluated fails to recover to a safe threshold or the power supply to be evaluated suffers a permanent fault within a first preset time period, the fault type of the power supply to be evaluated is determined based on the real-time preset parameter information of the power supply to be evaluated and the degradation feature vector. Based on the pre-defined mapping rules between fault types and backup modules in the system, the target backup module is matched and switched over.

[0007] Optionally, the step of analyzing the preset parameters using a degradation feature extraction algorithm to generate a degradation feature vector includes: The preset parameters are input into the degradation feature extraction algorithm; wherein, the degradation feature extraction algorithm includes the recursive least squares method; The preset parameters are identified using the recursive least squares method to obtain the equivalent series resistance of the filter capacitor. Based on the initial on-resistance, junction temperature, and the relationship between junction temperature and resistance of the power transistor, the on-resistance offset of the power switch is dynamically calculated. A degradation feature vector is generated based on the equivalent series resistance of the filter capacitor and the offset of the on-resistance of the power switch transistor.

[0008] Optionally, the degradation sensing compensation algorithm includes a temperature compensation term and a capacitance compensation term; The temperature compensation term is determined based on the temperature compensation coefficient, the initial equivalent series resistance of the filter capacitor, the voltage change value, and the equivalent series resistance of the filter capacitor. The capacitance compensation term is determined based on the capacitance compensation coefficient, the initial capacitance value, the estimated capacitance value, and the voltage change value.

[0009] Optionally, the step of determining the anomaly type of the power source to be evaluated based on the degradation feature vector includes: The duration of the continuous voltage drop of the power supply to be detected is determined by the degradation feature vector. If the duration of the continuous drop does not exceed the second preset duration, it is determined that the power supply to be evaluated has experienced a transient disturbance. If the duration of the continuous drop exceeds the second preset duration, it is determined that the power supply to be evaluated has suffered a permanent failure.

[0010] Optionally, the step of determining the fault type of the power supply to be evaluated based on the real-time preset parameter information and the degradation feature vector includes: Based on the real-time preset parameter information of the power supply to be evaluated, the output voltage ripple, output current, output voltage, and junction temperature of the power supply to be tested are determined. The change in the equivalent series resistance of the filter capacitor is determined based on the degradation feature vector. If the increase in the output voltage ripple of the power supply under test is greater than a preset level, and the change in the equivalent series resistance of the filter capacitor is greater than a preset change threshold, a filter capacitor degradation fault is determined to have occurred. If the input current increases sharply, the output voltage drops sharply, and the junction temperature rises sharply, then a short-circuit breakdown fault of the power switch of the power supply under evaluation is determined to have occurred.

[0011] Optionally, the steps of matching the target backup module for switching, based on the preset mapping rules between fault types and backup modules in the system, include: In the event of a filter capacitor degradation fault in the power supply to be evaluated, switch to a capacitor branch with a low equivalent series resistance. In the event that the power supply under evaluation experiences a short-circuit breakdown fault in the power switch, the power switch with the highest health assessment value is switched to.

[0012] This invention also provides an integrated circuit power supply compensation and redundancy hybrid self-healing control device, wherein the device includes: The data acquisition module is used to acquire preset parameters of the power supply to be evaluated. The generation module is used to analyze the preset parameters using a degradation feature extraction algorithm to generate a degradation feature vector; An anomaly type determination module is used to determine the anomaly type of the power supply to be evaluated based on the degradation feature vector, wherein the anomaly type includes: transient disturbance and permanent fault; The first calculation module is used to call the degradation perception compensation algorithm to determine the PWM duty cycle adjustment amount based on the degradation feature vector when the power supply to be evaluated experiences a transient disturbance. The compensation module is used to generate a PWM duty cycle adjustment command based on the PWM duty cycle adjustment amount and trigger the supercapacitor to discharge so that the power supply voltage to be evaluated is restored to a safe threshold. The fault type determination module is used to determine the fault type of the power supply to be evaluated based on the real-time preset parameter information of the power supply to be evaluated and the degradation feature vector when the voltage of the power supply to be evaluated fails to recover to the safety threshold or the power supply to be evaluated is permanently faulty within a first preset time period. The switching module is used to match the target backup module and switch it according to the mapping rules between the preset fault types and backup modules in the system.

[0013] Optionally, the generation module includes: The first submodule is used to input the preset parameters into the degradation feature extraction algorithm; wherein, the degradation feature extraction algorithm includes the recursive least squares method; The second submodule is used to identify the preset parameters using the recursive least squares method to obtain the equivalent series resistance of the filter capacitor. The third submodule is used to dynamically calculate the on-resistance offset of the power switch based on the initial on-resistance, junction temperature, and junction temperature-resistance relationship model of the power transistor. The fourth submodule is used to generate a degradation feature vector based on the equivalent series resistance of the filter capacitor and the offset of the on-resistance of the power switch transistor.

[0014] Optionally, the degradation sensing compensation algorithm includes a temperature compensation term and a capacitance compensation term; The temperature compensation term is determined based on the temperature compensation coefficient, the initial equivalent series resistance of the filter capacitor, the voltage change value, and the equivalent series resistance of the filter capacitor. The capacitance compensation term is determined based on the capacitance compensation coefficient, the initial capacitance value, the estimated capacitance value, and the voltage change value.

[0015] Optionally, the exception type determination module is specifically used for: The duration of the continuous voltage drop of the power supply to be detected is determined by the degradation feature vector. If the duration of the continuous drop does not exceed the second preset duration, it is determined that the power supply to be evaluated has experienced a transient disturbance. If the duration of the continuous drop exceeds the second preset duration, it is determined that the power supply to be evaluated has suffered a permanent failure.

[0016] Optionally, the fault type determination module is specifically used for: Based on the real-time preset parameter information of the power supply to be evaluated, the output voltage ripple, output current, output voltage, and junction temperature of the power supply to be tested are determined. The change in the equivalent series resistance of the filter capacitor is determined based on the degradation feature vector. If the increase in the output voltage ripple of the power supply under test is greater than a preset level, and the change in the equivalent series resistance of the filter capacitor is greater than a preset change threshold, a filter capacitor degradation fault is determined to have occurred. If the input current increases sharply, the output voltage drops sharply, and the junction temperature rises sharply, then a short-circuit breakdown fault of the power switch of the power supply under evaluation is determined to have occurred.

[0017] Optionally, the switching module is specifically used for: In the event of a filter capacitor degradation fault in the power supply to be evaluated, switch to a capacitor branch with a low equivalent series resistance. In the event that the power supply under evaluation experiences a short-circuit breakdown fault in the power switch, the power switch with the highest health assessment value is switched to.

[0018] This invention also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other via the communication bus. The memory is used to store computer programs. When the processor executes the program stored in the memory, it implements any of the above-described integrated circuit power supply compensation and redundancy hybrid self-healing control methods.

[0019] The integrated circuit power supply compensation and redundancy hybrid self-healing control scheme disclosed in this invention collects preset parameters of the power supply to be evaluated; analyzes the preset parameters using a degradation feature extraction algorithm to generate a degradation feature vector; determines the anomaly type of the power supply to be evaluated based on the degradation feature vector; in the event of a transient disturbance in the power supply to be evaluated, calls a degradation sensing compensation algorithm to determine the PWM duty cycle adjustment amount based on the degradation feature vector; generates a PWM duty cycle adjustment command based on the PWM duty cycle adjustment amount and triggers supercapacitor discharge to restore the voltage of the power supply to be evaluated to a safe threshold; if the voltage of the power supply to be evaluated fails to restore to the safe threshold or if the power supply to be evaluated suffers a permanent fault within a first preset time period, determines the fault type of the power supply to be evaluated based on the real-time preset parameter information and degradation feature vector; and matches the target backup module for switching according to the mapping rules between preset fault types and backup modules in the system. In this scheme, after detecting a power supply anomaly, the first line of defense, namely adaptive compensation, is activated first to deal with transient disturbances; the second line of defense, namely redundancy switching, is activated only when compensation is ineffective or a permanent fault is confirmed. This dual-defense collaborative mechanism not only minimizes test interruptions caused by erroneous switching, but also ensures rapid functional recovery of the power system in the event of a real fault by quickly isolating the faulty module and switching to a healthy backup module. This achieves full-process power system autonomy, from "perceived degradation" to "transient suppression" and then to "permanent recovery." This solution effectively addresses the problems of protection inaccuracies and fault misjudgments in existing solutions. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the steps of a hybrid self-healing control method for integrated circuit power supply compensation and redundancy according to an embodiment of this application. Figure 2 This is a flowchart illustrating the steps of a hybrid self-healing control method for integrated circuit power supply compensation and redundancy according to an embodiment of this application. Figure 3This is a flowchart illustrating the steps of an integrated circuit power supply performance degradation prediction method according to an embodiment of this application; Figure 4 This is a flowchart of a multi-channel secondary power supply health assessment method for an integrated circuit aging bench, provided in one embodiment of this application; Figure 5 This is a structural block diagram illustrating an integrated circuit power supply compensation and redundancy hybrid self-healing control device according to an embodiment of this application. Detailed Implementation

[0021] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0022] The integrated circuit power supply compensation and redundancy hybrid self-healing control method provided in this application will be described in detail below with reference to the accompanying drawings, through specific embodiments and application scenarios.

