Test control method and system for direct-current power supply

Abstract

The invention discloses a test control method and system for a direct-current power supply, and relates to the field of simulation testing, and the method comprises the steps: constructing a three-dimensional adaptive model to complete the matching calibration of system software and hardware, and achieving the load calibration through the comparison of dynamic load simulation and actual measurement data; in the testing process, a three-stage starting mechanism is adopted to power on step by step, fault protection is configured, and parameter scanning and real-time recording are executed according to a preset mode and a testing point; a double-closed-loop feedback mechanism is introduced, an inner loop adjusts output parameters, and an outer loop predicts power grid disturbance and performs advanced compensation; a safety threshold value is dynamically generated in combination with real-time environment data, linkage protection of approaching early warning and overrun protection is achieved, and a reliable effective data set is formed through fusion processing and double verification of multi-source test data. The system has the advantages that the precision and intelligence of the whole test process are realized through dynamic load simulation, a multi-level safety protection system is constructed through a three-level starting mechanism, and the precision, safety and data value of the test are comprehensively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of analog test, more particularly, to a test control method and system of a direct current power supply. BACKGROUND

[0002] With the continuous growth of demand for high-performance direct current power supply in the fields of industrial equipment, electric vehicles and new energy, higher requirements are put forward for its output accuracy, dynamic response and reliability. In recent years, automation testing and digital control technology have gradually popularized. Through the integration of programmable electronic load, data acquisition module and host computer software, multi-parameter and sequential testing are realized, which significantly improves the efficiency and accuracy.

[0003] Most of the current direct current power supply test methods on the market do not construct a multi-dimensional adaptive model, but rely on manual matching of hardware and software parameters. The calibration process does not combine power rated parameters, test requirements and environmental benchmark data, and is easily affected by temperature and humidity, power grid fluctuations, resulting in test basis deviation. Dynamic load simulation is mostly limited to single load type, lacks real-time calibration mechanism, and load simulation deviation often exceeds ±1%, which cannot truly restore the complex load characteristics such as pulse and nonlinearity in actual application. The start-up process is mostly simplified with self-checking and loading steps, without graded start-up and abnormal node recording, which can easily cause equipment failure during the start-up phase. Feedback regulation mostly uses single-loop control, which can only passively correct output deviation, lacks disturbance prediction compensation based on neural networks, and lags in responding to power grid mutations and load fluctuations. The safety threshold is mostly a fixed value, which is not dynamically adjusted in combination with the test phase and the environment, and the early warning and protection mechanism is single, which poses a safety hazard. Data processing is only simple filtering without hierarchical fusion and double verification, the effective data discrimination ability is weak, the output data lacks standardization and relevance, and it is difficult to provide high-quality support for power performance analysis. SUMMARY

[0004] In order to improve the existing method and system, a test control method and system of a direct current power supply are provided. The method realizes the precision and intelligence of the whole test process through three-dimensional adaptive calibration, dynamic load simulation and double-loop feedback regulation. A multi-level safety protection system is constructed through a three-stage start-up mechanism and dynamic safety threshold, which comprehensively improves the precision, safety and data value of the test.

[0005] To achieve the above purpose, the technical solution adopted by the present application is: A test control method of a direct current power supply, comprising: Constructing a three-dimensional adaptive model based on the rated parameters of the direct current power supply to be tested, test requirements and test environment benchmark parameters, matching the hardware configuration and software parameters of the test system and performing adaptive calibration; Simulate dynamic load characteristics based on the actual application scenario of the to-be-tested direct-current power supply, compare the actual absorbed voltage and current data of the load with the preset load parameters, and calibrate the load based on the comparison result; Start the test through a three-stage starting mechanism. The first stage is self-checking of the control module, the second stage is no-load starting of the to-be-tested direct-current power supply, and the third stage is gradual access of the load. In case of abnormality during the starting process, stop the machine and record the abnormal node. Generate multiple mode test sequences based on preconfigured test requirements. In each mode, perform parameter scanning test according to the preset test point distribution. During the test process, record the power supply parameters in real time; Construct a double-closed-loop feedback regulation mechanism. The inner loop compares the collected power supply parameters with the preset target value to generate a regulation signal. The outer loop monitors the power grid parameters and establishes a disturbance influence model based on historical test data to predict the influence trend of the disturbance on the power supply output parameters and output a compensation signal in advance. Generate safety boundary thresholds based on the rated parameters of the to-be-tested power supply, the test stage, and real-time environmental data. Compare the collected power supply parameters with the dynamic safety threshold in real time. When the monitored value approaches and exceeds the dynamic safety threshold, trigger the early warning mechanism and start the interlock protection, respectively. Collect multi-source data during the test process, process the data using a hierarchical fusion algorithm, and determine the effectiveness of the data through a double-checking mechanism to obtain an effective data set.

