A power adapter life prediction system and method

Abstract

The application relates to the technical field of electrical performance detection, and discloses a power adapter life prediction system and method. The system comprises a reference establishment module, which obtains working condition quantities and sampling values composed of steady-state load current and reference temperature in an initial operation stage, and establishes a multi-working condition reference matrix; an event-triggered sampling module, which obtains measured values and working condition quantities of an activated parameter domain when a discrete operation event occurs, and records a domain time stamp; a data normalization mapping module, which calculates a single degradation index according to the measured values and reference values, generates a domain degradation value and a working condition confidence coefficient; a health degree fusion module, which obtains a time-effect confidence coefficient according to the domain time stamp, corrects an initial weight in combination with the working condition confidence coefficient, and calculates a whole machine comprehensive health degree; and a life prediction and early warning module, which predicts a remaining life length according to the whole machine comprehensive health degree and outputs a graded early warning state. The application reduces sampling and operation loads, and realizes power adapter degradation state evaluation, remaining life prediction and graded early warning.

Description

Technical Field

[0001] This invention relates to the field of electrical performance testing technology, specifically to a power adapter life prediction system and method. Background Technology

[0002] A power adapter is a power electronic device that converts alternating current (AC) into direct current (DC) for electronic devices. It typically includes an input rectifier and filter circuit, a high-frequency switching converter circuit, an output rectifier and filter circuit, a feedback control circuit, and a safety protection circuit. During long-term operation, the internal electrolytic capacitors, power switching devices, magnetic components, feedback devices, and safety protection devices of a power adapter are susceptible to parameter drift, performance degradation, and changes in operating characteristics due to load and temperature variations. This, in turn, affects output stability, power conversion efficiency, and safety protection performance. Therefore, assessing the degradation status of a power adapter and predicting its remaining lifespan through electrical performance testing is crucial for fault prediction and health management.

[0003] Current power adapter status detection methods typically involve continuously collecting single parameters such as output voltage ripple, device temperature, or power conversion efficiency using sensors, comparing the collected results with fixed thresholds, and outputting a fault or replacement prompt when the parameter reaches the corresponding limit. Some safety protection performance tests also require external testing equipment to detect protection action thresholds, protection response times, or tripping execution status while the device is stopped.

[0004] However, the degradation process of different components within a power adapter is reflected simultaneously in multiple aspects, including output quality, efficiency, thermal state, and safety protection status. Relying solely on a single parameter for threshold judgment makes it difficult to comprehensively reflect the overall degradation state of the power adapter. Furthermore, continuously and frequently collecting multiple parameters increases the data processing and storage burden on the microcontroller, which is detrimental to long-term operation within a power adapter with limited computing resources. In addition, existing detection methods based on fixed thresholds typically only determine whether a single parameter has reached the fault boundary, lacking comprehensive processing of multiple degradation parameters and their temporal trends. This makes it difficult to continuously obtain health indicators that characterize the overall degradation level of the power adapter, and also makes it difficult to predict the remaining lifespan based on the gradual aging process. Therefore, it is not conducive to tiered early warning and maintenance arrangements before power supply anomalies occur.

[0005] Therefore, this invention proposes a power adapter life prediction system and method to address the shortcomings of existing technologies. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a power adapter life prediction system and method, which solves the problems that existing single parameters are difficult to objectively assess the overall degradation status of power adapters, and that continuous high-frequency sampling increases the computational load on microcontrollers and makes it difficult to perform remaining life prediction on the device side.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A power adapter life prediction system, comprising: The reference establishment module is configured to acquire operating conditions consisting of steady-state load current and reference temperature during the initial operation phase of the power adapter, and collect corresponding sampled values ​​to form a multi-operating condition reference matrix containing reference values. The event-triggered sampling module is configured to acquire the measured value of at least one parameter domain among the output quality domain, efficiency thermal domain, and safety protection domain activated by the current discrete operating event when the discrete operating event occurs, synchronously acquire the operating condition quantity at the time of the event, and record the domain time stamp corresponding to the activated parameter domain. The data normalization mapping module is configured to obtain the benchmark value from the multi-condition benchmark matrix based on the operating condition quantity, compare the measured value with the benchmark value to obtain a single-item degradation index, generate a domain degradation value from the single-item degradation index in the same parameter domain, and generate an operating condition confidence coefficient based on the degree of deviation of the operating condition quantity from the central benchmark operating condition in the multi-condition benchmark matrix. The health fusion module is configured to obtain a timeliness confidence coefficient based on the domain timescale, generate a joint confidence coefficient by combining the operating condition confidence coefficient and the timeliness confidence coefficient, correct the initial weights based on the joint confidence coefficients to obtain corrected weights, and weight the corrected weights with the domain degradation values ​​to obtain the overall health of the machine. The lifespan prediction and early warning module is configured to predict the remaining lifespan based on the overall health of the machine and output a graded early warning status.

[0008] Preferably, in the reference establishment module, the step of acquiring the operating conditions consisting of steady-state load current and reference temperature during the initial operation phase of the power adapter, and collecting corresponding sampled values ​​to form a multi-operating-condition reference matrix containing reference values ​​includes: Read the steady-state load current and the reference temperature at the current moment, and combine the synchronously read steady-state load current and the reference temperature into an array to form the operating condition quantity; The sampling value is formed by synchronously triggering sampling within the same polling cycle after acquiring the operating condition value, or by asynchronously sampling using discrete operating events that occur during the initial operating phase. The full-load output current range of the power adapter is divided into multiple consecutive load current ranges at equal intervals, and the operating temperature range is divided into multiple consecutive reference temperature ranges at equal intervals. The intersection of the load current range and the reference temperature range is determined as the grid coordinate position, and the sampled value is assigned to the corresponding operating condition range according to the corresponding operating condition quantity. The baseline value is obtained by performing mean filtering on the historical cumulative sample values ​​falling within the same operating condition range, and the baseline value calculated by each parameter channel is written into the corresponding storage address to construct the multi-operating condition baseline matrix.

[0009] Preferably, in the event-triggered sampling module, the steps of acquiring the measured value of at least one parameter domain among the output quality domain, efficiency thermal domain, and safety protection domain activated by the current discrete operating event when the discrete operating event occurs, synchronously acquiring the operating condition quantity at the time of the event, and recording the domain timescale corresponding to the activated parameter domain include: Capture the discrete operating events, including power-on events, load jump events, timer cycle completion events, leakage current protection activation events, and active excitation test events; The output quality domain is activated when the power-on event or the load jump event is captured; the efficiency thermal domain is activated when the timer reaches its cycle event; and the safety protection domain is activated when the leakage protection action event or the active excitation test event is triggered. Obtain the measured values ​​corresponding to the activated parameter domain, including the peak output ripple, voltage recovery time, start-up stabilization time, real-time power conversion efficiency, critical hot spot temperature, device temperature rise rate, protection action threshold drift, protection response time, and trip coil impedance. The steady-state load current after the latched transient fluctuation subsides is combined with the reference temperature to form the operating condition quantity. The total power-on time of the current device when the current parameter acquisition is completed is extracted and recorded as the domain time stamp corresponding to the activated parameter domain.

[0010] Preferably, in the event-triggered sampling module, the step of obtaining the measured value of at least one parameter domain among the output quality domain, efficiency thermal domain, and safety protection domain activated by the current discrete running event further includes: Extract temperature data from two points with a set time difference and perform time difference calculation to calculate the slope of the change in the temperature of the key hot spot within a unit time, so as to obtain the device temperature rise rate as the measured value.

[0011] Preferably, in the data normalization mapping module, the step of obtaining the benchmark value from the multi-condition benchmark matrix based on the operating condition quantity, and comparing the measured value with the benchmark value to obtain the single degradation index includes: Using the starting physical address of the multi-condition reference matrix as the base address, the memory offset corresponding to the condition quantity is calculated, and the reference value corresponding to the condition quantity is calculated using the bilinear interpolation algorithm. The intermediate degradation value is obtained by performing algebraic operations on the extracted measured value, the reference value, and the end-of-life value. The numerator of the intermediate degradation value is the difference between the measured value and the reference value or the difference between the reference value and the measured value. The denominator of the intermediate degradation value is the difference between the end-of-life value and the reference value or the difference between the reference value and the end-of-life value. The degradation is then calculated by superimposing the degradation values ​​of the two values ​​to obtain the positive floating-point protection value. The intermediate degradation values ​​less than zero are set to zero, and the intermediate degradation values ​​greater than one are set to one, to obtain the single degradation index.

[0012] Preferably, in the data normalization mapping module, the step of generating domain degradation values ​​from individual degradation indices within the same parameter domain, and generating working condition confidence coefficients based on the degree of deviation of the working condition quantity from the central reference working condition in the multi-working-condition reference matrix, includes: The domain degradation value is generated by arithmetically averaging multiple individual degradation indices within the same parameter domain. Obtain the rated typical current and rated typical temperature corresponding to the central reference condition in the multi-condition reference matrix; Based on the square of the first difference between the normalized steady-state load current and the rated typical current in the operating condition quantity, and the square of the second difference between the normalized reference temperature and the rated typical temperature in the operating condition quantity, the spatial Euclidean distance between the operating condition quantity and the central reference operating condition is calculated. The negative value of the product of the spatial Euclidean distance and the attenuation adjustment coefficient is used as the exponent of the natural constant to obtain the confidence coefficient of the operating condition.

[0013] Preferably, in the health fusion module, the step of obtaining the timeliness confidence coefficient based on the domain timescale and generating a joint confidence coefficient by combining the operating condition confidence coefficient and the timeliness confidence coefficient includes: Read the total power-on time of the current device, and calculate the time lag by subtracting the total power-on time of the current device from the domain time scale; The time lag is multiplied by the time decay constant, and the negative value is used as the exponent of the natural constant to obtain the time confidence coefficient. The joint confidence coefficient is generated by multiplying the operating condition confidence coefficient and the time-related confidence coefficient.

[0014] Preferably, in the health fusion module, the step of obtaining the corrected weights based on the joint confidence coefficient to correct the initial weights includes: The initial weights are multiplied by the joint confidence coefficients to obtain the intermediate retention weights; The total weight missing share is obtained by subtracting the value one from the sum of the intermediate retained weights of all the parameter domains. The relative allocation ratio is obtained by dividing the joint confidence coefficient of a specific parameter domain by the sum of the joint confidence coefficients of all parameter domains. The total weight missing share is multiplied by the relative allocation ratio and then added to the intermediate retention weight of the specific parameter domain to obtain the corrected weight of the specific parameter domain.