[0023] As attached Figure 1 As shown, the integrated circuit power supply compensation and redundancy hybrid self-healing control method of this application embodiment includes the following steps: Step 101: Collect preset data of the power source to be evaluated.

[0024] The integrated circuit power supply compensation and redundancy hybrid self-healing control method provided in this application can be applied to electronic devices. The electronic devices are equipped with an integrated circuit power supply compensation and redundancy hybrid self-healing control computer program. When the computer program is executed by the processor, it implements the integrated circuit power supply compensation and redundancy hybrid self-healing control method of the embodiments of this application.

[0025] During the operation of the power supply system, high-precision sensors collect key parameters such as output voltage, current, heat dissipation temperature, and load power in real time. A degradation feature extraction algorithm is used to identify key degradation characteristic parameters online, such as the equivalent series resistance and capacitance of the output filter capacitor, and the on-resistance of the MOSFET (power switch). These parameters are constructed into a degradation feature vector and embedded into a dynamic adaptive compensation model. The collected key power supply parameters reflect the operating status and health of the power supply system.

[0026] Step 102: Analyze the preset parameters using a degradation feature extraction algorithm to generate a degradation feature vector.

[0027] In one optional embodiment, the method of analyzing preset parameters using a degradation feature extraction algorithm to generate a degradation feature vector may include the following sub-steps: Sub-step 1: Input the preset parameters into the degradation feature extraction algorithm; Among them, the degradation feature extraction algorithm includes the recursive least squares method; Sub-step 2: Identify the preset parameters using the recursive least squares method to obtain the equivalent series resistance of the filter capacitor.

[0028] Sub-step 3: Based on the initial on-resistance, junction temperature, and the relationship model between junction temperature and resistance of the power transistor, dynamically calculate the on-resistance offset of the power switch.

[0029] Sub-step 4: Generate a degradation feature vector based on the equivalent series resistance of the filter capacitor and the offset of the on-resistance of the power switch transistor.

[0030] In practical implementation, a real-time monitoring module can continuously collect key parameters of the power supply under evaluation, such as the output voltage, current, and heatsink temperature of a secondary power supply. These key parameters serve as raw data input to the Degradation Feature Extraction (DE-PE) algorithm, which uses recursive least squares to identify the equivalent series resistance (ESR) and capacitance (C) of the output filter capacitor online. This algorithm utilizes the inherent high-frequency ripple current and voltage of the power supply's PWM (Pulse Width Modulation) switching action as excitation signals, eliminating the need for additional test signals and enabling parameter identification under normal power supply operation. Its calculation model is as follows:

[0031] Among them, U k Let I be the voltage value of the kth sample. k The current value is the value of the kth sample. The average voltage. This is the average current.

[0032] Simultaneously based on the junction temperature-resistance relationship model Dynamically calculate the MOSFET on-resistance offset, where, The temperature coefficient of the material. T is the initial on-resistance of the device at 25°C. j The calculated junction temperature. The final output is the degradation eigenvector [ΔESR, ΔR]. ds(on) (T)] provides a basis for perceiving the degraded environment for subsequent decision-making.

[0033] Step 103: Determine the anomaly type of the power source to be evaluated based on the degradation feature vector.

[0034] The anomaly types include transient disturbances and permanent faults.

[0035] In one alternative embodiment, the anomaly type of the power source to be evaluated can be determined based on the degradation feature vector as follows: The duration of voltage drop of the power supply under test is determined by the degradation feature vector; if the duration of voltage drop does not exceed the second preset duration, it is determined that the power supply under test has experienced a transient disturbance; if the duration of voltage drop exceeds the second preset duration, it is determined that the power supply under test has experienced a permanent fault.

[0036] For example, the second preset duration can be set to 100μs, 150μs or 200μs, etc. The specific value of the second preset duration can be flexibly set by those skilled in the art, and no specific limitation is made in this embodiment.

[0037] When the anomaly type is a transient disturbance, the first line of defense, namely adaptive compensation, is activated to deal with the transient disturbance. When the first line of defense compensation is ineffective or is confirmed as a permanent fault, the second line of defense, namely redundancy switching, is activated. Steps 104 to 105 are the execution process of the first line of defense, namely adaptive compensation, and steps 106 to 107 are the execution process of the second line of defense, namely redundancy switching.

[0038] Step 104: When a transient disturbance occurs in the power supply to be evaluated, the degradation perception compensation algorithm is invoked to determine the PWM duty cycle adjustment amount based on the degradation feature vector.

[0039] Upon determining that a transient disturbance has occurred in the power supply to be evaluated, the degradation sensing compensation algorithm, i.e., DPAC, is immediately activated. In an optional embodiment, the degradation sensing compensation algorithm includes a temperature compensation term and a capacitance compensation term; the temperature compensation term is determined based on a temperature compensation coefficient, the initial equivalent series resistance of the filter capacitor, the voltage change value, and the equivalent series resistance of the filter capacitor; the capacitance compensation term is determined based on a capacitance compensation coefficient, the initial capacitance value, the estimated capacitance value, and the voltage change value.

[0040] The degradation-aware compensation algorithm embeds the degradation feature vector into a dynamic compensation model. The model expression of this algorithm can be expressed as follows:

[0041] in, K represents the PWM duty cycle adjustment. p K i R is the compensation coefficient. ds0 C0 is the initial parameter, C est This is an estimated capacitance value. The algorithm model is characterized by the introduction of a degradation parameter for dynamic correction, which improves compensation accuracy and environmental adaptability.

[0042] Step 105: Generate a PWM duty cycle adjustment command based on the PWM duty cycle adjustment amount, and trigger the supercapacitor to discharge so that the power supply voltage to be evaluated is restored to the safe threshold.

[0043] The algorithm model dynamically generates PWM duty cycle adjustment commands based on the real-time system state and degradation level. In actual implementation, adaptive compensation can be achieved by executing the PWM duty cycle adjustment commands through a compensator. The parameters of the compensator can be optimized online using the gradient descent method to minimize the steady-state error of the output voltage. This ensures that even with component performance degradation, the system can still quickly and accurately suppress transient disturbances such as microsecond-level voltage drops, thereby avoiding unnecessary redundant switching triggered by short-term disturbances.

[0044] Step 106: If the voltage of the power supply to be evaluated fails to recover to the safety threshold or the power supply to be evaluated suffers a permanent fault within the first preset time period, the fault type of the power supply to be evaluated is determined based on the real-time preset parameter information and degradation feature vector of the power supply to be evaluated.

[0045] If the power supply voltage to be evaluated fails to recover to the safety threshold within the first preset time period, it indicates that the first line of defense, i.e., adaptive compensation, has failed. Both the first preset time period and the safety threshold can be flexibly set by those skilled in the art, and this embodiment does not impose specific limitations on them. For example, the safety threshold can be set to >92% of the rated value.

[0046] In one optional embodiment, determining the fault type of the power supply to be evaluated based on its real-time preset parameter information and degradation feature vector may include the following sub-steps: Sub-step 1: Based on the real-time preset parameter information of the power supply to be evaluated, determine the output voltage ripple, output current, output voltage, and junction temperature of the power supply to be tested.

[0047] Sub-step 2: Determine the change in the equivalent series resistance of the filter capacitor based on the degradation feature vector.