[0006] Preferably, a three-dimensional adaptation model is constructed based on the rated parameters of the to-be-tested direct-current power supply, test requirements, and test environment reference parameters to match and adaptively calibrate the hardware configuration and software parameters of the test system. Specifically, it includes: The rated parameters of the to-be-tested direct-current power supply include output voltage, current range, power, and ripple rejection ratio design values. The test requirements include steady-state accuracy, dynamic response, overload capacity, and environmental adaptability test. Real-time collection of environmental reference parameters through an integrated temperature and humidity sensor, including temperature, humidity, and power grid input voltage fluctuation range. Based on the collected rated parameters of the power supply, test requirements, and environmental reference parameters, a three-dimensional adaptation model is built to establish the mapping relationship between the parameters. The three-dimensional adaptation model is used to match the hardware and software parameters of the test system, and the matched sensors and load modules are pre-calibrated.

[0007] Preferably, simulate dynamic load characteristics based on the actual application scenario of the to-be-tested direct-current power supply, compare the actual absorbed voltage and current data of the load with the preset load parameters, and calibrate the load based on the comparison result. Specifically, it includes: Based on the specific application scenario of the to-be-tested direct-current power supply, dynamic load characteristics are simulated to generate various dynamic loads, including pulse load, nonlinear load, step load, and random fluctuation load. The actual voltage and current data absorbed by the load are continuously collected and compared with the preset load parameters. If the deviation exceeds ±1%, the amplitude and phase of the load driving signal are adjusted, and the nonlinear error and temperature drift of the load compensation module are compensated to correct the load simulation deviation.

[0008] Preferably, the test is started through a three-stage starting mechanism. The first stage is a control module self-check, the second stage is a no-load start of the to-be-tested direct-current power supply, and the third stage is a gradual load connection. Based on abnormal conditions occurring during the starting process, a shutdown protection is triggered and the specific node of the abnormality is recorded. Specifically, it includes: The first stage of control module self-check is performed to verify the connectivity between modules and check the working status of each module to ensure that there is no hardware failure or communication interruption problem; The second stage of no-load start and monitoring is performed. The to-be-tested direct-current power supply is controlled to enter the no-load start mode, the output voltage of the power supply is increased from 0V to the test starting voltage at a preset slope, and the ripple voltage and noise voltage of the power supply output under no load are monitored in real time to determine whether there is no-load current leakage; The third stage of graded loading is performed. After the no-load start is normal, the load is gradually connected to the power supply at a load ratio of 10%, 30%, 50%, 80%, and 100%, and the fluctuation amplitude of the power supply output parameters is continuously monitored; If the ripple voltage exceeds 50% of the design threshold, there is current leakage, or the parameter fluctuation exceeds ±3% during the graded loading, the shutdown protection mechanism is immediately triggered, and the node data of the abnormality is recorded.

[0009] Preferably, the multiple-mode test sequence is generated based on the preconfigured test requirements. In each mode, parameter scanning test is performed according to the preset test point distribution. During the test process, the power supply parameters are recorded in real time. Based on the preconfigured test requirements, a test sequence including constant voltage mode, constant current mode, constant power mode, and mixed mode is generated; For each test mode, test points are set, including rated value, limit value, critical value, and random sampling value. The limit value test uses a step-by-step approximation method, and the increment / decrement amplitude is 5% of the rated value; Based on the set test sequence and test points, parameter scanning test is performed. During the test process, the output voltage accuracy, output current accuracy, load regulation, source effect regulation, dynamic response time, and output ripple peak-to-peak value parameters of the power supply are monitored in real time.

[0010] Preferably, the double closed-loop feedback regulation mechanism is constructed, the inner loop compares the collected power parameters with the preset target value to generate a regulation signal, and the outer loop monitors the power grid parameters, establishes a disturbance influence model based on historical test data, predicts the influence trend of the disturbance on the power output parameters, and outputs a compensation signal in advance, specifically including: A double closed-loop feedback regulation mechanism is constructed, the inner loop is an output parameter feedback, and the outer loop is a disturbance prediction compensation; The inner loop compares the collected actual data with the preset target value to generate a regulation signal, and controls the PWM drive module of the power supply to correct the output deviation through the regulation signal; The outer loop collects power input voltage fluctuation, environmental temperature change, and load mutation data, establishes a disturbance influence model based on an LSTM neural network model according to historical test data, predicts the influence trend of the disturbance on the power output parameters, and outputs a compensation signal in advance according to the prediction result; When a sudden disturbance of grid voltage sag or load mutation amplitude exceeding the threshold value is detected, an emergency regulation mode is started, the feedback regulation period is shortened, and a pre-stored disturbance compensation strategy library is called to offset the influence of the sudden disturbance.