[0015] Preferably, in the lifespan prediction and early warning module, the step of predicting the remaining lifespan based on the overall health of the machine and outputting a graded early warning status includes: The overall health status of the machine is combined with the total power-on time of the current device into a two-dimensional data pair. The two-dimensional data pairs with a total number of valid samples greater than or equal to the minimum number of samples are extracted to form a time series. The time series was fitted by linear regression using the least squares method to calculate the degradation slope and initial intercept of the degradation trend prediction equation. When the degradation slope is less than or equal to zero, the degradation slope is set to the lower limit of the degradation slope, and the initial intercept is recalculated using the lower limit of the degradation slope. Substitute a safety margin threshold less than one into the degradation trend prediction equation to calculate the theoretical endpoint time. Subtract the theoretical endpoint time from the current total power-on time of the device to obtain the remaining lifespan.

[0016] The present invention also provides a method for predicting the lifespan of a power adapter, comprising the following steps: During the initial operation phase of the power adapter, the operating conditions consisting of steady-state load current and reference temperature are acquired, and the corresponding sampled values ​​are collected to form a multi-operating condition reference matrix containing reference values. When a discrete operation event occurs, the measured value of at least one parameter domain among the output quality domain, efficiency thermal domain, and safety protection domain activated by the current discrete operation event is obtained, the operating condition quantity at the time of the event is obtained synchronously, and the domain time stamp corresponding to the activated parameter domain is recorded. The reference value is obtained from the multi-condition reference matrix based on the operating condition quantity. The measured value is compared with the reference value to obtain a single degradation index. A domain degradation value is generated from the single degradation index in the same parameter domain. The operating condition confidence coefficient is generated based on the degree of deviation of the operating condition quantity from the central reference operating condition in the multi-condition reference matrix. Based on the domain timescale, the timeliness confidence coefficient is obtained. The operating condition confidence coefficient and the timeliness confidence coefficient are combined to generate a joint confidence coefficient. The initial weights are corrected based on the joint confidence coefficient to obtain the corrected weights. The corrected weights are then weighted with the domain degradation value to obtain the overall health of the machine. Based on the overall health status of the machine, the remaining lifespan is predicted and a graded early warning status is output.

[0017] This invention provides a power adapter lifespan prediction system and method. It has the following beneficial effects: 1. This invention obtains measured values ​​of the output quality domain, efficiency thermal domain, and safety protection domain through a discrete operation event triggering mechanism, and compares them with the benchmark values ​​in the multi-condition benchmark matrix, solving the problem that a single parameter is insufficient to reflect the overall degradation status of the power adapter. The fusion of degradation information from multiple parameter domains provides data support for fault prediction and health management, and outputs the overall health status of the entire unit, remaining lifespan, and graded early warning status.

[0018] 2. This invention generates a joint confidence coefficient from the operating condition confidence coefficient and the time-related confidence coefficient, and corrects the initial weights based on the joint confidence coefficient. Then, it calculates the overall health of the machine by combining the domain degradation value, so as to reduce the evaluation bias caused by fixed weights under different operating conditions and data time lag conditions. By performing trend fitting on the time series formed by the overall health of the machine with the current total power-on time of the equipment, the remaining lifespan is predicted, providing data reference for graded early warning and maintenance operations.

[0019] 3. This invention triggers parameter acquisition through discrete running events and uses timers to supplement the sampled data in continuous running states when periodic events are reached. This reduces the computational load on the microcontroller caused by continuous high-frequency sampling, enabling the calculation of domain degradation values, the fusion of overall machine health, and the prediction of remaining lifespan to be completed in the microcontroller, and outputting graded early warning status. Attached Figure Description

[0020] Figure 1 This is a diagram of a power adapter life prediction system architecture according to the present invention; Figure 2 This is a flowchart of a power adapter life prediction method according to the present invention; Figure 3 This is a schematic diagram of the benchmark establishment mechanism of the present invention; Figure 4 This is a schematic diagram of event-triggered sampling according to the present invention; Figure 5 This is a schematic diagram of the data normalization mapping of the present invention; Figure 6 This is a schematic diagram of the health fusion calculation of the present invention; Figure 7 This is a schematic diagram illustrating the lifespan prediction and early warning system of the present invention. Detailed Implementation

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Reference Figure 1 This invention provides a power adapter life prediction system. The functional modules of the power adapter life prediction system are deployed in a microcontroller inside the power adapter and run by calling the underlying hardware resources of the microcontroller. The microcontroller integrates an analog-to-digital converter, an internal data bus, and non-volatile memory. The signal input terminal of the microcontroller is electrically connected to the parameter acquisition circuits set in each functional node of the power adapter. The communication output terminal of the microcontroller is connected to a warning indication circuit. The warning indication circuit includes an indicator light and a communication circuit. The indicator light is used to output the light warning status, and the communication circuit is used to output maintenance prompt information including the overall health status of the device, the remaining lifespan, and the graded warning status.

[0023] The power adapter life prediction system includes a baseline establishment module, an event-triggered sampling module, a data normalization mapping module, a health fusion module, and a life prediction and early warning module. These modules reside in the microcontroller as program firmware and transfer data and call parameters through the microcontroller's internal data bus.

[0024] The reference establishment module establishes a read / write connection with the non-volatile memory through the internal data bus. During the initial operation phase of the power adapter, the reference establishment module acquires the operating conditions consisting of steady-state load current and reference temperature, and collects the corresponding sampled values. The reference establishment module processes the sampled values ​​to form a multi-operating-condition reference matrix containing reference values.

[0025] The event-triggered sampling module establishes a signal connection with the microcontroller's internal running event monitoring circuit and hardware timer. When a discrete running event occurs, the event-triggered sampling module generates a response action, acquiring the measured value of at least one parameter domain among the output quality domain, efficiency thermal domain, and safety protection domain activated by the current discrete running event. The output quality domain, efficiency thermal domain, and safety protection domain are parameter sets divided according to the power adapter's performance indicators. Within the sampling processing cycle after a single discrete running event is triggered, the event-triggered sampling module synchronously acquires the operating conditions at the time of the event and calls the cumulative running timer to record the domain timestamps corresponding to the activated parameter domains. The event-triggered sampling module transmits the acquired measured values, operating conditions, and domain timestamps to the data normalization mapping module and the health fusion module.

[0026] The data normalization mapping module receives data transmitted by the event-triggered sampling module and retrieves the multi-condition benchmark matrix generated by the benchmark establishment module from the non-volatile memory via the address pointer. The data normalization mapping module obtains the corresponding benchmark value from the multi-condition benchmark matrix based on the condition quantity. The data normalization mapping module compares the measured value with the benchmark value to obtain the single-item degradation index. The single-item degradation index within the same parameter domain generates the domain degradation value. Based on the degree of deviation of the condition quantity from the central benchmark condition in the multi-condition benchmark matrix, the data normalization mapping module generates the condition confidence coefficient. The data normalization mapping module outputs the generated domain degradation value and condition confidence coefficient to the health fusion module.

[0027] The health fusion module receives the calculation results output by the data normalization mapping module and the domain timescale provided by the event-triggered sampling module. The health fusion module obtains the time-of-failure confidence coefficient based on the domain timescale, generates a joint confidence coefficient by combining the operating condition confidence coefficient and the time-of-failure confidence coefficient, corrects the initial weights based on the joint confidence coefficient, and weights the corrected weights with the domain degradation value to obtain the overall health of the machine. The health fusion module outputs the calculated overall health of the machine to the life prediction and early warning module.

[0028] The logic input of the life prediction and early warning module is connected to the output of the health fusion module, and the drive output is connected to the early warning indicator circuit outside the power adapter. The life prediction and early warning module transfers the overall health of the whole machine to non-volatile memory. The life prediction and early warning module predicts the remaining lifespan based on the overall health of the whole machine and outputs a graded early warning status. The drive early warning indicator circuit outputs maintenance prompt information.

[0029] Reference Figure 2 This invention provides a power adapter lifespan prediction method, which is executed by the aforementioned power adapter lifespan prediction system and specifically includes the following steps: S1, during the initial operation phase of the power adapter, acquire the operating conditions consisting of steady-state load current and reference temperature, and collect the corresponding sampled values ​​to form a multi-operating condition reference matrix containing reference values. S2, when a discrete operation event occurs, acquire the measured value of at least one parameter domain among the output quality domain, efficiency thermal domain, and safety protection domain activated by the current discrete operation event, simultaneously acquire the operating conditions at the time of the event, and record the domain timestamp corresponding to the activated parameter domain; S3. Based on the working condition quantity, obtain the benchmark value from the multi-working condition benchmark matrix, compare the measured value with the benchmark value to obtain the single-item degradation index, generate the domain degradation value from the single-item degradation index in the same parameter domain, and generate the working condition confidence coefficient based on the degree of deviation of the working condition quantity from the central benchmark working condition in the multi-working condition benchmark matrix. S4. Obtain the timeliness confidence coefficient based on the domain timescale, generate a joint confidence coefficient from the operating condition confidence coefficient and the timeliness confidence coefficient, correct the initial weights based on the joint confidence coefficients, and weight the corrected weights with the domain degradation values ​​to obtain the overall health of the machine. S5 predicts the remaining lifespan based on the overall health of the machine and outputs a graded warning status.

[0030] The following combination Figures 3 to 7 The implementation process of the above functional modules and corresponding steps will be explained separately.

[0031] Reference Figure 3 The benchmark establishment module executes step S1, which includes the following steps: S11, determine the initial running phase and obtain basic parameters.

[0032] In this embodiment, the power adapter's power-on duration is recorded by the cumulative running timer inside the microcontroller. When the power-on duration is less than the set initial reference establishment time threshold, the reference establishment module determines that the power adapter is in the initial operation phase. When the power-on duration is greater than or equal to the initial reference establishment time threshold, the reference establishment module ends the initial reference establishment state, and the sampling module is triggered by the event described later to perform routine monitoring within the life cycle.

[0033] In a preferred embodiment, the initial reference establishment time threshold is set to 50 to 150 hours. The initial reference establishment time threshold corresponds to the reference acquisition duration used to form a multi-condition reference matrix during the initial commissioning of the power adapter, and the initial reference establishment state ends after the time threshold is reached.

[0034] During the initial operation phase, the reference establishment module activates the analog-to-digital conversion channel with a polling cycle of, for example, 1 to 10 minutes, to read the steady-state load current and reference temperature at the current moment. The polling cycle is used to limit the time interval between two adjacent load condition data acquisitions during the initial operation phase, so that the non-volatile memory can obtain sampled values ​​covering different load current and reference temperature ranges within the initial reference establishment time threshold.