[0048] Sub-step 3: If the increase in output voltage ripple of the power supply under test is greater than a preset level, and the change in the equivalent series resistance of the filter capacitor is greater than a preset change threshold, then a filter capacitor degradation fault is determined to have occurred.

[0049] The preset change threshold can be set to 30%, 35%, or 25%, etc., and this application embodiment does not impose specific restrictions on it.

[0050] Sub-step 4: If the input current increases sharply, the output voltage drops sharply, and the junction temperature rises sharply, then it is determined that a short-circuit breakdown fault has occurred in the power switch of the power supply to be evaluated.

[0051] Step 107: Based on the mapping rules between the preset fault types and backup modules in the system, match the target backup module and switch it.

[0052] In an optional embodiment, the method of matching and switching the target backup module according to the mapping rules between the preset fault types and backup modules in the system can be as follows: if the power supply to be evaluated experiences a filter capacitor degradation fault, switch to the capacitor branch with a low equivalent series resistance; if the power supply to be evaluated experiences a power switch tube short circuit breakdown fault, switch to the power switch tube with the highest health assessment value.

[0053] In the event of a permanent failure in the power supply to be evaluated, the device that has suffered the permanent failure can be further identified. Generally, a permanent failure of a device means that the device has been broken down. In this case, the corresponding broken down device with the highest health assessment value can be selected from the backup module for switching.

[0054] The health assessment value of the backup module can be calculated through periodic self-checks, historical operating data, and monitoring of key parameters to determine its health status score. The algorithm automatically outputs an operation sequence to complete power topology reconfiguration within a certain time. Finally, the system returns the monitoring status, forming a closed-loop autonomous system, achieving full-process self-healing from "degradation perception → transient suppression → permanent recovery".

[0055] This invention discloses a hybrid self-healing control method for integrated circuit power supply compensation and redundancy. The method collects preset parameters of the power supply to be evaluated; analyzes the preset parameters using a degradation feature extraction algorithm to generate a degradation feature vector; determines the anomaly type of the power supply to be evaluated based on the degradation feature vector; in the event of a transient disturbance in the power supply to be evaluated, calls a degradation sensing compensation algorithm to determine the PWM duty cycle adjustment amount based on the degradation feature vector; generates a PWM duty cycle adjustment command based on the PWM duty cycle adjustment amount and triggers supercapacitor discharge to restore the voltage of the power supply to be evaluated to a safe threshold; if the voltage of the power supply to be evaluated fails to recover to the safe threshold within a first preset time period or if the power supply to be evaluated suffers a permanent fault, determines the fault type of the power supply to be evaluated based on the real-time preset parameter information and degradation feature vector; and matches a target backup module for switching according to the mapping rules between preset fault types and backup modules in the system. The method provided in this application, upon detecting a power supply anomaly, prioritizes the activation of the first line of defense, adaptive compensation, to address transient disturbances; only when compensation is ineffective or a permanent fault is confirmed is the second line of defense, redundant switching, activated. This dual-defense collaborative mechanism not only minimizes test interruptions caused by erroneous switching, but also ensures rapid functional recovery of the power system in the event of a real fault by quickly isolating the faulty module and switching to a healthy backup module. This achieves full-process power system autonomy, from "perceived degradation" to "transient suppression" and then to "permanent recovery." This solution effectively addresses the problems of protection inaccuracies and fault misjudgments in existing solutions.

[0056] The following is combined Figure 2This application describes the integrated circuit power supply compensation and redundancy hybrid self-healing control method provided in this application.

[0057] The integrated circuit power supply compensation and redundancy hybrid self-healing control method provided in this specific example constructs a two-level collaborative decision-making system for degradation awareness: First, the system collects key parameters such as output voltage, current, temperature, and load power in real time. Then, using a degradation feature extraction algorithm, it identifies key degradation characteristic parameters online, including the equivalent series resistance and capacitance of the output filter capacitor, and the on-resistance of the MOSFET. These key degradation characteristic parameters are constructed into a degradation feature vector and embedded into a dynamic adaptive compensation model, which incorporates a degradation-aware compensation algorithm.

[0058] This adaptive compensation model dynamically generates PWM duty cycle adjustment commands based on the real-time system state and degradation level. The compensator executes these PWM duty cycle adjustment commands for adaptive compensation. Its core advantage lies in the fact that the compensator parameters can be optimized online using gradient descent to minimize the steady-state error of the output voltage. This ensures that even with component performance degradation, the system can still quickly and accurately suppress transient disturbances, thereby avoiding unnecessary redundant switching triggered by short-term disturbances. Secondly, the degradation-aware two-level collaborative decision-making system also constructs an intelligent switching path library based on the mapping between fault types and redundant module states. When the adaptive compensation of the first line of defense fails, or the system is determined to have a permanent fault (such as device breakdown), this line of defense will be activated. Based on the aforementioned degradation characteristics and real-time fault information, the system accurately determines the fault type, such as capacitor degradation, MOSFET short circuit, etc., and automatically matches the backup module with the highest health from the redundant modules according to the preset mapping rules, generating the optimal switching operation sequence to complete the rapid and lossless reconstruction of the power supply topology. The degradation-aware two-tier collaborative decision-making system contains sophisticated hierarchical triggering logic: upon detecting an anomaly, the first line of defense, adaptive compensation, is activated to supplement transient disturbances; when compensation is ineffective or a permanent fault is confirmed, the second line of defense, redundancy switching, is activated. This collaborative mechanism not only minimizes test interruptions caused by erroneous switching but also ensures rapid system recovery at the functional level by quickly isolating the faulty module and switching to a healthy backup module when a real fault occurs, achieving full-process autonomy from "perceiving degradation" to "transient suppression" and then to "permanent recovery."

[0059] This specific example illustrates the self-healing control method for integrated circuit power supply analysis, using a secondary power supply as an example. The process may include the following: S1: Real-time monitoring.

[0060] The real-time monitoring module continuously collects key parameters such as the output voltage, current, and heat sink temperature of the secondary power supply.

[0061] S2: The key parameters obtained from real-time monitoring are used to input the degradation feature extraction algorithm DE-PE.

[0062] The key parameters obtained from real-time monitoring are used as raw data input to DE-PE. The algorithm identifies the equivalent series resistance (ESR) and capacitance (C) of the output filter capacitor online using the recursive least squares method. This algorithm utilizes the inherent high-frequency ripple current and voltage of the secondary power supply's PWM switching operation as the excitation signal, eliminating the need for additional test signals and enabling parameter identification under normal operating conditions of the secondary power supply. Its calculation model is as follows:

[0063] Among them, U k Let I be the voltage value of the kth sample. k The current value is the value of the kth sample. The average voltage. This is the average current.

[0064] Simultaneously based on the junction temperature-resistance relationship model Dynamically calculate the MOSFET on-resistance offset (where R0 is the temperature coefficient of the material, R0 is the initial on-resistance of the device at 25°C, and T0 is the temperature coefficient of the material. j (For the calculated junction temperature). Output the degradation eigenvector [Δ , Δ This provides a basis for perceiving the degraded environment in subsequent decision-making.

[0065] S3: Fault detection.

[0066] The fault detection module identifies whether the secondary power supply is experiencing a transient disturbance or a permanent fault based on the degradation feature vector.

[0067] S4: When the fault detection module detects a transient disturbance, it activates the first line of defense: the Degradation Perception Compensation Algorithm (DPAC) for adaptive compensation.

[0068] When the fault detection module detects a transient disturbance, such as a voltage drop lasting less than 100 μs, it immediately activates the degradation-aware compensation algorithm DPAC. This degradation-aware compensation algorithm embeds the degradation feature vector into the dynamic compensation model:

[0069] in, K represents the PWM duty cycle adjustment. p K i R is the compensation coefficient. ds0 C0 is the initial parameter, Cest This is an estimated capacitance value. The characteristic of this degradation-aware compensation algorithm lies in the introduction of degradation parameters for dynamic correction, which improves compensation accuracy and environmental adaptability.

[0070] The degradation-aware compensation algorithm fine-tunes the coefficients of the digital compensator online using the gradient descent method. , The optimization process aims to minimize the steady-state error of the output voltage and is constrained by system loop stability constraints derived from the degenerate eigenvector, ensuring that the power supply system remains stable after the compensation parameters are adjusted. A PWM duty cycle adjustment command (e.g., +12.5%) is generated and the supercapacitor is triggered to discharge, restoring the voltage to a safe threshold (e.g., >92% of the rated value).