[0011] Preferably, the safety boundary threshold is generated based on the rated parameters of the power supply to be tested, the test stage, and the real-time environmental data, and the real-time collected power parameters are compared with the dynamic safety threshold in real time. When the monitored value approaches and exceeds the dynamic safety threshold, the early warning mechanism is triggered and the interlock protection is started, specifically including: The safety boundary threshold is generated based on the rated parameters of the power supply to be tested, the current test stage, and the real-time environmental data, including the maximum allowed output voltage, the maximum allowed output current, the maximum allowed temperature rise, and the minimum allowed insulation resistance; The output end parameters, shell temperature, internal component temperature, and insulation resistance value of the power supply are continuously collected, and the collected real-time data are compared with the dynamic safety boundary threshold in real time. If the monitored value reaches 90% of the threshold, the early warning mechanism is triggered, and the test load is reduced and the test parameters are adjusted; If the monitored value exceeds the safety boundary threshold, the shutdown operation is performed in the gradient of current reduction, voltage reduction, and power-off, and the electrical connection between the load and the power supply is cut off.

[0012] Preferably, the multi-source data in the test process is collected, a hierarchical fusion algorithm is used for data processing, and a double verification mechanism is used for validity determination of the data to obtain an effective data set, specifically including: All kinds of data in the test process are summarized, including power output parameter data, load feedback data, environmental monitoring data, disturbance record data, and safety protection data, and a hierarchical fusion algorithm is used for data processing; Through data cleaning in the first step of hierarchical fusion, abnormal values and repeated records in the data are identified and removed, and invalid data interference is removed; By analyzing the correlation between different data types through the second step of hierarchical fusion, a corresponding mapping between power supply output parameters and environmental changes, load status, and disturbance factors is constructed. The effectiveness verification of the third step of the hierarchical integration verifies the consistency of the values ​​of voltage, current, and power, as well as the timing matching of the dynamic response curve with load changes. The key test points of rated values ​​and extreme values ​​are repeatedly tested, and the verified data are integrated to form a standardized and effective dataset.

[0013] Furthermore, a test and control system for a DC power supply is proposed, comprising: 3D Adaptation and Calibration Module: Based on power specifications, test requirements, and environmental parameters, a model is built to complete the automatic matching and initial calibration of the test system's hardware and software; Dynamic load simulation module: Simulates various dynamic load characteristics under real application scenarios and performs online calibration of load simulation accuracy through real-time data feedback; The three-level safe startup module controls the test startup process in the order of system self-test - no-load startup - graded loading, and performs protection and recording in case of abnormality; Multi-mode test sequence execution module: Automatically generates and executes parameter scanning test sequences containing multiple working modes and test points based on preset test requirements; Dual closed-loop control module: The inner loop quickly corrects the output deviation, while the outer loop predictively compensates for external disturbances, thus maintaining the stability and accuracy of the testing process. Safety boundary protection module: Calculates and updates safety thresholds in real time, and ensures test safety through multi-level early warning and gradient interlocking protection mechanisms; Multi-source data fusion processing module: collects data from the entire testing process, cleans it, performs correlation analysis and double verification, and finally outputs a standardized and valid dataset; Processor: The processor is used to handle the calculation process of each formula and the construction calculation process of each model.

[0014] Compared with the prior art, the advantages of the present invention are: By constructing a 3D adaptation model and combining power supply rated parameters, test requirements, and environmental data, precise matching and calibration of hardware and software are achieved, laying a reliable foundation for testing. Dynamic load simulation can reproduce various load characteristics of real application scenarios, and deviations are controlled within ±1% through real-time calibration, improving scenario adaptability. A three-level startup mechanism with layer-by-layer verification avoids abnormal risks during the startup phase, and multi-mode test sequences cover all-dimensional test points, making parameter scanning more comprehensive. The innovative dual-loop feedback regulation, which integrates real-time inner-loop correction and LSTM-based outer-loop disturbance prediction compensation, can also cope with sudden disturbances and significantly improve test stability. Dynamic safety thresholds combined with multi-level early warning and gradient protection ensure test safety from the source. In addition, the layered fusion algorithm and dual verification mechanism perform refined processing of multi-source data, eliminating invalid data and verifying correlations. The final standardized dataset provides high-quality support for power supply performance analysis, achieving a synergistic improvement in test efficiency, accuracy, and safety. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the method proposed in this invention; Figure 2 This is a schematic diagram of the three-dimensional adaptation model proposed in this invention; Figure 3 This is a schematic diagram of the load calibration proposed in this invention; Figure 4 This is a schematic diagram of the three-level startup mechanism proposed in this invention; Figure 5 This is a schematic diagram illustrating the generation of multi-mode test sequences proposed in this invention; Figure 6 This is a schematic diagram of the dual closed-loop feedback regulation mechanism proposed in this invention; Figure 7 This is a schematic diagram illustrating the generation of the safety boundary threshold proposed in this invention; Figure 8 This is a schematic diagram illustrating the method for obtaining a valid dataset as proposed in this invention. Detailed Implementation