[0035] The steady-state load current is obtained through the current sensing resistor and differential amplifier circuit in the secondary output circuit of the power adapter, and the reference temperature is obtained through the temperature sampling point located inside the power adapter housing, avoiding the main heat-generating components.

[0036] For the signal extraction and conditioning circuits of steady-state load current and reference temperature, those skilled in the art can arrange them in conjunction with conventional power supply design data. The hardware connection method is a well-known technology in the field and will not be described in detail here.

[0037] The benchmark establishment module combines the synchronously read steady-state load current and reference temperature into an array to form operating conditions.

[0038] When extracting steady-state load current and reference temperature, the microcontroller's hardware timer triggers dual-channel injection synchronous conversion of the analog-to-digital converter (ADC), ensuring that the current and temperature signals are synchronized on the time axis, for example, with the sampling time difference controlled within the microsecond range. Operating parameters are used to characterize the load stress and temperature conditions of the power adapter at a specific sampling moment.

[0039] S12, synchronously reads multi-parameter domain measurement data.

[0040] In a preferred embodiment, during the initial operation phase, the benchmark establishment module controls the parameter acquisition circuit inside the power adapter to operate; wherein, for conventional parameters that need to be extracted continuously in a periodic manner, synchronous sampling is triggered within the same polling cycle in which the operating condition quantity is acquired; for parameter channels that need to be formed by event conditions or active test conditions, the benchmark establishment module uses discrete operating events that occur naturally or are actively triggered during the initial operation phase to perform asynchronous sampling, and associates the sampled values ​​with the operating condition quantity at the time the corresponding event occurs.

[0041] The parameter acquisition circuit includes an output quality domain sampling branch, an efficiency thermal domain sampling branch, and a safety protection domain sampling branch. The output quality domain sampling branch is used to acquire signals corresponding to the output ripple peak value, voltage recovery time, and start-up stabilization time. The efficiency thermal domain sampling branch is used to acquire signals corresponding to the real-time power conversion efficiency, key hot spot temperature, and device temperature rise rate. The safety protection domain sampling branch is used to acquire signals corresponding to the protection action threshold drift, protection response time, and trip coil impedance.

[0042] The benchmark establishment module synchronously acquires the original sampling signals of each key node in the above three parameter domains, and after processing by analog-to-digital converter or microcontroller internal timer, or by extracting temperature data of two points within a short observation time window and performing time difference operation, the corresponding sampling values ​​are formed.

[0043] For parameter channels that require event conditions or active test conditions to form, such as voltage recovery time, start-up stabilization time, protection response time, protection action threshold drift, and trip coil impedance, the reference establishment module obtains the corresponding initial sampled values ​​during the initial operation phase using power-on events, load jump events, leakage current protection action events, or active excitation test events. For example, the initial sampled value of start-up stabilization time is obtained through power-on events, the initial sampled value of voltage recovery time is obtained through load jump events, and the initial sampled values ​​of protection response time and trip coil impedance are obtained through active excitation test events.

[0044] Specifically, the original sampling signals include, for example, the output voltage AC ripple amplitude representing the output quality domain, the voltage signal corresponding to the primary-to-secondary power ratio representing the efficiency thermal domain, and the sampling signal corresponding to the protection action threshold drift representing the safety protection domain.

[0045] In this embodiment, the sampled values ​​have the same channel type as the measured values ​​during subsequent normal operation, and are used to form the reference values ​​corresponding to each parameter channel. The sampled values ​​are used to reflect the electrical and thermal state of the power adapter during the initial operation phase.

[0046] S13, Establish working condition grid and data mapping.

[0047] The reference establishment module divides the non-volatile memory of the microcontroller into continuous storage areas and establishes a multi-condition reference matrix indexed by the load current range and the reference temperature range.

[0048] In practical implementation, the reference establishment module divides the full-load output current range of the power adapter into multiple consecutive load current intervals at equal intervals, and divides the operating temperature range into multiple consecutive reference temperature intervals at equal intervals. For example, when the full-load output current of the power adapter is 5A and the number of load current intervals is 10, the output current from 0 to 5A is divided into 10 intervals, and the span of each load current interval is 0.5A.

[0049] To avoid mapping anomalies caused by data exceeding limits, the benchmark establishment module truncates extreme operating condition data to the outermost range for mapping processing when the data exceeds the normal operating limits due to transient interference.

[0050] The benchmark establishment module extracts the operating condition quantities generated in S11, compares the steady-state load current included in the operating condition quantities with the upper and lower boundary values ​​of the load current range, and compares the reference temperature included in the operating condition quantities with the upper and lower boundary values ​​of the reference temperature range. The upper and lower boundary values ​​of the load current range are obtained by equally dividing the full-load output current range of the power adapter, and the upper and lower boundary values ​​of the reference temperature range are obtained by equally dividing the operating temperature range. The benchmark establishment module determines the intersection of the load current range and the reference temperature range boundary conditions as the grid coordinate position of the operating condition quantity in the multi-operating condition benchmark matrix, and assigns the sampled values ​​synchronously acquired in S12 to the corresponding operating condition range according to the corresponding operating condition quantity.

[0051] S14, Filtering, Noise Reduction and Matrix Storage.

[0052] To reduce random discrete fluctuations and high-frequency measurement noise introduced by a single sampling, the benchmark establishment module performs mean filtering on historical cumulative sample values ​​falling within the same operating condition range.

[0053] Mean filtering smooths random errors using an arithmetic mean, resulting in a baseline value that characterizes the initial operating state.

[0054] Let it fall on the . The load current range and the first The total number of valid sampled data within a specific operating condition interval formed by the intersection of several reference temperature intervals is: , No. The first trigger to collect data The sampled values ​​of each parameter channel are denoted as ,in, In a preferred embodiment Then, under a specific operating condition range, the first... The baseline value corresponding to each parameter channel The calculation is as follows: ; In the formula, The total number of valid sampled data; For the first The first trigger to collect data The sampled values ​​of each parameter channel.

[0055] If the set initial baseline establishment time threshold is reached, the total number of valid sampled data within a certain operating condition interval is... If the value is zero, the benchmark establishment module will perform linear interpolation calculations using existing benchmark values ​​in adjacent working condition intervals to fill in the missing data, thereby ensuring the data completeness of the multi-working condition benchmark matrix.

[0056] When no existing reference value exists in adjacent operating condition intervals, the reference establishment module calls the reference value of the nearest valid operating condition interval to the corresponding operating condition interval to fill in the gap; when no reference value of any valid operating condition interval exists in the multi-operating condition reference matrix, the reference establishment module calls the factory calibration reference value stored in the non-volatile memory corresponding to each parameter channel to fill in the gap.

[0057] The benchmark establishment module obtains the corresponding benchmark value through algebraic summation and arithmetic average operation. The benchmark value is used to represent the initial health reference value of each parameter channel under specific working condition boundary conditions.

[0058] The reference establishment module establishes pointer associations between the reference values ​​calculated by each parameter channel and the corresponding load current range and reference temperature range, and sequentially writes the associated data blocks into the corresponding storage address of the multi-condition reference matrix, thereby constructing the multi-condition reference matrix.

[0059] The multi-condition reference matrix constitutes a discretized reference data set, which is used to provide data alignment and state comparison basis for the subsequent degradation index calculation under random triggered conditions after the power adapter finishes the initial reference establishment phase and enters normal life cycle operation.

[0060] Reference Figure 4 The event-triggered sampling module executes step S2, which includes the following steps: S21, Classification monitoring and trigger response of discrete operating events.

[0061] In this embodiment, the event-triggered sampling module establishes a signal connection with the running event monitoring circuit and hardware timer inside the microcontroller. The running event monitoring circuit continuously performs edge detection on the operating parameters of the power adapter to capture discontinuous changes in operating conditions.

[0062] When an internal bus voltage rise occurs due to AC power connection, the operation event monitoring circuit detects and captures the power-on event. The operation event monitoring circuit obtains the load current values ​​before and after the event through the secondary-side current sensing resistor and differential amplifier circuit, and records the steady-state load current before the event as... The steady-state load current after the event is denoted as And calculate the load current jump amplitude: ; In the formula, This represents the load current jump amplitude. This refers to the steady-state load current before the event occurred. This represents the steady-state load current after the event occurs.

[0063] when When the current is greater than 500mA, the event monitoring circuit detects and captures a load jump event; when When the current is less than or equal to 500mA, the microcontroller treats it as a normal steady-state load current fluctuation and does not trigger event logging.

[0064] The microcontroller has an internally powered low-power timer. When the low-power timer accumulates a sampling period of 12 to 48 hours, it triggers a timer cycle completion event. The sampling period is used to define the periodic sampling interval of the efficiency thermal domain during continuous operation.

[0065] The event monitoring circuit also captures leakage protection action events by monitoring the flip-flop state of the output pin of the leakage detection chip, or determines the execution of active excitation test events when the microcontroller actively sends a tiny test pulse to the external safety detection circuit.

[0066] The pulse width of the micro test pulse is 1ms to 10ms, and the test current corresponding to the micro test pulse is 5% to 20% of the rated trigger current of the protection action.

[0067] When any discrete operating event that triggers parameter acquisition occurs, the event-triggered sampling module generates a hardware interrupt response, suspends the regular background task, and prepares to execute parameter acquisition for a specific parameter domain. These discrete operating events include power-on events, load jump events, timer cycle completion events, leakage current protection activation events, and active excitation test events.

[0068] S22, Classification and extraction of measured values ​​in the parameter domain and parameter encapsulation.

[0069] After a hardware interrupt response is generated, the event-triggered sampling module acquires the measured value of at least one parameter domain among the output quality domain, efficiency thermal domain, and safety protection domain activated by the current discrete running event.

[0070] Because the degradation characteristics of different parameter domains differ in time scale and triggering conditions, the event-triggered sampling module executes the corresponding sampling activation rules: When a power-on event or load jump event is captured, the event triggers the sampling module to activate the sampling channel of the output quality domain.

[0071] The output quality domain is used to characterize the dynamic performance of the internal filter network and feedback loop. As internal filter electrolytic capacitors and other components age, the equivalent series resistance increases, and the output voltage ripple and transient recovery characteristics change accordingly.