[0071] The essence of gradient descent is to establish an optimization problem with controller parameters as independent variables and system performance indicators as dependent variables. The specific steps are as follows: 1) Define the loss function: Quantify the control objective into a mathematical indicator and construct the mean squared error loss function:

[0072] in, The value of the loss function. The size of the sliding sampling window. This is the output voltage reference value. For the first The actual output voltage value at each sampling time.

[0073] 2) Calculate the gradient: Calculate the loss function Regarding the parameters to be optimized , Partial derivatives:

[0074] in , This gradient represents the sensitivity of the system output to the controller parameters. In practical systems, this gradient value can be estimated online using the perturbation method: for , Apply a slight perturbation and observe. The change in is used to approximate the gradient.

[0075] 3) Parameter update: Along the inverse direction of the gradient (i.e., the direction of the fastest descent of the loss function), new controller n parameters:

[0076] Where n represents the number of iterations. The learning rate is used to control the update step size.

[0077] 4) Iteration and Convergence: Repeat the gradient calculation and parameter update steps to form an optimization loop until the loss function is reached. The output voltage error converges to near the minimum value, or meets the preset performance requirements.

[0078] This gradient descent-based online optimization process can run continuously and automatically, and can track the drift of the object model caused by component aging, load characteristic changes and ambient temperature fluctuations in real time, so that the controller parameters are always dynamically maintained near the optimal value, thereby significantly improving the adaptive compensation accuracy and robustness of the system throughout its entire life cycle.

[0079] S5: Determine if the voltage has returned to the safe threshold; if yes, return to execute S1; if no, execute S6.

[0080] If the voltage recovers to the safe threshold, the adaptive compensation is considered successful. If the compensation is successful, the system maintains its operating state and returns to the real-time monitoring closed loop, then returns to execute S1. If the voltage does not recover to the safe threshold, the adaptive compensation is considered to have failed.

[0081] S6: Activate the second line of defense: Path matching algorithm FPM.

[0082] The fault type determination incorporates degradation characteristics. For example, if a significant increase in output voltage ripple is detected, and the DE-PE algorithm simultaneously identifies ΔESR far exceeding the threshold, it is determined to be a filter capacitor degradation fault; if a sharp increase in input current is detected, accompanied by a drop in output voltage and a sharp increase in junction temperature, it is determined to be a power switching transistor (MOSFET) short-circuit breakdown fault.

[0083] After the fault type is determined, the process transitions to the Path Matching (FPM) algorithm. This algorithm establishes a mapping rule between fault types and switching paths: The mapping rules can be set as follows: if ΔESR > 30%, switch to the low ESR capacitor branch; if the fault type is MOSFET short-circuit breakdown failure, switch to the standby module with the highest health assessment value.

[0084] The health assessment value of the backup module can be calculated through periodic self-checks, historical operating data, and monitoring of key parameters to determine its health status score. The algorithm automatically outputs an operation sequence to complete power topology reconfiguration within a certain time. Finally, the system returns the monitoring status, forming a closed-loop autonomous system, achieving full-process self-healing from "degradation perception → transient suppression → permanent recovery".

[0085] This application also provides a method for predicting the performance degradation of an integrated circuit power supply. When the degradation state of the power supply is predicted to reach the repair standard, the above-described embodiment is executed to perform integrated circuit power supply compensation and redundancy hybrid self-healing control.

[0086] As attached Figure 3 As shown, the integrated circuit power supply performance degradation prediction method of this application embodiment includes the following steps: Step 301: Smooth and denoise the first health index data of the collected sample power sources using the Kalman filter algorithm to obtain the historical health index dataset.

[0087] The integrated circuit power supply performance degradation prediction method provided in this application can be applied to electronic devices. The electronic devices are equipped with an integrated circuit power supply performance degradation prediction computer program. When the computer program is executed by the processor, it implements the integrated circuit power supply performance degradation prediction method of this application.

[0088] In this embodiment, the first health index data of the sample power source is used as the basic data for the supervised learning dataset. In actual implementation, the health index data of the sample power source throughout its entire life cycle can be collected, or only the health index data of the latter half of the life cycle can be collected. This embodiment does not impose specific limitations on this. The collected first health index data of the sample power source can be regarded as the original health index (HI) data sequence. To eliminate the influence of sensor noise and environmental interference on the original health index (HI) data sequence and improve the input quality and prediction accuracy of the subsequent prediction model, this scheme first uses a Kalman filter to smooth and denoise the real-time collected original HI data.

[0089] In the specific implementation process, the true health index of the power supply circuit can be regarded as a system state that changes over time, while the data collected by the sensor is a noisy observation. Using a one-dimensional Kalman filter, at each time step, the current state is first "predicted" based on the optimal estimate of the previous time step, and then the current "observation" is used to "correct" this prediction, thereby obtaining the optimal estimate for the current time step.

[0090] In an optional embodiment, the method of smoothing and denoising the first health index data of the collected sample power supply using the Kalman filter algorithm to obtain the historical health index dataset may include the following sub-steps: Sub-step 1: For each time step in the first health index data of the collected sample power supply, determine the measured health index value and the predicted health index value at the current time. Sub-step 2: Based on the current measured health index value and the current predicted health index value, and using the Kalman gain, determine the optimal estimated value of the health index at the current time; The optimal estimate of the health index at the current moment includes: the mean of the optimal estimate of the health index at the current moment, and the uncertainty of the optimal estimate of the health index at the current moment; the optimal estimate of the health index at the current moment is used as the predicted value of the health index at the next moment.

[0091] Sub-step 3: Use the optimal estimate of the health index corresponding to each time step as the historical health index dataset.

[0092] One specific way to generate a historical health index dataset is as follows: First, the parameters are defined as shown in the table below:

[0093] Secondly, the goal is to find the optimal estimate of the health index at the current moment. First, we need to obtain the current health index measurement value. and the predicted value of the health index at the current moment .

[0094] In one-dimensional Kalman filtering, the formulas for calculating the mean and variance of the optimal estimate of the health index are as follows:

[0095]

[0096] The health index after Kalman filtering at time t is ,in , which is the Kalman gain.

[0097] Among the parameters used in the above calculations, the predicted health index value at the current moment is taken as the optimal estimate of the health index at the previous moment, and the formula is as follows: The variance of the current health index prediction is the sum of the variance of the best estimate from the previous time step and a fixed prediction variance, as shown in the formula: .

[0098] By iterating repeatedly using the above formula, the mean of the optimal estimate of the health index at each time step can be obtained. The curve that changes over time is the health index after Kalman filtering.

[0099] Step 302: Use a sliding window approach to cut the historical health index dataset into multiple first data sequences of preset window lengths, and determine the actual remaining lifespan label corresponding to each first data sequence.

[0100] The preset window length can be flexibly adjusted by those skilled in the art. In this embodiment, no specific limitation is made. For example, the preset window length can be set to include 5 time steps, 10 time steps, etc.

[0101] In this embodiment, a particle swarm optimization-based LSTM-Attention model is used for power remaining lifetime prediction. After obtaining the historical health index dataset after Kalman filtering in step 301, it is necessary to determine the initial parameters of the particle swarm optimization-based LSTM-Attention model and generate a supervised learning dataset based on the historical health index dataset. When generating the supervised learning dataset, it is necessary to use a sliding window approach to truncate the historical health index dataset into multiple first data sequences of preset window lengths and determine the true remaining lifetime label corresponding to each first data sequence.

[0102] To overcome the problems of error accumulation and low computational efficiency in traditional rolling forecasting methods, this invention proposes a more advanced end-to-end remaining useful life (HI) prediction framework. The core idea of ​​this framework is to redefine the model's prediction objective: instead of predicting the health index at the next time step, it directly predicts the time from the current moment until the HI value first reaches the preset failure threshold. The required time span. Therefore, the first step is to generate labels for the data sequence, which includes the following sub-steps: Sub-step 3021: On the complete historical HI dataset, extract an input sequence of length L using a sliding window. , where t is the starting point of the window.