[0016] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0017] A test and control system for a DC power supply, comprising: 3D Adaptation and Calibration Module: Based on power specifications, test requirements, and environmental parameters, a model is built to complete the automatic matching and initial calibration of the test system's hardware and software; Dynamic load simulation module: Simulates various dynamic load characteristics under real application scenarios and performs online calibration of load simulation accuracy through real-time data feedback; The three-level safe startup module controls the test startup process in the order of system self-test - no-load startup - graded loading, and performs protection and recording in case of abnormality; Multi-mode test sequence execution module: Automatically generates and executes parameter scanning test sequences containing multiple working modes and test points based on preset test requirements; Dual closed-loop control module: The inner loop quickly corrects the output deviation, while the outer loop predictively compensates for external disturbances, thus maintaining the stability and accuracy of the testing process. Safety boundary protection module: Calculates and updates safety thresholds in real time, and ensures test safety through multi-level early warning and gradient interlocking protection mechanisms; Multi-source data fusion processing module: collects data from the entire testing process, cleans it, performs correlation analysis and double verification, and finally outputs a standardized and valid dataset; Processor: The processor is used to handle the calculation process of each formula and the calculation process of constructing each model.

[0018] See Figure 1 As shown, a test control method for a DC power supply includes: Step 1: Construct a three-dimensional adaptation model based on the rated parameters of the DC power supply under test, test requirements, and test environment benchmark parameters; match the hardware configuration and software parameters of the test system and perform adaptation calibration. Step 2: Simulate dynamic load characteristics based on the actual application scenario of the DC power supply under test. Collect the actual absorbed voltage and current data of the load in real time, compare them with the preset load parameters, and perform load calibration based on the comparison results. Step 3: Start the test through a three-level startup mechanism. The first level is the control module self-test, the second level is the no-load startup of the DC power supply under test, and the third level is the gradual connection of the load. If an abnormal situation occurs during the startup process, the shutdown protection is triggered and the abnormal node is recorded. Step 4: Generate a multi-mode test sequence based on the pre-configured test requirements. In each mode, perform parameter scanning tests according to the preset test point distribution. During the test, record the power parameters in real time. Step 5: Construct a dual closed-loop feedback regulation mechanism. The inner loop compares the collected power supply parameters with the preset target value to generate a regulation signal. The outer loop monitors the grid parameters, establishes a disturbance impact model based on historical test data, predicts the impact trend of disturbances on power supply output parameters, and outputs compensation signals in advance. Step 6: Generate a safety boundary threshold based on the rated parameters of the power supply under test, the test stage, and real-time environmental data. Compare the power supply parameters with the dynamic safety threshold in real time. When the monitored value approaches or exceeds the dynamic safety threshold, trigger the early warning mechanism and start the interlocking protection, respectively. Step 7: Collect multi-source data during the testing process, use a hierarchical fusion algorithm for data processing, and use a dual verification mechanism to determine the validity of the data to obtain a valid dataset.

[0019] See Figure 2 As shown, a three-dimensional adaptation model is constructed based on the rated parameters of the DC power supply under test, test requirements, and test environment benchmark parameters. This model is then matched with the hardware configuration and software parameters of the test system, and adaptation calibration is performed. Specifically, this includes: The rated parameters of the DC power supply under test include output voltage, current range, power, and ripple rejection ratio design value. Testing requirements include steady-state accuracy, dynamic response, overload capacity, and environmental adaptability testing. Environmental baseline parameters, including temperature, humidity, and power grid input voltage fluctuation range, are collected in real time through an integrated temperature and humidity sensor. A three-dimensional adaptation model is built based on the collected power supply rated parameters, test requirements, and environmental benchmark parameters. The mapping relationship between parameters is established, and the hardware and software parameters of the test system are matched through the three-dimensional adaptation model. The matched sensors and load modules are then pre-calibrated.