[0072] The event-triggered sampling module obtains the maximum amplitude of the secondary AC ripple component through a hardware peak hold circuit and an analog-to-digital converter, and uses the maximum amplitude as the output ripple peak value. In a load jump event, a high-frequency counter records the time required from the occurrence of the load jump to the output voltage re-entering the steady-state error band, and uses the recorded time as the voltage recovery time. In a power-on event, the time required from the occurrence of the power-on action to the first establishment of the output voltage within the steady-state error band is recorded, and the recorded time is used as the start-up stabilization time. In a preferred embodiment, the steady-state error band is within the range of ±1% to ±5% of the nominal output voltage.

[0073] When the trigger timer reaches a periodic event, the event-triggered sampling module activates the sampling channel of the efficiency thermal domain. The efficiency thermal domain is used to characterize power conversion efficiency and thermal state changes. The event-triggered sampling module calculates the real-time power conversion efficiency by synchronously acquiring the primary input power and the secondary output power, and reads the temperature of key hot spots on the surface of the heating device through a resistance divider network of the thermistors.

[0074] The event-triggered sampling module further calculates the slope of the key hot spot temperature change per unit time to obtain the device temperature rise rate.

[0075] The event-triggered sampling module acquires the primary input power after input rectification and before output filtering, respectively. With secondary output power And obtain the real-time power conversion efficiency (i.e., ...) through division operation. ).

[0076] Set the starting time within the short-time observation window The temperature obtained is At the end time The temperature obtained is The device temperature rise rate is then calculated using the following difference equation: ; In the formula, This refers to the device's temperature rise rate, typically expressed in degrees Celsius per second. and These are the start and end times, respectively. To filter out high-frequency thermal noise interference, the time difference between the two is 1 to 5 seconds. and The temperature in degrees Celsius obtained for the corresponding node.

[0077] When a leakage current protection action event or an active excitation test event is triggered, the event trigger sampling module activates the sampling channel of the safety protection domain. The safety protection domain is used to characterize the protection action threshold, protection response time, and trip execution status.

[0078] The event-triggered sampling module obtains the protection action threshold drift by reading the absolute difference between the leakage current when the leakage protection is actually triggered and the factory-nominated safety threshold; it records the action delay from the issuance of the trigger signal to the physical disconnection of the trip relay using a timer, and uses this delay as the protection response time; and it obtains the trip coil impedance by applying a constant test current and reading the voltage value across the trip coil. In a preferred embodiment, the constant test current is 5% to 20% of the rated drive current of the trip coil.

[0079] The event-triggered sampling module will output ripple peak value, voltage recovery time, start-up stabilization time, real-time power conversion efficiency, critical hot spot temperature, device temperature rise rate, protection action threshold drift, protection response time, trip coil impedance, and form measured values ​​according to their respective parameter domains.

[0080] For the impedance matching, filtering, and amplification conditioning of the above-mentioned parameter signals, those skilled in the art can use conventional analog front-end circuits for arrangement. The circuit structure is a well-known technology in the field and will not be described in detail here.

[0081] The parameter domain is a set of parameters divided according to different performance indicators of the power adapter. The measured values ​​are used to quantify the degree of performance degradation of the power adapter within the corresponding parameter domain.

[0082] S23, Operating condition lockout and timing calibration.

[0083] To ensure that discrete sampling data corresponds to the operating conditions at the time of the event, the event-triggered sampling module synchronously acquires the operating conditions at the time of the event within the sampling processing cycle after a single discrete operating event is triggered. The sampling processing cycle is the processing cycle in which the event-triggered sampling module completes the acquisition of the measured values ​​of the activated parameter domain and the latching of the operating conditions after responding to a single discrete operating event.

[0084] In practice, the microcontroller latches the steady-state load current and reference temperature after the transient fluctuations subside through an analog-to-digital converter, and combines the steady-state load current and reference temperature into operating parameters.

[0085] To prevent secondary events from overwriting register data during data conversion, the microcontroller triggers the hardware latch control signal of the data register when responding to a hardware interrupt, thereby reducing concurrent interference.

[0086] The event-triggered sampling module calls the microcontroller's internal cumulative running timer to extract the total power-on time of the current device when the current parameter acquisition is completed, and records the domain timestamp corresponding to the activated parameter domain, synchronously generating an activated parameter domain marker. The total power-on time of the current device is the cumulative power-on duration output by the cumulative running timer when the current parameter acquisition is completed, in hours. The domain timestamp is used to identify the position of a single discrete sampling event on the device power-on time axis and is used for subsequent evaluation of the timeliness of the acquired data. The activated parameter domain marker is used to identify the parameter domain actually activated in this discrete running event. The actually activated parameter domain includes one or more of the output quality domain, efficiency thermal domain, and safety protection domain, so that subsequent modules can distinguish the update status of the data in this cycle.

[0087] S24, cross-module data synchronization and transmission.

[0088] The event-triggered sampling module writes the categorized and encapsulated measured values, the corresponding bound operating conditions, the domain time stamp identifying the sampling time, and the activated parameter field flags used to identify which parameter fields were activated in this event into the microcontroller's internal shared memory area. The internal shared memory area uses a First-In-First-Out (FIFO) queue for buffering to adapt to subsequent data processing.

[0089] The event-triggered sampling module transmits the acquired measured values, operating conditions, and activated parameter domain markers to the data normalization mapping module and the health fusion module by triggering the data ready flag, thereby completing the asynchronous data acquisition and distribution process based on the event-triggered sampling mechanism.

[0090] Reference Figure 5 The data normalization mapping module executes step S3, which includes the following steps: S31, storage addressing of the reference matrix and acquisition of reference value interpolation.

[0091] In this embodiment, the data normalization mapping module receives the data transmitted by the event-triggered sampling module and sends a read instruction to the microcontroller's non-volatile memory through the memory read mechanism.

[0092] The data normalization mapping module uses the starting physical address of the multi-condition reference matrix as the base address and calculates the memory offset using the acquired condition parameters, including steady-state load current and reference temperature. Specifically, the data normalization mapping module first determines the corresponding row number based on the load current range into which the steady-state load current falls, and determines the corresponding column number based on the reference temperature range into which the reference temperature falls. Then, according to the row and column storage order of the multi-condition reference matrix in non-volatile memory, the module converts the row number, column number, and the data block length corresponding to a single condition range to obtain the memory offset relative to the starting physical address.

[0093] The data normalization mapping module uses the memory offset as the addressing increment of the physical address, adds it to the base address to form a non-volatile memory access address pointer, and calls the multi-condition benchmark matrix generated by the benchmark establishment module from the non-volatile memory through the non-volatile memory access address pointer.

[0094] Since naturally triggered operating conditions are usually located at non-integer nodes of the discrete grid, in order to avoid quantization truncation errors caused by directly rounding to the nearest integer, the data normalization mapping module locks the four adjacent calibration grid points around the operating condition based on the actual floating-point coordinates of the operating condition in the two-dimensional operating condition coordinates.

[0095] Subsequently, the data normalization mapping module uses a bilinear interpolation algorithm to calculate the proportion of the distance between the actual coordinate point and four adjacent grid points in the X-axis (load current direction) and Y-axis (reference temperature direction) as the allocation weight, and performs two-dimensional linear interpolation calculations in the horizontal and vertical directions in sequence to obtain the reference value corresponding to the operating condition.

[0096] If the currently captured operating condition quantity exceeds the matrix boundary range formed by the load current range and the reference temperature range in the multi-operating condition reference matrix, the data normalization mapping module will restrict the coordinates corresponding to the operating condition quantity to the outermost effective grid edge to avoid addressing out of bounds and causing addressing anomalies.

[0097] Through the above address pointer operations and interpolation, the data normalization mapping module can obtain the reference value under the corresponding operating conditions under random load and temperature conditions.

[0098] S32, Dimensionless Comparison and Calculation of Individual Degradation Index.

[0099] The single degradation index is a dimensionless scalar with a value between 0 and 1, where 0 indicates close to the initial healthy state and 1 indicates close to the end of life.

[0100] When the corresponding operating condition interval in the multi-operating condition reference matrix is ​​filled to form a reference value, the filled reference value is used as the normalized reference. The normalized reference is the reference value obtained from the multi-operating condition reference matrix based on the current operating condition quantity.

[0101] Different parameters have different dimensions and scales of change. For example, the ripple voltage is in the millivolt range in the output quality domain, the response time is in the millisecond range in the safety protection domain, or the temperature is in the Celsius range in the efficiency thermal domain. The data normalization mapping module converts different parameters to a unified relative scale through a single degradation index.

[0102] The data normalization mapping module performs algebraic operations on the extracted measured values, baseline values, and lifetime end values ​​pre-programmed into non-volatile memory.

[0103] Let the first The measured values ​​extracted from each parameter channel are The corresponding baseline value is , No. The lifetime endpoint value corresponding to each parameter channel is ,in .

[0104] When the parameter channel is one whose value increases with the degree of degradation, the data normalization mapping module calculates the intermediate degradation value according to the following formula: ; When the parameter channel is one whose value decreases as the degree of degradation increases, the data normalization mapping module calculates the intermediate degradation value according to the following formula: ; In the formula, The numerator represents the intermediate degradation value; the numerator represents the deviation of the current parameter from the initial healthy state, and the denominator represents the allowable range of change of the parameter from the baseline value to the end-of-life value.

[0105] In order to avoid certain parameter channels and When numerical values ​​overlap or are close to each other, resulting in a denominator of zero, a degenerate floating-point protection value is typically added to the denominator term. , .

[0106] Subsequently, the data normalization mapping module obtains the individual degradation index in the following manner: ; In the formula, This is a single degradation index.

[0107] Life end value Each parameter channel is stored in a non-volatile memory and is called by the data normalization mapping module when calculating a single degradation index.

[0108] Among them, life end value These are the channel boundary values ​​when the corresponding parameter channel reaches the allowable operating boundary. This includes output ripple peak value, voltage recovery time, start-up stabilization time, critical hot spot temperature, device temperature rise rate, protection action threshold drift, protection response time, trip coil impedance, and end-of-life value. greater than the corresponding benchmark value The upper boundary value; for real-time power conversion efficiency, the lifetime endpoint value. Less than the corresponding benchmark value The lower boundary value.