[0103]

[0104] Sub-step 3022: For each extracted input sequence, i.e., the first data sequence Determine the true remaining useful life label corresponding to the first number sequence. .

[0105] In one optional embodiment, the method for determining the true remaining useful life tag corresponding to each first data sequence may specifically include the following sub-steps: S1: Determine the end time point of the window corresponding to the first data sequence; Determine the time step corresponding to the end time point of the window. Each point in time corresponds to a time step.

[0106] S2: Starting from the end time point, scan the supervised learning dataset in the future direction to obtain the first time point that meets the preset conditions; Preset conditions can be set to , This is a preset failure threshold.

[0107] from Starting from this moment, scan the historical HI dataset towards the future to find the first one that satisfies the condition. The time point corresponding to the time step That is, the first time point.

[0108] S3: Calculate the time step difference between the first time point and the end time point; The time step difference can be calculated using the following formula.

[0109] S4: Generate the actual remaining useful life label corresponding to the first data sequence based on the time step difference and the time difference between time steps.

[0110] The first data sequence can be calculated using the following formula. RUL time stamp: Where △t represents the time for one step.

[0111] Step 303: For each first data sequence, store the first data sequence and its corresponding real remaining lifespan label to generate a supervised learning dataset.

[0112] All generated The sample pairs are combined to form the final supervised learning dataset used for model training.

[0113] Step 304: Determine the initial parameters of the LSTM-Attention prediction model based on the particle swarm optimization algorithm and the supervised learning dataset.

[0114] After obtaining a smoothed health index sequence through Kalman filtering, in order to accurately predict the remaining effective lifetime of the main power supply and determine the optimal switching time, this application proposes a Long Short-Term Memory (LSTM) prediction model based on Particle Swarm Optimization (PSO).

[0115] This solution addresses key parameters of traditional LSTM models, such as the time window length L and the number of hidden layer neurons. Learning rate For problems that are difficult to determine, the particle swarm optimization algorithm, which has global optimization capabilities and fast convergence characteristics, is used to adaptively optimize the model hyperparameters to determine the initial parameter values ​​of the model, in order to train a model with better performance.

[0116] In one alternative embodiment, determining the initial parameters of the LSTM-Attention prediction model based on the particle swarm optimization algorithm and the supervised learning dataset may include the following sub-steps: S1: Set up the particle swarm and randomly initialize the position and velocity of each particle in the particle swarm within a preset range. Each particle corresponds to a combination of parameters.

[0117] In practical implementation, the particle swarm size can be set to... (For example =20), maximum number of iterations is Each particle Represents a set of potential parameter combinations, whose position vector is denoted as The velocity vector is represented as The position and velocity of each particle are randomly initialized within a preset range.

[0118] S2: For each particle, the particle is used as a parameter of the LSTM-Attention prediction model. The LSTM-Attention prediction model is iteratively trained using a supervised learning dataset. The corresponding prediction error is then calculated. Specifically, the root mean square error (RMSE) on the supervised learning dataset can be used as the fitness function. For each parameter combination represented by a particle, a corresponding LSTM-Attention model is constructed, trained, and its prediction error is calculated.

[0119] in To determine the number of samples in the validation set, For the actual remaining lifespan, These are the model's predicted values. The smaller the fitness value, the better the parameter combination.

[0120] S3: During the iterative training of the LSTM-Attention prediction model using a supervised learning dataset, record the historical best position and the global historical best position of the particle in each iteration. S4: Update the particle's velocity and position based on the particle's corresponding historical best position and global historical best position; In each iteration, the best position in the history of the individual particle is recorded. and the global historical best position In actual implementation, the velocity and position of each particle can be updated according to the following formula:

[0121]

[0122] in, For inertial weights, and The learning factor (usually 2.0) , It is a random number between [0,1].

[0123] S5: After completing the iterative training of each particle in the particle swarm, determine the parameter combination of the particle corresponding to the global optimal position as the initial parameters of the LSTM-Attention prediction model.

[0124] When the maximum number of iterations is reached If the fitness value is less than a preset threshold, the iteration of particles in the particle swarm is considered complete. The global optimal position is output. Corresponding parameter combinations As the initial parameters for the final LSTM-Attention model.

[0125] Step 305: Optimize and iteratively train the initial parameters of the LSTM-Attention prediction model based on the supervised learning dataset to obtain the target LSTM-Attention prediction model.

[0126] To accurately capture the complex nonlinear dynamics and long-term dependencies in health index data sequences, this application constructs a deep prediction model, namely the LSTM-Attention prediction model, which integrates a Long Short-Term Memory network and an Attention Mechanism. This LSTM-Attention prediction model includes an input layer, an LSTM encoding layer, an attention layer, and a fully connected output layer.

[0127] The specific process of optimizing and iteratively training the initial parameters of the LSTM-Attention prediction model based on the supervised learning dataset can be referenced from existing methods of training models based on training samples, and no specific restrictions are imposed here.

[0128] Step 306: Collect the second health index data of the power supply to be monitored within a preset window length.

[0129] The data collected in this step is the latest data at a preset window length, and the purpose is to predict the latest health status of the power supply under test.

[0130] The preset window length is the same as the preset window length set when the historical health index dataset is truncated into a first data sequence of multiple preset window lengths using a sliding window method.

[0131] Step 307: Based on the target LSTM-Attention prediction model, predict the second health index data to obtain the remaining service life of the power supply to be monitored.

[0132] In one optional embodiment, the method of predicting the remaining lifespan of the power supply to be monitored based on the target LSTM-Attention prediction model may include the following sub-steps: Sub-step 1: Generate a second data sequence based on the second health index data and input it into the input layer.

[0133] In practical implementation, the health index data can also be filtered and smoothed using the Kalman filter algorithm.

[0134] The HI data sequence is smoothed using Kalman filtering and then segmented by a sliding window (window length L). For example, at time t, the input is... The output is a three-dimensional tensor with dimensions [batch size, time step, feature dimension], which will be used as the input to the LSTM encoding layer.

[0135] Sub-step 2: The input layer sequentially inputs the second data sequence into the LSTM encoding layer according to the time steps.

[0136] Sub-step 3: The LSTM encoding layer generates a sequence of hidden state vectors containing all time steps for the input second data sequence.

[0137] The hidden state vector sequence includes: batch size, time step, and number of LSTM hidden units.

[0138] Input sequence That is, the second data sequence is fed into the LSTM layer step by step. At each time step... The LSTM unit will adjust according to the current input. and the hidden state of the previous moment With cell state Calculate and update the hidden state at the current time step. and cell state This process encodes and transmits historical information. Among other things, The range is from 1 to L.

[0139] A sequence of hidden state vectors containing all time steps, denoted as... The hidden state vector sequence has dimensions of [batch size, time step L, number of LSTM hidden units]. The hidden state vector sequence H completely preserves the contextual information of the original input sequence, i.e., the second data sequence, at different time points, and is the basis for information filtering in subsequent Attention layers.

[0140] Sub-step 4: The attention layer dynamically weights the hidden state vector sequence to generate a context vector.

[0141] The context vector contains key time step information.

[0142] The attention layer in the scheme provided in this application simulates the human attention mechanism, dynamically weighting the hidden state vector sequence H output by the LSTM encoding layer. This enables the target LSTM-Attention prediction model to automatically identify and focus on key historical time points that contribute most to the prediction of future HI values, such as early signs of failure or turning points in degradation trends, while ignoring secondary or irrelevant information.

[0143] In one optional embodiment, the attention layer dynamically weights the hidden state vector sequence to generate the context vector, which may specifically include the following sub-steps: Sub-step 41: Calculate attention score Each hidden state is computed using a small feedforward neural network. Importance score This network is learnable; it automatically learns how to score different hidden states during training. The calculation formula is as follows:

[0144] in, , and These are the learnable weight matrix, bias terms, and context vectors in the network. This is the activation function, used to introduce nonlinearity.

[0145] Sub-step 42: Attention weight normalization To transform the scores into a probability distribution, the Softmax function is used to calculate the scores at all time steps. Normalization is performed to obtain the final attention weights. :

[0146] Weight The value is between 0 and 1, and the sum of all weights is 1. The larger the value, the higher the hidden state at time step i. The more important it is for the final prediction.