[0020] Specifically, an adaptation model is built based on the collected rated parameters of the power supply under test as the basic dimension, the test requirements as the functional dimension, and the environmental benchmark parameters as the adaptation dimension. This model automatically matches the hardware configuration of the test system, selects the corresponding resistive or electronic type according to the power level of the power supply, determines the sampling rate specification of the data acquisition module according to the test type, and matches the sensor range with the rated voltage and current range to ensure that the hardware is accurately matched with the test object and test requirements. At the same time, the software parameters are matched synchronously, the data acquisition cycle and single-item test duration are set according to the test accuracy requirements, and the parameter judgment thresholds are defined according to the power supply design standards to form an initial configuration scheme for hardware and software collaboration. By calling the built-in calibration parameter library, the system matches the corresponding standard calibration curve and temperature compensation coefficient according to the sensor model and load module type; it applies stepped reference signals (such as zero point, full scale point, and midpoint) to the sensor through a standard signal source, collects sensor feedback data in real time and compares it with the reference value, and dynamically corrects its linearity and zero drift; it applies preset static and dynamic load modes (such as constant resistance and constant current switching) to the load module, and iteratively adjusts the amplitude / phase compensation parameters of its drive circuit based on the deviation between the actual absorbed power and the command value; it simultaneously integrates real-time temperature and humidity data to perform online compensation for sensor temperature drift and temperature characteristics of load power devices.

[0021] See Figure 3As shown, based on the actual application scenario of the DC power supply under test, the dynamic load characteristics are simulated. Real-time data on the actual absorbed voltage and current of the load are collected and compared with preset load parameters. Load calibration is then performed based on the comparison results, specifically including: Based on the specific application scenario of the DC power supply under test, dynamic load characteristics are simulated to generate various dynamic loads, including pulse load, nonlinear load, stepped load, and random fluctuating load. The system continuously collects the actual voltage and current data absorbed by the load and compares them with the preset load parameters. If the deviation exceeds ±1%, it adjusts the amplitude and phase of the load drive signal and compensates for the nonlinear error and temperature drift of the load module to correct the load simulation deviation.

[0022] Specifically, dynamic loads are generated to simulate the complex load characteristics faced by the power supply in actual working conditions, such as pulse loads generated by the start and stop of industrial equipment, nonlinear loads caused by frequency converters, stepped loads during battery charging, and random loads caused by power grid fluctuations, to ensure that the test environment is close to reality. Simulate the load change process in actual working conditions, such as the pulse load when industrial equipment starts and stops, and the nonlinear load fluctuation when electronic circuits are working. Simultaneously, use high-frequency acquisition to capture the actual voltage and current data absorbed by the load in real time, record the actual load value at each time point, and form a continuous load operation data curve. The actual load data collected is compared with the preset load parameters point by point, and the deviation value between the two is calculated. The deviation is focused on key stages such as load type switching and peak fluctuation. A deviation judgment standard is set. When the deviation value of three consecutive collection cycles exceeds ±1% or the single deviation exceeds ±3%, it is judged that the load simulation accuracy is not up to standard, and the load calibration mechanism is immediately triggered. Preset load parameters include: Waveform characteristic parameters: Pulsed load: peak current, rise / fall time, duty cycle; Nonlinear loads: harmonic spectrum, power factor; Stepped load: step amplitude, duration, step interval; Random load: fluctuation frequency range, maximum offset amplitude; Electrical performance benchmark values: Target voltage / current curve, theoretical power consumption curve, impedance change timing; To address nonlinear errors, a compensation algorithm is used to correct the load output characteristics, making them conform to the preset nonlinear curve. To address deviations caused by temperature drift, the calibration coefficient is adjusted to offset the temperature effect by combining the real-time temperature data of the load module. After calibration, the deviation between the actual load data and the preset value is continuously monitored until the deviation stabilizes within ±1%, forming a closed-loop control of simulation-acquisition-comparison-calibration-verification.

[0023] See Figure 4 As shown, the test is initiated through a three-level startup mechanism. The first level is a self-test of the control module, the second level is a no-load startup of the DC power supply under test, and the third level is a gradual connection of the load. Based on any abnormal situations that occur during startup, a shutdown protection mechanism is triggered and the abnormal nodes are recorded. Specific abnormal nodes include: Perform a self-test of the first-level control module to verify whether the connectivity between modules is normal, and check the working status of each module to ensure that there are no hardware faults or communication interruptions. The second-level no-load start-up and monitoring is performed. The DC power supply under test is controlled to enter the no-load start-up mode. The power supply output voltage is controlled to rise from 0V to the test start voltage at a preset slope. The ripple voltage and noise voltage of the power supply output under no-load conditions are monitored in real time to determine whether there is no-load current leakage. After performing the third level of graded loading and starting without any abnormalities, gradually connect the load to the power supply according to the load ratios of 10%, 30%, 50%, 80%, and 100%, and continuously monitor the fluctuation range of the power supply output parameters. If the ripple voltage exceeds 50% of the design threshold, current leakage is found, or parameter fluctuations exceed ±3% during graded loading, the shutdown protection mechanism is immediately triggered, and the node data where the abnormality occurred is recorded.