[0109] The lifetime endpoint value of the output ripple peak value is taken as the upper limit of the rated output ripple allowable of the power adapter; the lifetime endpoint value of the voltage recovery time is taken as the longest time allowed for the output voltage to recover to the steady-state error band; the lifetime endpoint value of the start-up settling time is taken as the longest time allowed for the output voltage to establish up to the steady-state error band; the lifetime endpoint value of the real-time power conversion efficiency is taken as the rated minimum conversion efficiency of the power adapter; the lifetime endpoint value of the critical hot spot temperature is taken as the upper limit of the allowable operating temperature of the corresponding hot spot device; the lifetime endpoint value of the device temperature rise rate is taken as the upper limit of the allowable temperature rise rate of the corresponding hot spot device; the lifetime endpoint value of the protection action threshold drift is taken as the upper limit of the allowable drift of the protection action; the lifetime endpoint value of the protection response time is taken as the upper limit of the allowable response time of the protection action; the lifetime endpoint value of the trip coil impedance is taken as the upper limit of the allowable impedance deviation of the trip coil.

[0110] The data normalization mapping module iterates through the currently activated sampled parameter channels and calculates the corresponding individual degradation index for each one.

[0111] For parameter channels within the activated parameter domain that did not obtain valid data in this discrete running event, the data normalization mapping module calls the most recently valid single-item degradation index of the parameter channel to participate in the calculation of subsequent domain degradation values; if the parameter channel does not have a most recently valid single-item degradation index, then the single-item degradation index of the parameter channel is set to 0.

[0112] The above segmentation process sets intermediate degradation values ​​less than 0 to 0 and intermediate degradation values ​​greater than 1 to 1.

[0113] This embodiment preferably uses nine degradation parameters, which, after normalization, generate nine individual degradation indices, specifically: the peak output ripple degradation index in the output quality domain. Voltage recovery time degradation index Start-up stabilization time degradation index Real-time power conversion efficiency degradation index in the efficiency thermal domain Key hotspot temperature degradation index Device temperature rise rate degradation index ; and the degradation index of the protection action threshold drift in the safety protection domain. Protection response time degradation index Tripping coil impedance degradation index .

[0114] S33, fusion of single-item degradation index within the parameter domain and generation of domain degradation value.

[0115] Based on the source of the parameters, the above nine individual degradation indices are grouped into the same parameter domain for calculation.

[0116] The data normalization mapping module performs an arithmetic average of multiple individual degradation indices within the same parameter domain. Specifically, it outputs the degradation value in the quality domain. Efficiency thermal domain degradation value and security protection domain degradation value Generate them according to the following formulas respectively: ; ; ; In the formula, This represents the arithmetic mean operation when there are three input values. This represents the average value obtained by adding the three input values ​​together and then dividing by 3.

[0117] Domain degradation values ​​are used to characterize the overall degree of degradation of a specific parameter domain.

[0118] If, within a specific period, a sensor malfunction causes a loss of valid data for a certain parameter channel, the data normalization mapping module obtains the individual degradation index of the corresponding parameter channel according to the parameter channel processing rules in S32 for those without valid data, and then uses the processed individual degradation index to calculate the arithmetic mean.

[0119] Through the above arithmetic averaging process, the data normalization mapping module can smooth the discrete fluctuations of a single feature parameter and output the domain degradation value.

[0120] S34, Derivation of Spatial Deviation Measurement and Operating Condition Confidence Coefficient.

[0121] The operating condition confidence coefficient is used to quantify the impact of the current test operating condition deviating from the central reference operating condition on the degradation assessment results.

[0122] During operation, the power adapter may experience extreme full load or high temperature conditions. Test data obtained under extreme conditions may be superimposed with transient effects caused by high temperature or high current.

[0123] Therefore, compared with intermediate operating conditions such as normal temperature and half-load, the operating condition confidence coefficients corresponding to test data under extreme operating conditions are lower. To quantify the degree of deviation of operating conditions, the data normalization mapping module calculates the spatial Euclidean distance between the operating condition quantities and the central reference operating condition in the multi-operating condition reference matrix.

[0124] The central reference operating condition is the operating condition in which both the load current range and the reference temperature range are located in the middle position; when the number of load current ranges or reference temperature ranges is even, the range with the lower boundary value among the two ranges closest to the middle position is taken as the central reference operating condition.

[0125] Let the normalized steady-state load current in the operating parameters be... Reference temperature is The rated typical current corresponding to the center reference operating condition is Rated typical temperature is .

[0126] in, and All values ​​have been normalized according to the full-load output current range of the power adapter. and All values ​​have been normalized according to the operating temperature range of the power adapter, and the normalized values ​​are all between 0 and 1. Spatial Euclidean distance The calculation formula is as follows: ; In the formula, This refers to the normalized steady-state load current in the operating parameters. The reference temperature after normalization in the operating parameters; The rated typical current corresponding to the center reference operating condition; The rated typical temperature corresponding to the central reference operating condition; and These are the preset weights for the current dimension and the temperature dimension, respectively, and both satisfy the following: ; in, The percentage of the impact of the steady-state load current on the deviation of the operating condition. The percentage of the impact of the corresponding reference temperature on the deviation from the operating condition. For power adapters where changes in rated output current have a greater impact on degradation assessment than changes in reference temperature. Greater than In a preferred embodiment, Take 0.6, Take 0.4; for power adapters where the impact of reference temperature change on degradation assessment is greater than that of rated output current change, Greater than In another preferred embodiment, Take 0.3, Take 0.7.

[0127] Subsequently, the data normalization mapping module introduces an attenuation adjustment coefficient to convert the spatial Euclidean distance into a working condition confidence coefficient. The conversion function for the working condition confidence coefficient is as follows: ; In the formula, The confidence coefficient for the operating condition; The spatial Euclidean distance; The attenuation adjustment coefficient, in a preferred embodiment, ranges from 0.5 to 2.0. Specifically, in a preferred embodiment... A value of 1.2 is used to control the rate at which the confidence coefficient for the operating condition decreases with the distance of deviation from the operating condition; when it is necessary to increase the impact of extreme operating condition data on weight reduction, Take a value close to 2.0 when it is necessary to reduce the impact of extreme operating condition data on weight reduction. Take the value closest to 0.5; For the natural constant An exponential function with base 0.

[0128] As can be seen from the transformation function, the more significant the deviation of the working condition from the central reference working condition, the greater the spatial Euclidean distance. The larger the value, the lower the calculated confidence coefficient for the operating condition. The value range is greater than 0 and less than or equal to 1.

[0129] The data normalization mapping module outputs the generated domain degradation values ​​and operating condition confidence coefficients to the health fusion module, and updates the corresponding operating condition confidence coefficients according to the parameter domains they belong to.

[0130] When a certain parameter domain is activated in this discrete running event, the data normalization mapping module will use the operating condition confidence coefficient calculated in this operation. As the working condition confidence coefficient in the corresponding parameter domain Store, where Used to represent the output quality domain, efficiency thermal domain, or safety protection domain.

[0131] Reference Figure 6 The health fusion module executes step S4, which includes the following steps: S41, Calculation of time lag and acquisition of timeliness confidence coefficient.

[0132] In one embodiment, before proceeding with the calculation of time lag, the health fusion module determines the data source for the current period's fusion calculation based on the activated parameter domain marker.

[0133] When this evaluation cycle is triggered by a discrete operating event used to trigger parameter acquisition, for parameter domains already activated in this cycle, the health fusion module uses the domain degradation value, operating condition confidence coefficient, and domain timestamp obtained in this cycle; for parameter domains not activated in this cycle, the health fusion module calls the most recently valid domain degradation value, operating condition confidence coefficient, and domain timestamp for the corresponding parameter domain. If a parameter domain does not have a most recently valid domain degradation value, operating condition confidence coefficient, or domain timestamp, then the domain degradation value of the corresponding parameter domain is set to 0, the operating condition confidence coefficient is set to 1, and the domain timestamp is set to the current total power-on time of the device.

[0134] The degradation state of internal components in a power adapter changes with power-on time. During periods when sampling is not triggered, there may be time differences between historical sampling data and the current operating state. To quantify the impact of the time lag in historical sampling data on the lifetime assessment results, the health fusion module obtains the timeliness confidence coefficient based on the domain timescale.

[0135] The timeliness confidence coefficient is a decay factor ranging from 0 to 1, used to characterize the validity of historical measurement data at the current moment.

[0136] The health fusion module reads the current total power-on time of the device inside the microcontroller and performs a subtraction operation between the current total power-on time and the latest acquired domain timestamp of each parameter domain to obtain the time lag.

[0137] Let the total power-on time of the current device be... The domain timescale of a specific parameter domain is , and All values ​​are expressed in hours. The time lag is calculated as follows: ; In the formula, The time lag; subscript The indexes are used to identify different parameter domains, specifically covering the output quality domain, efficiency thermal domain, and safety protection domain.

[0138] when When less than 0, Set to 0; when When greater than or equal to 0, maintain The calculated value is then used. Subsequently, a time decay constant is introduced to exponentially convert the time lag into a time-response confidence coefficient. The conversion formula for the time-response confidence coefficient is: ; In the formula, This is the confidence coefficient for timeliness; For the first The time decay constant corresponding to each parameter domain ranges from 0.01 per hour to 0.1 per hour, and is used to characterize the rate at which the reliability of the data corresponding to the domain time scale decreases as the residence time increases. For the natural constant It is an exponential function with base 0. As the time lag increases, the calculated timeliness confidence coefficient decreases.

[0139] In a preferred embodiment, the output quality domain, the efficiency thermal domain, and the safety protection domain each correspond to an independent time decay constant. , and .

[0140] Since the time decay constant is used to control the rate at which the confidence coefficient decreases with time lag, in a preferred embodiment, when the degradation state of a certain parameter domain changes faster within a unit power-on time, the corresponding time decay constant is taken as close to 0.1 per hour; when the degradation state of a certain parameter domain changes slower within a unit power-on time, the corresponding time decay constant is taken as close to 0.01 per hour. In a preferred embodiment, the time decay constant of the output quality domain... Taking 0.05 per hour, the time decay constant of the efficiency thermal domain The time decay constant of the safety protection domain is set to 0.08 per hour. Take 0.02 per hour.

[0141] S42, Construction of joint confidence coefficients and reduction of initial weights.

[0142] The health fusion module generates a joint confidence coefficient from the operating condition confidence coefficient and the time lag confidence coefficient. The joint confidence coefficient is used to characterize the reliability of data in a certain parameter domain under the combined effects of operating condition deviation and time lag.

[0143] Let the first The confidence coefficients for the operating conditions in each parameter domain are: The confidence coefficient for timeliness is Then the joint confidence coefficient for the corresponding parameter domain is calculated as follows: ; In the formula, This represents the joint confidence coefficient.