[0147] Sub-step 43: Generate context vector The obtained attention weights With the corresponding LSTM hidden state Perform a weighted summation to generate a single, fixed-length context vector.

[0148]

[0149] Context vector It is a summary of the entire input sequence, which aggregates information from all time steps, but highlights the information at the moments that the model considers most critical.

[0150] Sub-step 5: The fully connected output layer generates the remaining lifetime of the power supply to be monitored based on the key time step information in the context vector.

[0151] The fully connected output layer serves as the model's final predictor, taking the highly condensed context vector generated by the Attention layer. Mapped to the final prediction result. Outputs a single numerical value, the predicted remaining lifespan.

[0152] In an optional embodiment, to eliminate false triggering of switching due to predicted fluctuations, this embodiment of the application includes a confirmation logic based on a time window. Specifically, a threshold for the startup time of backup power supply intervention is set. The system calculates the RUL in real time, and only if the RUL predicted in n consecutive times satisfies If the monitored power supply is about to fail, a switching command is immediately sent to the soft-switching control circuit. The monitored power supply is the main power supply, and the power supply switched by the soft-switching control circuit is the backup power supply. Timely switching to the backup power supply in the event of impending failure enables self-healing control of the integrated circuit power supply.

[0153] The integrated circuit power supply performance degradation prediction method disclosed in this invention involves smoothing and denoising the first health index data of the collected sample power supply using a Kalman filter algorithm to obtain a historical health index dataset. The historical health index dataset is then truncated into multiple first data sequences of preset window lengths using a sliding window approach, and the true remaining lifespan label corresponding to each first data sequence is determined. The first data sequences and their corresponding true remaining lifespan labels are stored to generate a supervised learning dataset. Initial parameters of the LSTM-Attention prediction model are determined based on the particle swarm optimization algorithm and the supervised learning dataset. The initial parameters of the LSTM-Attention prediction model are iteratively optimized and trained using the supervised learning dataset to obtain a target LSTM-Attention prediction model. Second health index data of the power supply to be monitored is collected within a preset window length. The remaining lifespan of the power supply to be monitored is then predicted based on the target LSTM-Attention prediction model. The method provided by this invention has several advantages. First, it employs an end-to-end prediction framework, enabling the model to directly learn the complex nonlinear mapping from the circuit health state sequence to its remaining lifetime, fundamentally avoiding the error accumulation problem inherent in traditional rolling prediction methods. Second, it combines an Attention mechanism that adaptively focuses on key fault features with Kalman filtering preprocessing that effectively removes noise, ensuring the accuracy and stability of the RUL prediction results. Third, it transforms high-precision prediction capabilities into precise and dynamic switching control. Since the model directly outputs RUL values ​​in time units, it provides a direct and quantitative basis for switching decisions, solving the problems of coarse time granularity and uncertain switching timing in traditional prediction methods. This allows the system to achieve power handover before a serious main power supply failure, effectively protecting the safety of the downstream load circuits. In summary, the solution disclosed in this invention transforms the reliability assurance of the power system from a traditional passive post-fault response mode to an active pre-fault warning and avoidance mode, improving the reliability and safety of integrated circuits.

[0154] This application also provides a method for assessing the health of a multi-channel secondary power supply on an integrated circuit aging bench. After assessing the health of the secondary power supply (i.e., the integrated circuit power supply mentioned above), the integrated circuit power supply performance degradation prediction method described above can be executed. In practice, the integrated circuit power supply performance degradation prediction method can be executed only when a health risk is assessed in the power supply, or it can be executed periodically. This application does not impose specific limitations on this approach.

[0155] Figure 4A flowchart is shown of a method for assessing the health of a multi-channel secondary power supply on an integrated circuit aging bench according to an embodiment of this application, which specifically includes the following steps.

[0156] Step S401: During the real-time operation phase of the integrated circuit aging bench, collect multi-dimensional real-time data of the multi-channel secondary power supply of the integrated circuit aging bench under the current operating conditions; In one embodiment of this application, after collecting multi-dimensional real-time data of the multi-channel secondary power supply of the integrated circuit aging bench under the current operating conditions, the method further includes: generating multiple real-time data segments with partial overlap by sliding slices of the multi-dimensional real-time data of the multi-channel secondary power supply under the current operating conditions according to a fixed window size, and obtaining a set of standardized real-time data segments by independently performing Z-Score standardization processing on each dimension of each real-time data segment.

[0157] Step S402: By inputting the multi-dimensional real-time data of the multi-channel secondary power supply under the current operating condition into the pre-trained deep neural network model, the fused feature vector, reconstructed waveform and prediction variance value of the multi-channel secondary power supply under the current operating condition are obtained. In one embodiment of this application, the deep neural network model includes a hybrid spatiotemporal encoder, a reconstruction decoder, and an uncertainty estimation head; wherein the hybrid spatiotemporal encoder includes a temporal encoder and a spatial encoder.

[0158] In one embodiment of this application, the step of inputting multi-dimensional real-time data of the multi-channel secondary power supply under the current operating condition into a pre-trained deep neural network model to obtain the fused feature vector, reconstructed waveform, and prediction variance of the multi-channel secondary power supply under the current operating condition includes: inputting the multi-dimensional real-time data of the multi-channel secondary power supply under the current operating condition into a time-domain encoder in the pre-trained deep neural network model to extract time-domain features, thereby obtaining a time-domain feature vector of the multi-channel secondary power supply under the current operating condition; inputting the multi-dimensional real-time data of the multi-channel secondary power supply under the current operating condition into a spatial encoder in the pre-trained deep neural network model to extract spatial features, thereby obtaining a spatial feature vector of the multi-channel secondary power supply under the current operating condition; and concatenating the time-domain feature vector output by the time-domain encoder with the spatial feature vector output by the spatial encoder to obtain the fused feature vector of the multi-channel secondary power supply under the current operating condition.

[0159] In one embodiment of this application, the step of inputting the multi-dimensional real-time data of the multi-channel secondary power supply under the current operating condition into a pre-trained deep neural network model to obtain the fused feature vector, reconstructed waveform, and prediction variance value of the multi-channel secondary power supply under the current operating condition includes: inputting the fused feature vector of the multi-channel secondary power supply under the current operating condition into the reconstruction decoder in the pre-trained deep neural network model to obtain the reconstructed waveform of the multi-channel secondary power supply under the current operating condition.

[0160] In one embodiment of this application, the step of obtaining the fused feature vector, reconstructed waveform, and predicted variance of the multi-channel secondary power supply under the current operating condition by inputting the multi-dimensional real-time data of the multi-channel secondary power supply under the current operating condition into a pre-trained deep neural network model includes: obtaining the predicted variance of the multi-channel secondary power supply under the current operating condition by inputting the fused feature vector of the multi-channel secondary power supply under the current operating condition into the uncertainty estimation head of the pre-trained deep neural network model.

[0161] Step S403: Using the fused feature vector, reconstructed waveform, and predicted variance of the multi-channel secondary power supply under the current operating conditions, obtain the health assessment results of the multi-channel secondary power supply of the integrated circuit aging bench.

[0162] In one embodiment of this application, obtaining the health assessment result of the multi-channel secondary power supply of the integrated circuit aging bench using the fused feature vector, reconstructed waveform, and prediction variance value of the multi-channel secondary power supply under the current operating condition includes: calculating the Mahalanobis distance between the fused feature vector of the multi-channel secondary power supply under the current operating condition and a pre-built health fingerprint database, and converting the Mahalanobis distance into a consistency health index; calculating the root mean square error between the reconstructed waveform and the original waveform of the multi-channel secondary power supply under the current operating condition, and converting the root mean square error into a reconstruction health index; calculating the dynamic weighting coefficient for assessing the current operating state of the integrated circuit aging bench using the prediction variance value of the multi-channel secondary power supply under the current operating condition; and calculating the comprehensive health index of the multi-channel secondary power supply under the current operating condition using the consistency health index, reconstruction health index, and dynamic weighting coefficient, and using the comprehensive health index as the health assessment result of the multi-channel secondary power supply of the integrated circuit aging bench.