[0024] Specifically, the slope setting is dynamically determined based on the power supply rating, test start voltage, and ambient temperature. For high-power power supplies (>1kW): a gentler slope (e.g., 0.5-1V / s) is used to avoid surge current caused by sudden voltage application; for low-power power supplies (<200W): a steeper slope (e.g., 5-10V / s) is used to accelerate the startup process; for high-temperature environments (>40℃): the slope is reduced by 20%-30% to prevent excessive temperature rise. The start-up control module performs a comprehensive self-test, focusing on verifying the working status of the core components of the test system; it checks the signal transmission continuity of each data acquisition channel, confirming the voltage and current acquisition links are uninterrupted by using simulated signal input; it checks the command response capability of the load drive module, sending an unloaded drive signal to verify whether the module is responding normally; it tests the connection stability of the communication module to ensure that there is no delay or loss in the transmission of control commands and feedback data; if any module is found to have abnormal connectivity or a malfunction during the self-test, the startup process is immediately paused. After the self-test passes, the second-level no-load start-up procedure of the DC power supply under test is executed. The voltage is gradually increased from 0V to the test start voltage according to the preset adjustable slope. The ripple voltage and noise voltage data of the power supply output under no-load conditions are collected in real time, and the power supply output terminal is continuously monitored for any no-load current leakage. During this period, the peak value and fluctuation frequency of the ripple voltage, the amplitude of the noise voltage, and the presence or absence of leakage current are closely monitored. The collected parameters are compared with the preset safety thresholds in real time to ensure the basic performance of the power supply under no-load conditions is stable. After the no-load test passes, the third stage of gradual load connection begins. Loads are applied in stages according to proportions of 10%, 30%, 50%, 80%, and 100%. After each stage, a 3-10 second stabilization period is maintained to allow time for the power supply output parameters to adjust. During each stabilization period, the fluctuation range of the power supply's output voltage and current is continuously monitored, and the parameter change curves are recorded. If the parameter fluctuations are stable during the loading process and the fluctuation range is controlled within the allowable range during the stabilization period, the next stage of load loading is executed. If abnormal fluctuations such as sudden increases or decreases in parameters occur, the loading process is immediately paused. Design thresholds refer to the permissible limits in the power supply's manufacturer's technical specifications, including ripple voltage threshold (typically 1%-3% of the rated output voltage), noise voltage threshold (generally 1.5-2 times the ripple threshold), and leakage current threshold (set according to safety standards). During graded loading, the instantaneous fluctuations of output voltage and output current are monitored, and the fluctuation amplitude is obtained by dividing the difference between the maximum and minimum values ​​of output voltage and output current by the rated output voltage and output current values.

[0025] See Figure 5 As shown, a multi-mode test sequence is generated based on pre-configured test requirements. In each mode, parameter scanning tests are performed according to the preset test point distribution. During the test, power parameters are recorded in real time, including: Generate test sequences including constant voltage mode, constant current mode, constant power mode and mixed mode based on pre-configured test requirements; Test points are set for each test mode, including rated value, limit value, critical value and random sample value. The limit value test adopts a step-by-step approximation method, with each increment or decrement being 5% of the rated value. Based on the set test sequence and test points, parameter scanning tests are performed. During the test, the power supply's output voltage accuracy, output current accuracy, load regulation, source effect regulation, dynamic response time, and output ripple peak-to-peak parameters are monitored in real time.

[0026] Specifically, the step-by-step execution process first performs basic mode tests such as constant voltage and constant current. In constant voltage mode, the target value of the power supply output voltage is fixed, and the load current is gradually adjusted through the load module, with real-time monitoring of voltage accuracy and load regulation. In constant current mode, the target value of the output current is fixed, the load resistance is adjusted, and the current accuracy and source effect regulation are recorded. After the basic mode tests are completed, constant power mode and hybrid mode are executed: constant power mode maintains stable output power by coordinating voltage and current adjustment, and power stability and parameter coordination adjustment response speed are monitored. In hybrid mode, the working mode is switched strictly according to preset rules, and the parameter transition curve and stabilization time are recorded at each switch.