[0144] Based on this, the health fusion module corrects the initial weights based on the joint confidence coefficient.

[0145] At the beginning of each evaluation cycle, the health fusion module assigns initial weights to each parameter field from the dynamic weight allocation table in the microcontroller's non-volatile memory, based on the overall health degradation stage of the power adapter in the previous cycle. If it is the first evaluation cycle, the power adapter is classified as an early degradation stage.

[0146] In this embodiment, the dynamic weight allocation table stores multiple degradation stages and their corresponding initial weight combinations in a tabular format. The health fusion module calls the corresponding initial weight combination based on the degradation stage to which the overall health of the machine belongs in the previous cycle.

[0147] Specifically, the dynamic weight allocation table includes at least three weight combinations for the early, middle, and late stages of degradation. In the early stage of degradation, the initial weight of the output quality domain is 0.40, the initial weight of the efficiency-thermal domain is 0.35, and the initial weight of the safety protection domain is 0.25. In the middle stage of degradation, the initial weight of the output quality domain is 0.30, the initial weight of the efficiency-thermal domain is 0.35, and the initial weight of the safety protection domain is 0.35. In the late stage of degradation, the initial weight of the output quality domain is 0.20, the initial weight of the efficiency-thermal domain is 0.25, and the initial weight of the safety protection domain is 0.55.

[0148] When the overall health of the system in the previous cycle is less than or equal to 0.30, the health fusion module calls the initial weight combination corresponding to the early stage of degradation; when the overall health of the system in the previous cycle is greater than 0.30 and less than or equal to 0.70, the health fusion module calls the initial weight combination corresponding to the middle stage of degradation; when the overall health of the system in the previous cycle is greater than 0.70, the health fusion module calls the initial weight combination corresponding to the late stage of degradation.

[0149] The health fusion module uses the calculated joint confidence coefficient to reduce the initial weights of each parameter domain, resulting in intermediate retained weights.

[0150] Let the first The initial weights for each parameter domain are: The corresponding intermediate retention weight is ,but The calculation is as follows: ; Using the multiplicative correction method described above, if the sampled data of a certain parameter domain is far from the current time or the operating conditions at the time of sampling deviate significantly, the joint confidence coefficient decreases accordingly, and the intermediate retention weight decreases as a result.

[0151] S43, Confidence coefficient compensation redistribution of total weight missing shares.

[0152] After the above weight reduction, the sum of the intermediate retained weights of all parameter domains is usually less than 1, which no longer satisfies the normalization condition. If it is directly used for weighted calculation, it will affect the calculation result of the overall health of the machine.

[0153] To this end, the health fusion module calculates the total weight missing share and redistributes it based on the relative proportion of the current joint confidence coefficient of each parameter domain in the overall confidence coefficient.

[0154] Let the total weight of the missing shares be... The total number of parameter fields is ,in, This corresponds to the output quality domain, efficiency-thermal domain, and safety protection domain. The total weight is then missing a portion. The calculation formula is as follows: ; Due to the The intermediate retention weight of each parameter domain From the corresponding initial weights With joint confidence coefficient Multiplying them together, and with the sum of the initial weights for each parameter domain being 1, and the joint confidence coefficient between 0 and 1, therefore: ; Therefore, the total weight is missing share. Greater than or equal to 0.

[0155] Next, the health fusion module calculates the first... The relative proportion of each parameter domain in the total confidence coefficient of all parameter domains . No. The joint confidence coefficients for the parameter domains are: , No. The joint confidence coefficients for the parameter domains are: ,in The summation index is used to iterate through all parameter fields.

[0156] To prevent computational overflow anomalies caused by the joint confidence coefficients of all parameter domains approaching zero, resulting in a zero denominator, the health fusion module sets a positive floating-point protection value for weight allocation. ,in: .

[0157] When the sum of the joint confidence coefficients of all parameter domains is greater than At that time, the health fusion module will... Dividing the joint confidence coefficient of the nth parameter domain by the sum of the joint confidence coefficients of all parameter domains yields the nth... The relative allocation ratio of each parameter domain .

[0158] When the sum of the joint confidence coefficients of all parameter domains is less than or equal to At that time, the health fusion module will... Initial weights for each parameter domain As the first The relative allocation ratio of each parameter domain Relative distribution ratio The calculation formula is as follows: ; The above segmentation process results in a relative allocation ratio When the joint confidence coefficient is valid, the weights are allocated according to the overall confidence ratio. When the joint confidence coefficient approaches zero, the weights are returned to the initial weight allocation, thus avoiding a zero denominator and maintaining the continuity of the weight allocation process.

[0159] In both of the above cases, the health fusion module will have a missing share of the total weight. and the corresponding relative allocation ratio Multiply and add up to the nth. The intermediate retention weight of each parameter domain , obtained the Correction weights for each parameter domain . No. Correction weights for each parameter domain The calculation formula is as follows: ; Since the sum of the relative allocation ratios of each parameter domain is 1, the sum of the corrected weights of each parameter domain satisfies the normalization condition: .

[0160] The corrected weights constitute a dynamically adjusted set of evaluation weights. The health fusion module adjusts the weight allocation tendency of the life assessment to a high-confidence parameter domain with higher data timeliness and lower operating condition deviation based on the corrected weights, so as to reduce the impact of event-triggered asynchronous sampling on weight allocation.

[0161] S44, corrects the weighted summation and overall health output of the machine.

[0162] The overall health score is used to quantify the current overall degradation level of the power adapter. The higher the value, the higher the degree of degradation of the power adapter. The health score fusion module will adjust the weights and domain degradation values ​​to obtain the overall health score of the entire device.

[0163] Specifically, according to the aforementioned parameter domain division, the health fusion module uses the correction weights corresponding to the output quality domain as the final weights for the output quality domain. The corrected weights corresponding to the efficiency thermal domain are used as the final weights of the efficiency thermal domain. The modified weights corresponding to the security protection domains are used as the final weights of the security protection domains. .

[0164] Let the overall health of the machine in the current period be output as... The specific weighted summation formula is as follows: ; That is equivalent to: ; In the formula, the correction weights satisfy the normalization constraint: All domain degradation values ​​are dimensionless scalars, therefore the overall health of the machine is... Limited to a numerical range of 0 to 1, the closer the value is to 1, the higher the overall degree of degradation.

[0165] After weighted fusion is completed, the health fusion module outputs the calculated overall health of the machine to the life prediction and early warning module, and overwrites and temporarily stores the overall health of the machine in the local cache of the microcontroller as the overall health of the machine in the previous cycle required for the next evaluation cycle.

[0166] The above processing corrects the temporal differences caused by natural operating events through confidence coefficient correction and weight compensation mechanisms, so that the final output health assessment index has the ability to resist random interference.

[0167] Reference Figure 7 The lifespan prediction and early warning module executes step S5, which includes the following steps: S51, Time Series Data Sequence Extraction and Least Squares Linear Regression Fitting.

[0168] In this embodiment, the logic input of the lifespan prediction and early warning module is connected to the output of the health fusion module, and the drive output is connected to the early warning indication circuit outside the power adapter. To form a historical time series reflecting the device degradation rate, the lifespan prediction and early warning module transfers the acquired overall device health status to non-volatile memory.

[0169] During the long-term service life of a power adapter, the physical wear and tear of the internal electrolytic capacitors, magnetic devices, and semiconductor switches usually increases with the power-on time, and in the middle and late stages of its life, it can exhibit approximately linear or accelerated degradation characteristics.

[0170] Based on the above trends, the life prediction and early warning module opens a fixed-length circular queue buffer space in the non-volatile memory, combines the overall health of the whole machine with the current total power-on time of the device into a two-dimensional data pair, writes it sequentially into the circular queue buffer space, and uses a first-in-first-out overwrite method to prevent storage space overflow.

[0171] When the number of accumulated two-dimensional data pairs in the circular queue buffer space is less than the minimum number of samples, the lifetime prediction and early warning module only performs historical data accumulation and storage, and does not trigger the subsequent trend fitting and lifetime prediction process.

[0172] When the number of accumulated two-dimensional data pairs in the circular queue buffer space is greater than or equal to the minimum sample size, the lifespan prediction and early warning module extracts the two-dimensional data pairs to form a time series and performs a first-order linear regression fitting operation. For example, the minimum sample size is a fixed value between 8 and 20 groups.

[0173] Let the total number of valid samples extracted be , No. The time value in a pair of two-dimensional data is The corresponding overall health of the machine is The lifespan prediction and early warning module calculates the degradation trend prediction equation using the least squares method. Degeneracy slope in With initial intercept The specific calculation formula is as follows: ; ; In the formula, The time values ​​in the extracted samples The arithmetic mean, The overall health of the machine corresponding to the extracted samples. The arithmetic mean.

[0174] The degradation trend prediction equation is used to represent the overall health of the machine as a function of the total power-on time. To avoid environmental fluctuations or measurement noise affecting the fitted degradation slope, the following equation is used. Less than or equal to zero, when the degradation slope When the lifespan prediction and early warning module is less than or equal to zero, the degradation slope will be... Set to the lower limit of the degradation slope, where the lower limit of the degradation slope is 1×10. -6 Up to 1×10 per hour -4 In a preferred embodiment, the lower limit of the degradation slope is 1 × 10 per hour. -5 Every hour, the initial intercept is recalculated using the lower limit of the degradation slope. When the program logic calculates the degradation slope When the value is greater than zero, the lifetime prediction and early warning module maintains the degradation slope. The calculated value and the initial intercept The calculated value.

[0175] S52, substitute the safety margin threshold and calculate the remaining lifespan.

[0176] Based on the degradation trend prediction equation generated above, the life prediction and early warning module predicts the remaining lifespan according to the overall health of the machine.

[0177] Considering that the power adapter may no longer meet the rated output ripple or electromagnetic compatibility design specifications before the overall health of the device reaches the theoretical failure boundary 1, the life prediction and early warning module uses a safety margin threshold of less than 1 as the boundary for determining the effective service life of the equipment. As a preferred approach, the safety margin threshold is a fixed value between 0.80 and 0.90, and is less than the theoretical failure boundary 1 of the overall health of the device.

[0178] Let the safety margin threshold be The lifespan prediction and early warning module will Substituting into the degradation trend prediction equation, the theoretical endpoint time when the equipment reaches the safety boundary is calculated through inverse algebraic solution. : ; in, This represents the total power-on time of the current device.