[0163] The multi-channel secondary power supply health assessment scheme for integrated circuit aging bench provided in this application has the following advantages: (1) It has a deep channel correlation perception capability and avoids "island-style" assessment: The introduction of graph attention network (GAT) and mask collaborative reconstruction mechanism breaks the limitation of traditional methods that independently assess a single channel. The model not only focuses on the waveform distortion of a single channel, but also can keenly capture the abnormal linkage of "inconsistent pace" between multiple channels (such as abnormal current sharing and parallel imbalance), so as to identify early faults where the single channel indicators seem normal but the system has hidden dangers. (2) It uses a dynamic fusion strategy with uncertainty perception to significantly improve robustness: For the common load changes and electromagnetic interference in industrial sites, this application uses Bayesian uncertainty estimation to quantify the "confidence" of the model in real time. Through the dynamic weight generation mechanism, the system can smoothly switch between "high sensitivity consistency detection under steady state" and "high robustness reconstruction detection under transient interference", which effectively solves the problem of false positives caused by operating condition drift in traditional methods. (3) A dual verification loop of "feature space + physical waveform" is constructed: This application outputs a consistency index (CHI) based on statistical distribution in parallel. C ) and the physical signal-based reconstruction error index (CHI) R The former is sensitive to minor degradation, while the latter is reliable for serious faults. The two complement and verify each other, greatly improving the credibility and interpretability of the evaluation results. (4) It has extremely high sensitivity to early minor faults: By extracting high-dimensional spatiotemporal features through a hybrid spatiotemporal encoder and combining them with Mahalanobis distance statistical measures, this application can amplify the minute offsets that are difficult to detect in the original signal in the feature space, thereby achieving predictive maintenance before the power supply performance deteriorates substantially.

[0164] The main purpose of this application is to provide a method for health assessment of multi-channel secondary power supplies in an integrated circuit aging test bench. This method aims to solve the problems of traditional methods in assessing multi-channel secondary power supplies, such as ignoring electrical coupling between channels, weak anti-interference capability under atypical operating conditions, and insufficient reliability of single assessment indicators. This will enable accurate identification and robust assessment of early minor faults in the secondary power supply system.

[0165] To achieve the above objectives, this application provides a multi-channel secondary power supply health assessment method for an integrated circuit aging bench, which includes two stages: offline model construction and training, and online real-time health assessment.

[0166] In the offline model building and training phase: First, multi-dimensional time-series data of multi-channel power supplies under different load conditions are collected and processed by sliding slicing and Z-score normalization. Second, a deep neural network including a hybrid spatiotemporal encoder, a reconstruction decoder, and an uncertainty estimation head is constructed and trained. The hybrid spatiotemporal encoder includes a temporal encoder and a spatial encoder: the temporal encoder uses a one-dimensional convolutional network (1D-CNN) and a Transformer encoder to extract high-frequency ripple and long-range temporal dependencies of a single channel; the spatial encoder uses a graph attention network (GAT) to construct a fully connected graph based on the electrical connections between channels, adaptively aggregating neighborhood information to capture the topological correlation features between channels. Simultaneously, a masked collaborative reconstruction mechanism is used for model training. By randomly masking some channel data, the model is forced to use the features of the remaining healthy channels and the learned inter-channel correlation weights to deduce the masked waveforms and simultaneously output the predicted uncertainty variance. Finally, steady-state data with low reconstruction errors are selected to construct a health fingerprint database, and the mean vector and covariance matrix in the latent feature space are calculated as health benchmarks.

[0167] During the online real-time evaluation phase: the real-time collected data is input into the trained deep neural network to perform a dynamic fusion evaluation based on two indicators: First, the Consistency Health Index (CHI) is calculated. C First, the Mahalanobis distance between the latent feature vectors (i.e., fused feature vectors) of real-time data and the health fingerprint database is used to quantify the degree of consistency between the current state and the healthy distribution. Second, the reconstruction error health index (CHI) is calculated. R The first step involves quantifying the deep neural network's interpretability of the current data by using the root mean square error between the reconstructed waveform and the original waveform, based on the output of the deep neural network. The second step is to perform adaptive fusion based on uncertainty awareness: the uncertainty estimation head of the deep neural network outputs the prediction variance in real time, generating dynamic weight coefficients through nonlinear mapping. When the system is in a steady-state, high-confidence scenario, the focus is on consistency health indicators. When the system encounters transient disturbances or atypical operating conditions leading to a surge in uncertainty, the dependence on feature distance is automatically reduced, and the focus shifts to reconstruction error indicators, ultimately generating a comprehensive health index (CHI). The original waveform is obtained by preprocessing real-time data.

[0168] This application presents a complete and rigorous technical loop, from data processing and the construction and training of deep neural network models to multidimensional evaluation. It can provide unprecedented in-depth health insights for critical power systems, effectively address the limitations of traditional methods, and provide a scientific basis for the maintenance and management of power systems. Figure 5 The structural block diagram of an integrated circuit power supply compensation and redundancy hybrid self-healing control device is shown in the embodiment of this application.

[0169] The integrated circuit power supply compensation and redundancy hybrid self-healing control device provided in this application includes the following functional modules: The acquisition module 501 is used to acquire preset parameters of the power supply to be evaluated; The generation module 502 is used to analyze the preset parameters using a degradation feature extraction algorithm to generate a degradation feature vector; Anomaly type determination module 503 is used to determine the anomaly type of the power supply to be evaluated based on the degradation feature vector, wherein the anomaly type includes: transient disturbance and permanent fault; The first calculation module 504 is used to call the degradation perception compensation algorithm to determine the PWM duty cycle adjustment amount based on the degradation feature vector when the power supply to be evaluated experiences a transient disturbance. The compensation module 405 is used to generate a PWM duty cycle adjustment command based on the PWM duty cycle adjustment amount and trigger the supercapacitor to discharge so that the power supply voltage to be evaluated is restored to a safe threshold. The fault type determination module 506 is used to determine the fault type of the power supply to be evaluated based on the real-time preset parameter information of the power supply to be evaluated and the degradation feature vector when the voltage of the power supply to be evaluated fails to recover to the safety threshold or the power supply to be evaluated is permanently faulty within a first preset time period. The switching module 507 is used to match the target backup module for switching based on the mapping rules between the preset fault types and backup modules in the system.

[0170] Optionally, the generation module includes: The first submodule is used to input the preset parameters into the degradation feature extraction algorithm; wherein, the degradation feature extraction algorithm includes the recursive least squares method; The second submodule is used to identify the preset parameters using the recursive least squares method to obtain the equivalent series resistance of the filter capacitor. The third submodule is used to dynamically calculate the on-resistance offset of the power switch based on the initial on-resistance, junction temperature, and junction temperature-resistance relationship model of the power transistor. The fourth submodule is used to generate a degradation feature vector based on the equivalent series resistance of the filter capacitor and the offset of the on-resistance of the power switch transistor.

[0171] Optionally, the degradation sensing compensation algorithm includes a temperature compensation term and a capacitance compensation term; The temperature compensation term is determined based on the temperature compensation coefficient, the initial equivalent series resistance of the filter capacitor, the voltage change value, and the equivalent series resistance of the filter capacitor. The capacitance compensation term is determined based on the capacitance compensation coefficient, the initial capacitance value, the estimated capacitance value, and the voltage change value.

[0172] Optionally, the exception type determination module is specifically used for: The duration of the continuous voltage drop of the power supply to be detected is determined by the degradation feature vector. If the duration of the continuous drop does not exceed the second preset duration, it is determined that the power supply to be evaluated has experienced a transient disturbance. If the duration of the continuous drop exceeds the second preset duration, it is determined that the power supply to be evaluated has suffered a permanent failure.

[0173] Optionally, the fault type determination module is specifically used for: Based on the real-time preset parameter information of the power supply to be evaluated, the output voltage ripple, output current, output voltage, and junction temperature of the power supply to be tested are determined. The change in the equivalent series resistance of the filter capacitor is determined based on the degradation feature vector. If the increase in the output voltage ripple of the power supply under test is greater than a preset level, and the change in the equivalent series resistance of the filter capacitor is greater than a preset change threshold, a filter capacitor degradation fault is determined to have occurred. If the input current increases sharply, the output voltage drops sharply, and the junction temperature rises sharply, then a short-circuit breakdown fault of the power switch of the power supply under evaluation is determined to have occurred.