[0027] See Figure 6As shown, a dual-loop feedback regulation mechanism is constructed. The inner loop compares the collected power supply parameters with preset target values ​​to generate a regulation signal. The outer loop monitors grid parameters, establishes a disturbance impact model based on historical test data, predicts the impact trend of disturbances on power supply output parameters, and outputs compensation signals in advance. Specifically, this includes: A dual-closed-loop feedback regulation mechanism is constructed, with the inner loop for output parameter feedback and the outer loop for disturbance prediction and compensation. The inner loop compares the collected actual data with the preset target value to generate an adjustment signal, which then controls the PWM drive module of the power supply to correct the output deviation. The outer ring collects data on grid input voltage fluctuations, ambient temperature changes, and load mutations. Based on historical test data, a disturbance impact model based on an LSTM neural network is established to predict the impact trend of disturbances on power supply output parameters. Based on the prediction results, compensation signals are output in advance. When a sudden disturbance such as a sharp drop in grid voltage or a load change exceeding a threshold is detected, an emergency adjustment mode is activated, shortening the feedback adjustment cycle and calling a pre-stored disturbance compensation strategy library to offset the impact of the sudden disturbance.

[0028] Specifically, the inner loop output parameter feedback adjustment process is initiated. A fixed data acquisition cycle of 1ms-5ms is set, and a high-speed acquisition module continuously captures real-time voltage and current data at the power supply output. The acquired real-time data is compared cycle-by-cycle with the target value pre-configured in step one, and the parameter deviation is calculated, with a focus on the trend of deviation changes. Based on the deviation characteristics, a targeted adjustment signal is generated. If the voltage is too high, a voltage reduction command is sent to the adjustment module; if the current is too low, a current boost control signal is output. By precisely adjusting the internal output link parameters of the power supply, the deviation is quickly offset, maintaining stable output parameters. The outer loop disturbance prediction and compensation operation continuously monitors three core disturbance factors: instantaneous fluctuations in grid input voltage, sudden changes in test environment temperature, and sudden increases and decreases in load power. Disturbance signals are captured by dedicated sensors and uploaded to the control module in real time. By accumulating historical test data, we collect disturbance characteristic data and corresponding output parameter change data to establish a correlation database. The disturbance characteristic data includes: disturbance type (grid input voltage fluctuation, ambient temperature change, load abrupt change, etc.), disturbance intensity (e.g., grid voltage drop ≥10%, load current change rate ≥15% / ms), duration (e.g., drop lasting more than 20ms), and environmental parameters (e.g., current temperature, humidity). The output parameter change data includes: the deviation of power supply output parameters (voltage, current, power) under the corresponding disturbance (e.g., voltage deviation ±2%), change trend (e.g., continuous rise / fall), and recovery time (e.g., recovery to stability within 10ms). The dynamic disturbance impact model (LSTM neural network) is trained by taking disturbance feature data as input and output parameter change data as output. The model learns the nonlinear mapping relationship between disturbances and the impact of output parameters. Model inputs include disturbance type (encoded numerically), disturbance intensity (e.g., a 10% drop in power grid voltage is denoted as 0.1), duration, and ambient temperature. Model outputs include the prediction deviation of output parameters (voltage / current) (e.g., ±2.5%) and the predicted trend (e.g., "increasing" / "decreasing"). The model parameters (number of hidden layer nodes, learning rate, number of iterations) are adjusted using gradient descent to minimize the mean squared error (MSE) between the predicted and actual values, ensuring the model's prediction accuracy for different disturbance scenarios (e.g., cosine similarity ≥ 90%). The system collects disturbance data in real time during the test using sensor modules (voltage sensor, current sensor, temperature and humidity sensor) and extracts key features: Disturbance type identification includes judging the disturbance type (such as power grid fluctuation, load change) by signal characteristics (such as voltage waveform drop, current change); Disturbance intensity calculation includes quantifying the disturbance degree (such as instantaneous drop in power grid input voltage / rated voltage × 100%, load current change rate di / dt); Environmental parameter fusion corrects the disturbance intensity by synchronously collecting current temperature and humidity. The perturbation feature parameters extracted in real time are input into the trained LSTM model, and the output prediction results are: Impact magnitude: the deviation value of the power supply output parameters caused by the perturbation; Impact trend: the direction of change of the output parameters; Confidence assessment: the confidence of the prediction results given by the model. If the confidence is lower than 80%, a fallback strategy is triggered (such as using historical average compensation parameters). Based on the predicted impact magnitude and trend, and combined with the pre-stored "disturbance feature-compensation parameter" mapping library, a targeted compensation signal is generated; historical disturbance scenarios are analyzed using an LSTM model to generate corresponding optimal compensation parameters, which are stored according to "disturbance type + intensity"; real-time disturbance features are matched with entries in the mapping library to retrieve the corresponding compensation parameters and generate a digital compensation signal; the compensation signal is superimposed on the inner loop of the dual closed-loop regulation mechanism to drive the power supply hardware module to correct the output deviation in advance and offset the potential impact of the disturbance. A voltage sag is triggered when the instantaneous drop in grid input voltage is ≥10% of the rated voltage or the sag lasts for more than 20ms. The load sag amplitude is determined using the current change rate (di / dt) as the core indicator. The base threshold is 15% of the power supply's rated current / ms, and dynamic corrections include a 20% reduction in the threshold under high-temperature environments (>40℃) and a 10% increase in the threshold during extreme testing. The feedback adjustment cycle is shortened through hardware acceleration. When a sudden disturbance triggers the emergency mode, the inner loop adjustment cycle is compressed from the usual 1-5ms to 200-500μs. The FPGA is used to process the PWM signal in real time and bypass the software protocol layer. Sensor data is directly read through memory mapping technology to eliminate communication delay. The pre-storage process of the compensation strategy library is based on historical disturbance data. It collects historical data of typical disturbance scenarios during DC power supply testing, including: disturbance characteristics: grid sag, load change, ambient temperature change, recording the type, intensity, and duration of the disturbance; power supply response data: output voltage / current deviation and dynamic response time of the power supply when the disturbance occurs; compensation parameters used at that time, and output deviation after compensation; Key time-series features are extracted from historical data and used as input to the LSTM model, including: disturbance type encoding: classifying and encoding grid sags, load mutations, etc.; disturbance intensity quantification: grid sags are represented by "percentage drop in rated voltage", and load mutations are represented by "rate of change of current (di / dt)"; environmental co-features: incorporating ambient temperature and humidity to correct for the impact of disturbances on the power supply; and data cleaning and normalization. LSTM neural network training is used to explore the nonlinear temporal relationship between disturbance features and optimal compensation parameters: Input layer: temporal data such as disturbance type, intensity, and environmental co-features (e.g., the disturbance intensity sequence of the past 5 sampling points); Output layer: corresponding optimal compensation parameters (e.g., PWM duty cycle adjustment, current loop gain coefficient); Minimize the mean square error (MSE) between "model predicted compensation parameters" and "historical effective compensation parameters" to ensure that the model can predict the optimal compensation scheme under different disturbance scenarios; The trained LSTM model generates the optimal compensation parameters for each type of historical disturbance scenario, and stores them as policy entries according to the type and intensity of the disturbance, forming a "disturbance feature-compensation parameter" mapping library. Strategy invocation and disturbance cancellation are executed through dynamic matching: the disturbance characteristics monitored in real time are matched with the entries in the strategy library (e.g., if the cosine similarity is >90%, it is directly invoked); the compensation signal is superimposed on the inner loop regulation output to drive the PWM module to correct the deviation in advance; the compensation effect is verified within 10ms, and if the deviation is still >2%, gradient compensation is started (increasing the compensation amount by 20%) until the output returns to stability (deviation ≤1%).