[0179] Subsequently, the lifespan prediction and early warning module calculates the difference between the theoretical endpoint time and the current total power-on time of the equipment to determine the remaining lifespan of the equipment before it reaches the replacement critical point. ; In the formula, This represents the remaining lifespan.

[0180] To avoid negative output values ​​due to recent data fluctuations, when the remaining lifetime duration... When the lifespan prediction and early warning module is less than zero, it will... Setting it to 0 indicates that the equipment has exceeded its safe service life. (The remaining service life is then considered.) When the lifespan prediction and early warning module is greater than or equal to zero, it remains unchanged. The calculated value.

[0181] When the number of two-dimensional data pairs accumulated in the circular queue buffer space is less than the minimum number of samples and the lifetime prediction process is not triggered, the lifetime prediction and early warning module calls the factory nominal design lifetime stored in the non-volatile memory corresponding to the power adapter model, and uses the difference between the factory nominal design lifetime and the current total power-on time of the device as the remaining lifetime duration; when the difference is less than zero, the remaining lifetime duration is set to 0.

[0182] The physical unit of remaining lifetime duration is consistent with the underlying timer reference, and in this embodiment it is represented as available hours; when it is necessary to output in days, the lifetime prediction and early warning module divides the available hours by 24 to generate the corresponding available days.

[0183] S53, hardware output for health status interval determination and graded early warning status.

[0184] The lifespan prediction and early warning module outputs graded early warning status and drives the external early warning indication circuit to output maintenance prompts.

[0185] The graded early warning status is used to classify the continuous degradation of equipment into multiple discrete control levels according to the potential failure risk; maintenance prompts are sent to external operators or host computer systems through light status or level data messages to output the current operating risks of the equipment and suggested maintenance actions.

[0186] In a preferred embodiment, the normal threshold is 0.40, the warning threshold is 0.75, and the safety margin threshold is 0.85, and the three satisfy the condition that the normal threshold is less than the warning threshold and the warning threshold is less than the safety margin threshold.

[0187] Among them, the normal threshold and the warning threshold are used to classify the warning status, and the safety margin threshold is used by the life prediction warning module to calculate the remaining life duration; the safety margin threshold is not used as the direct switching threshold for the indicator light's graded status.

[0188] When the overall health of the device is less than or equal to the normal threshold, the power adapter is considered to be in the normal range.

[0189] The lifespan prediction and early warning module outputs a static low level through a general purpose input / output (GPIO) pin, driving the indicator light to be constantly green. It also sends maintenance prompts containing overall health status, remaining lifespan data, and a good status bit field flag via a communication circuit (such as an RS485 communication interface or a CAN bus communication interface connected to a microcontroller).

[0190] When the overall health of the device is greater than the normal threshold but less than or equal to the warning threshold, the power adapter is determined to have entered the warning zone.

[0191] The lifespan prediction and early warning module outputs a square wave with varying duty cycle through a pulse width modulation timer control pin, driving the indicator light to switch to a yellow flashing state. Simultaneously, through the communication circuit, it sends maintenance prompts, including the overall health status of the machine, the latest calculated remaining lifespan data, and early warning status markers, to the monitoring host computer, prompting maintenance personnel to prepare spare parts or plan downtime maintenance time windows.

[0192] When the overall health of the device exceeds the warning threshold, the power adapter is deemed to have entered the replacement prompt range.

[0193] The lifespan prediction and early warning module outputs a high-level drive signal to set the indicator light to flash red rapidly.

[0194] In embodiments configured with a derating control interface, the life prediction and early warning module simultaneously pulls down the relevant hardware enable pins to perform derating operation protection, and sends maintenance prompts containing overall machine health, remaining lifespan data, and alarm bit field replacement flags via the communication circuit.

[0195] In one embodiment, the lifespan prediction and early warning module is also used to respond to external status query commands. When the communication circuit receives an external status query command and transmits it to the microcontroller, the lifespan prediction and early warning module reads the overall health status, remaining lifespan, and graded early warning status obtained from the most recent assessment, and encapsulates the overall health status, remaining lifespan, and graded early warning status into a status query result, which is then output to an external device through the communication circuit.

[0196] Through the above threshold comparison and signal output, the life prediction and early warning module converts the overall health status of the whole machine into corresponding maintenance prompt actions, which are used for power adapter operating status detection, fault prediction and health management.

[0197] To further illustrate the implementation process and technical effects of this invention, the following description is provided in conjunction with specific application scenarios and experimental data. The specific numerical values ​​and scenario parameters described below are only used to explain the implementation principle of this invention and do not limit the scope of protection of this invention.

[0198] A power adapter life prediction system is deployed in a consumer-grade power adapter with a rated input AC 100 to 240V, 50 to 60Hz, and a rated output DC 12V / 3A. This power adapter internally incorporates an STM32G0 series microcontroller and a leakage current detection chip, with a factory-specified rated trigger current of 5mA for protection actions. This embodiment uses the aforementioned system to detect and evaluate the operating status of the power adapter to achieve fault prediction and health management.

[0199] During the initial operation phase of the power adapter, the reference establishment module acquires the operating conditions formed by the combination of a steady-state load current of 2.0A and a reference temperature of 50°C. Under these operating conditions, the reference establishment module synchronously reads multi-parameter domain measurement data, processes the acquired raw sampled signals to form sampled values, and performs mean filtering on historical cumulative sampled values ​​falling within the corresponding intervals. These values ​​are then written to non-volatile memory to form a multi-operating-condition reference matrix containing reference values. The reference values ​​recorded in each parameter domain specifically include: Output quality range: peak output ripple 50mV, voltage recovery time 10ms, startup settling time 20ms; Efficiency thermal domain: Real-time power conversion efficiency 88%, critical hot spot temperature 55℃, device temperature rise rate 0.5℃ / s; Safety protection domain: protection action threshold drift 0.1mA, protection response time 15ms, trip coil impedance 100Ω.

[0200] During normal equipment operation, the microcontroller continuously captures power-on events, timer cycle completion events, and active excitation test events; all of these events are discrete operational events. The event-triggered sampling module generates a response action, fully activating the sampling channels of the output quality domain, efficiency thermal domain, and safety protection domain, and simultaneously acquiring the operating conditions at the time of the event (steady-state load current 2.0A, reference temperature 50℃). The event-triggered sampling module extracts the total power-on time of the current equipment at the time the current parameter acquisition is completed and records it as the domain timescale corresponding to the three parameter domains, each with a value of 500 hours. The acquired measured values ​​for each parameter domain specifically include: Output quality range: peak output ripple 80mV, voltage recovery time 15ms, startup settling time 25ms; Efficiency thermal domain: Real-time power conversion efficiency 85%, critical hot spot temperature 65℃, device temperature rise rate 0.8℃ / s; Safety protection domain: protection action threshold drift 0.5mA, protection response time 20ms, trip coil impedance 110Ω.

[0201] The data normalization mapping module receives data transmitted by the event-triggered sampling module and calls the lifetime end values ​​of each channel stored in the non-volatile memory, specifically: Output ripple peak value: 120mV; voltage recovery time: 25ms; start-up stabilization time: 40ms; real-time power conversion efficiency: 80%; critical hotspot temperature: 95℃; device temperature rise rate: 1.5℃ / s; protection action threshold drift: 1.0mA; protection response time: 30ms; trip coil impedance: 150Ω. Degradation calculation of positive floating-point protection value is performed. .

[0202] The data normalization mapping module compares the measured values ​​with the benchmark values ​​to obtain the individual degradation index.

[0203] For the real-time power conversion efficiency channel where the value decreases with increasing degradation, the intermediate degradation value is calculated as follows: ; For the remaining eight channels whose values ​​increase with the degree of degradation, calculations were performed according to the corresponding formulas. All intermediate degradation values ​​are between 0 and 1, and are directly used as the corresponding individual degradation indices. The results are as follows: Output quality domain single-term degradation index: peak output ripple Voltage recovery time Startup stabilization time Efficiency thermal domain single-term degradation index: real-time power conversion efficiency Key hot spot temperature 0.250, device temperature rise rate Safety protection domain single-item degradation index: protection action threshold drift amount Protection response time 0.333, trip coil impedance .

[0204] Domain degradation values ​​are generated by arithmetically averaging the individual degradation indices within the same parameter domain: Output quality domain degradation values Efficiency thermal domain degradation value Security protection domain degradation value .

[0205] The data normalization mapping module extracts the rated typical current of 1.5A and the rated typical temperature of 40℃ corresponding to the central reference condition in the multi-condition reference matrix. The rated output of 3A and the upper limit temperature of 80℃ are used as full-scale boundaries for normalization, resulting in a normalized steady-state load current of 0.667, a normalized reference temperature of 0.625, and normalized typical current and temperature of 0.500. The current dimension weight is then set. Temperature dimension weight .

[0206] Calculate spatial Euclidean distance: ; Set attenuation adjustment coefficient Generate operating condition confidence coefficients shared by all parameter domains: ; In the current evaluation period, the health fusion module reads the total power-on time of the current device as 505 hours. The difference between the total power-on time of the current device and the domain timescale (500 hours) of each parameter domain is calculated, and the time lag of each domain is 5 hours.

[0207] Using time decay constants of 0.05 h / h, 0.08 h / h, and 0.02 h / h for the output quality domain, efficiency thermal domain, and safety protection domain, respectively, generate the time-dependent confidence coefficients for each parameter domain: Output quality domain ; Efficiency thermal domain ; Security protection domain .

[0208] Multiply the operating condition confidence coefficient by the time-related confidence coefficient to generate the joint confidence coefficient for each parameter domain: ; ; ; The sum of the joint confidence coefficients for all parameter domains is 1.964.

[0209] Since the overall health of the machine in the previous cycle was less than or equal to 0.30, the health fusion module calls the initial weights corresponding to the early stage of degradation in the dynamic weight allocation table: output quality domain 0.40, efficiency thermal domain 0.35, and safety protection domain 0.25.

[0210] Multiplying the initial weights by the joint confidence coefficients yields the intermediate retained weights: ; ; ; Calculate the total weighted missing shares: ; The corrected weights are obtained by redistributing the data based on the relative proportions of each domain in the total confidence score. ; ; ; The overall health of the system is obtained by weighting the corrected weights with the corresponding domain degradation values: ; The lifespan prediction and early warning module transfers the acquired overall system health status to a circular queue buffer space. At this point, the number of accumulated two-dimensional data pairs in the circular queue buffer space exceeds the minimum sample size. Therefore, the lifespan prediction and early warning module extracts the two-dimensional data pairs to form a time series, and performs a first-order linear regression fitting on the time series using the least squares method to calculate the degradation slope of the degradation trend prediction equation. Initial intercept .