[0174] Optionally, the switching module is specifically used for: In the event of a filter capacitor degradation fault in the power supply to be evaluated, switch to a capacitor branch with a low equivalent series resistance. In the event that the power supply under evaluation experiences a short-circuit breakdown fault in the power switch, the power switch with the highest health assessment value is switched to.

[0175] The embodiments provided in this application Figure 5 The integrated circuit power supply compensation and redundancy hybrid self-healing control device shown can achieve Figure 1 The various processes implemented in the method implementation examples will not be described again here to avoid repetition.

[0176] The integrated circuit power supply compensation and redundancy hybrid self-healing control device provided in this application, upon detecting a power supply anomaly, prioritizes activating the first line of defense, adaptive compensation, to address transient disturbances. Only when compensation proves ineffective or a permanent fault is confirmed is the second line of defense, redundant switching, activated. This dual-defense collaborative mechanism minimizes test interruptions caused by erroneous switching and, in the event of a real fault, ensures rapid functional recovery of the power system by quickly isolating the faulty module and switching to a healthy backup module. This achieves full-process power supply autonomy, from "perceived degradation" to "transient suppression" and then to "permanent recovery." This solution effectively addresses the protection inaccuracies and fault misjudgments present in existing solutions.

[0177] This invention also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus.

[0178] Memory, used to store computer programs; When the processor executes the program stored in the memory, it implements the integrated circuit power supply compensation and redundancy hybrid self-healing control method shown in the above method embodiments.

[0179] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc.

[0180] The communication interface is used for communication between the aforementioned terminal and other devices.

[0181] The memory may include random access memory (RAM) or non-volatile memory, such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0182] In another embodiment of the present invention, a computer-readable storage medium is also provided, which stores instructions that, when executed on an electronic device, cause the electronic device to implement the integrated circuit power supply compensation and redundancy hybrid self-healing control method described in any of the above embodiments.

[0183] In another embodiment of the present invention, a computer program product containing instructions is also provided, which, when run on an electronic device, causes the electronic device to implement the integrated circuit power supply power analysis self-healing control method described in any of the above embodiments.

[0184] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0185] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A self-healing control method for integrated circuit power supply compensation and redundancy, characterized in that, The method includes: Collect the preset parameters of the power source to be evaluated; The preset parameters are analyzed using a degradation feature extraction algorithm to generate a degradation feature vector; Based on the degradation feature vector, the anomaly type of the power supply to be evaluated is determined, wherein the anomaly type includes: transient disturbances and permanent faults; When the power supply to be evaluated experiences a transient disturbance, the degradation sensing compensation algorithm is invoked to determine the PWM duty cycle adjustment amount based on the degradation feature vector; A PWM duty cycle adjustment command is generated based on the PWM duty cycle adjustment amount, and the supercapacitor is triggered to discharge so that the power supply voltage to be evaluated is restored to a safe threshold. If the voltage of the power supply to be evaluated fails to recover to a safe threshold or the power supply to be evaluated suffers a permanent fault within a first preset time period, the fault type of the power supply to be evaluated is determined based on the real-time preset parameter information of the power supply to be evaluated and the degradation feature vector. Based on the pre-defined mapping rules between fault types and backup modules in the system, the target backup module is matched and switched over.

2. The method according to claim 1, characterized in that, The step of analyzing the preset parameters using a degradation feature extraction algorithm to generate a degradation feature vector includes: The preset parameters are input into the degradation feature extraction algorithm; wherein, the degradation feature extraction algorithm includes the recursive least squares method; The preset parameters are identified using the recursive least squares method to obtain the equivalent series resistance of the filter capacitor. Based on the initial on-resistance, junction temperature, and the relationship between junction temperature and resistance of the power transistor, the on-resistance offset of the power switch is dynamically calculated. A degradation feature vector is generated based on the equivalent series resistance of the filter capacitor and the offset of the on-resistance of the power switch transistor.

3. The method according to claim 2, characterized in that, The degradation sensing compensation algorithm includes a temperature compensation term and a capacitance compensation term; The temperature compensation term is determined based on the temperature compensation coefficient, the initial equivalent series resistance of the filter capacitor, the voltage change value, and the equivalent series resistance of the filter capacitor. The capacitance compensation term is determined based on the capacitance compensation coefficient, the initial capacitance value, the estimated capacitance value, and the voltage change value.

4. The method according to claim 1, characterized in that, The step of determining the anomaly type of the power source to be evaluated based on the degradation feature vector includes: The duration of the continuous voltage drop of the power supply to be detected is determined by the degradation feature vector. If the duration of the continuous drop does not exceed the second preset duration, it is determined that the power supply to be evaluated has experienced a transient disturbance. If the duration of the continuous drop exceeds the second preset duration, it is determined that the power supply to be evaluated has suffered a permanent failure.

5. The method according to claim 1, characterized in that, The step of determining the fault type of the power supply to be evaluated based on the real-time preset parameter information and the degradation feature vector includes: Based on the real-time preset parameter information of the power supply to be evaluated, the output voltage ripple, output current, output voltage, and junction temperature of the power supply to be tested are determined. The change in the equivalent series resistance of the filter capacitor is determined based on the degradation feature vector. If the increase in the output voltage ripple of the power supply under test is greater than a preset level, and the change in the equivalent series resistance of the filter capacitor is greater than a preset change threshold, a filter capacitor degradation fault is determined to have occurred. If the input current increases sharply, the output voltage drops sharply, and the junction temperature rises sharply, then a short-circuit breakdown fault of the power switch of the power supply under evaluation is determined to have occurred.

6. The method according to claim 5, characterized in that, Based on the system's preset mapping rules between fault types and backup modules, the steps for matching and switching to the target backup module include: In the event of a filter capacitor degradation fault in the power supply to be evaluated, switch to a capacitor branch with a low equivalent series resistance. In the event that the power supply under evaluation experiences a short-circuit breakdown fault in the power switch, the power switch with the highest health assessment value is switched to.

7. A hybrid self-healing control device for integrated circuit power supply compensation and redundancy, characterized in that, The device includes: The data acquisition module is used to acquire preset parameters of the power supply to be evaluated. The generation module is used to analyze the preset parameters using a degradation feature extraction algorithm to generate a degradation feature vector; An anomaly type determination module is used to determine the anomaly type of the power supply to be evaluated based on the degradation feature vector, wherein the anomaly type includes: transient disturbance and permanent fault; The first calculation module is used to call the degradation perception compensation algorithm to determine the PWM duty cycle adjustment amount based on the degradation feature vector when the power supply to be evaluated experiences a transient disturbance. The compensation module is used to generate a PWM duty cycle adjustment command based on the PWM duty cycle adjustment amount and trigger the supercapacitor to discharge so that the power supply voltage to be evaluated is restored to a safe threshold. The fault type determination module is used to determine the fault type of the power supply to be evaluated based on the real-time preset parameter information of the power supply to be evaluated and the degradation feature vector when the voltage of the power supply to be evaluated fails to recover to the safety threshold or the power supply to be evaluated is permanently faulty within a first preset time period. The switching module is used to match the target backup module and switch it according to the mapping rules between the preset fault types and backup modules in the system.

8. The apparatus according to claim 7, characterized in that, The generation module includes: The first submodule is used to input the preset parameters into the degradation feature extraction algorithm; wherein, the degradation feature extraction algorithm includes the recursive least squares method; The second submodule is used to identify the preset parameters using the recursive least squares method to obtain the equivalent series resistance of the filter capacitor. The third submodule is used to dynamically calculate the on-resistance offset of the power switch based on the initial on-resistance, junction temperature, and junction temperature-resistance relationship model of the power transistor. The fourth submodule is used to generate a degradation feature vector based on the equivalent series resistance of the filter capacitor and the offset of the on-resistance of the power switch transistor.

9. The apparatus according to claim 8, characterized in that, The degradation sensing compensation algorithm includes a temperature compensation term and a capacitance compensation term; The temperature compensation term is determined based on the temperature compensation coefficient, the initial equivalent series resistance of the filter capacitor, the voltage change value, and the equivalent series resistance of the filter capacitor. The capacitance compensation term is determined based on the capacitance compensation coefficient, the initial capacitance value, the estimated capacitance value, and the voltage change value.

10. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a program stored in memory, it implements the integrated circuit power supply compensation and redundancy hybrid self-healing control method as described in any one of claims 1-6.

Images