[0211] Using a safety margin threshold of 0.85, calculate the theoretical endpoint time: ; The remaining lifespan is calculated by subtracting the theoretical endpoint time from the total power-on time of the current equipment. ; The lifespan prediction and early warning module predicts the remaining lifespan based on the overall health of the device and outputs a tiered early warning status. Since the overall health score of 0.324 is less than the normal threshold of 0.40, the power adapter is determined to be within the normal range. Subsequently, a static low level is output through the general-purpose input / output pins, driving the indicator light to illuminate steadily green. A maintenance prompt message containing the overall health score, remaining lifespan data, and a "good status" flag is then sent via the communication circuit.

[0212] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result using substantially the same method falls within the scope of the present invention.

Claims

1. A power adapter lifespan prediction system, characterized in that, include: The reference establishment module is configured to acquire operating conditions consisting of steady-state load current and reference temperature during the initial operation phase of the power adapter, and collect corresponding sampled values ​​to form a multi-operating condition reference matrix containing reference values. The event-triggered sampling module is configured to acquire the measured value of at least one parameter domain among the output quality domain, efficiency thermal domain, and safety protection domain activated by the current discrete operating event when the discrete operating event occurs, synchronously acquire the operating condition quantity at the time of the event, and record the domain time stamp corresponding to the activated parameter domain. The data normalization mapping module is configured to obtain the benchmark value from the multi-condition benchmark matrix based on the operating condition quantity, compare the measured value with the benchmark value to obtain a single-item degradation index, generate a domain degradation value from the single-item degradation index in the same parameter domain, and generate an operating condition confidence coefficient based on the degree of deviation of the operating condition quantity from the central benchmark operating condition in the multi-condition benchmark matrix. The health fusion module is configured to obtain a timeliness confidence coefficient based on the domain timescale, generate a joint confidence coefficient by combining the operating condition confidence coefficient and the timeliness confidence coefficient, correct the initial weights based on the joint confidence coefficients to obtain corrected weights, and weight the corrected weights with the domain degradation values ​​to obtain the overall health of the machine. The lifespan prediction and early warning module is configured to predict the remaining lifespan based on the overall health of the machine and output a graded early warning status.

2. The power adapter life prediction system according to claim 1, characterized in that, In the reference establishment module, the step of acquiring operating conditions consisting of steady-state load current and reference temperature during the initial operation phase of the power adapter, and collecting corresponding sampled values ​​to form a multi-operating-condition reference matrix containing reference values ​​includes: Read the steady-state load current and the reference temperature at the current moment, and combine the synchronously read steady-state load current and the reference temperature into an array to form the operating condition quantity; The sampling value is formed by synchronously triggering sampling within the same polling cycle after acquiring the operating condition value, or by asynchronously sampling using discrete operating events that occur during the initial operating phase. The full-load output current range of the power adapter is divided into multiple consecutive load current ranges at equal intervals, and the operating temperature range is divided into multiple consecutive reference temperature ranges at equal intervals. The intersection of the load current range and the reference temperature range is determined as the grid coordinate position, and the sampled value is assigned to the corresponding operating condition range according to the corresponding operating condition quantity. The baseline value is obtained by performing mean filtering on the historical cumulative sample values ​​falling within the same operating condition range, and the baseline value calculated by each parameter channel is written into the corresponding storage address to construct the multi-operating condition baseline matrix.

3. The power adapter life prediction system according to claim 1, characterized in that, In the event-triggered sampling module, the steps of acquiring the measured value of at least one parameter domain among the output quality domain, efficiency thermal domain, and safety protection domain activated by the current discrete operating event when the event occurs, synchronously acquiring the operating condition quantity at the time of the event, and recording the domain timescale corresponding to the activated parameter domain include: Capture the discrete operating events, including power-on events, load jump events, timer cycle completion events, leakage current protection activation events, and active excitation test events; The output quality domain is activated when the power-on event or the load jump event is captured; the efficiency thermal domain is activated when the timer reaches its cycle event; and the safety protection domain is activated when the leakage protection action event or the active excitation test event is triggered. Obtain the measured values ​​corresponding to the activated parameter domain, including the peak output ripple, voltage recovery time, start-up stabilization time, real-time power conversion efficiency, critical hot spot temperature, device temperature rise rate, protection action threshold drift, protection response time, and trip coil impedance. The steady-state load current after the latched transient fluctuation subsides is combined with the reference temperature to form the operating condition quantity. The total power-on time of the current device when the current parameter acquisition is completed is extracted and recorded as the domain time stamp corresponding to the activated parameter domain.

4. The power adapter life prediction system according to claim 3, characterized in that, In the event-triggered sampling module, the step of obtaining the measured value of at least one parameter domain among the output quality domain, efficiency thermal domain, and safety protection domain activated by the current discrete running event further includes: Extract temperature data from two points with a set time difference and perform time difference calculation to calculate the slope of the change in the temperature of the key hot spot within a unit time, so as to obtain the device temperature rise rate as the measured value.

5. The power adapter life prediction system according to claim 1, characterized in that, In the data normalization mapping module, the step of obtaining the benchmark value from the multi-condition benchmark matrix based on the operating condition quantity, and comparing the measured value with the benchmark value to obtain the single degradation index includes: Using the starting physical address of the multi-condition reference matrix as the base address, the memory offset corresponding to the condition quantity is calculated, and the reference value corresponding to the condition quantity is calculated using the bilinear interpolation algorithm. The intermediate degradation value is obtained by performing algebraic operations on the extracted measured value, the reference value, and the end-of-life value. The numerator of the intermediate degradation value is the difference between the measured value and the reference value or the difference between the reference value and the measured value. The denominator of the intermediate degradation value is the difference between the end-of-life value and the reference value or the difference between the reference value and the end-of-life value. The degradation is then calculated by superimposing the degradation values ​​of the two values ​​to obtain the positive floating-point protection value. The intermediate degradation values ​​less than zero are set to zero, and the intermediate degradation values ​​greater than one are set to one, to obtain the single degradation index.

6. The power adapter life prediction system according to claim 5, characterized in that, In the data normalization mapping module, the step of generating domain degradation values ​​from single-item degradation indices within the same parameter domain, and generating working condition confidence coefficients based on the degree of deviation of the working condition quantity from the central reference working condition in the multi-working-condition reference matrix, includes: The domain degradation value is generated by arithmetically averaging multiple individual degradation indices within the same parameter domain. Obtain the rated typical current and rated typical temperature corresponding to the central reference condition in the multi-condition reference matrix; Based on the square of the first difference between the normalized steady-state load current and the rated typical current in the operating condition quantity, and the square of the second difference between the normalized reference temperature and the rated typical temperature in the operating condition quantity, the spatial Euclidean distance between the operating condition quantity and the central reference operating condition is calculated. The negative value of the product of the spatial Euclidean distance and the attenuation adjustment coefficient is used as the exponent of the natural constant to obtain the confidence coefficient of the operating condition.

7. The power adapter life prediction system according to claim 1, characterized in that, In the health fusion module, the step of obtaining the timeliness confidence coefficient based on the domain timescale and generating a joint confidence coefficient from the operating condition confidence coefficient and the timeliness confidence coefficient includes: Read the total power-on time of the current device, and calculate the time lag by subtracting the total power-on time of the current device from the domain time scale; The time lag is multiplied by the time decay constant, and the negative value is used as the exponent of the natural constant to obtain the time confidence coefficient. The joint confidence coefficient is generated by multiplying the operating condition confidence coefficient and the time-related confidence coefficient.

8. The power adapter life prediction system according to claim 7, characterized in that, In the health fusion module, the step of obtaining the corrected weights based on the joint confidence coefficients after correcting the initial weights includes: The initial weights are multiplied by the joint confidence coefficients to obtain the intermediate retention weights; The total weight missing share is obtained by subtracting the value one from the sum of the intermediate retained weights of all the parameter domains. The relative allocation ratio is obtained by dividing the joint confidence coefficient of a specific parameter domain by the sum of the joint confidence coefficients of all parameter domains. The total weight missing share is multiplied by the relative allocation ratio and then added to the intermediate retention weight of the specific parameter domain to obtain the corrected weight of the specific parameter domain.

9. The power adapter life prediction system according to claim 1, characterized in that, In the lifespan prediction and early warning module, the step of predicting the remaining lifespan based on the overall health of the machine and outputting a graded early warning status includes: The overall health status of the machine and the total power-on time of the current device are combined into a two-dimensional data pair. The two-dimensional data pairs with a total number of valid samples greater than or equal to the minimum number of samples are extracted to form a time series. The time series was fitted by linear regression using the least squares method to calculate the degradation slope and initial intercept of the degradation trend prediction equation. When the degradation slope is less than or equal to zero, the degradation slope is set to the lower limit of the degradation slope, and the initial intercept is recalculated using the lower limit of the degradation slope. Substitute a safety margin threshold less than one into the degradation trend prediction equation to calculate the theoretical endpoint time. Subtract the theoretical endpoint time from the current total power-on time of the device to obtain the remaining lifespan.

10. A power adapter lifespan prediction method, applied to the power adapter lifespan prediction system as described in any one of claims 1 to 9, characterized in that, Includes the following steps: During the initial operation phase of the power adapter, the operating conditions consisting of steady-state load current and reference temperature are acquired, and the corresponding sampled values ​​are collected to form a multi-operating condition reference matrix containing reference values. When a discrete operation event occurs, the measured value of at least one parameter domain among the output quality domain, efficiency thermal domain, and safety protection domain activated by the current discrete operation event is obtained, the operating condition quantity at the time of the event is obtained synchronously, and the domain time stamp corresponding to the activated parameter domain is recorded. The reference value is obtained from the multi-condition reference matrix based on the operating condition quantity. The measured value is compared with the reference value to obtain a single degradation index. A domain degradation value is generated from the single degradation index in the same parameter domain. The operating condition confidence coefficient is generated based on the degree of deviation of the operating condition quantity from the central reference operating condition in the multi-condition reference matrix. Based on the domain timescale, the timeliness confidence coefficient is obtained. The operating condition confidence coefficient and the timeliness confidence coefficient are combined to generate a joint confidence coefficient. The initial weights are corrected based on the joint confidence coefficient to obtain the corrected weights. The corrected weights are then weighted with the domain degradation value to obtain the overall health of the machine. Based on the overall health status of the machine, the remaining lifespan is predicted and a graded early warning status is output.

Images