[0029] See Figure 7 As shown, a safety boundary threshold is generated based on the rated parameters of the power supply under test, the test stage, and real-time environmental data. The power supply parameters are compared in real-time with the dynamic safety threshold. When the monitored value approaches or exceeds the dynamic safety threshold, an early warning mechanism and interlocking protection are triggered, respectively. Specifically, this includes: Based on the rated parameters of the power supply under test, the current test stage, and real-time environmental data, safety boundary thresholds are generated, including the maximum allowable output voltage, the maximum allowable output current, the maximum allowable temperature rise, and the minimum allowable insulation resistance. The system continuously collects output parameters, casing temperature, internal component temperature, and insulation resistance value of the power supply. The collected real-time data is compared with the dynamic safety boundary threshold in real time. If the monitored value reaches 90% of the threshold, an early warning mechanism is triggered, and the test load is reduced and the test parameters are adjusted. If the monitored value exceeds the safety boundary threshold, the shutdown operation will be carried out in the order of reducing current, reducing voltage, and cutting off power, while disconnecting the electrical connection between the load and the power supply.

[0030] Specifically, the collected monitoring data are compared with the dynamic safety threshold at the millisecond level. When the monitored value reaches 90% of the safety threshold, the early warning mechanism is immediately triggered: the current test load is automatically reduced or the test parameters are adjusted, and an audible and visual warning signal is issued to remind the operator to pay attention to the equipment status. If the monitored value continues to rise or directly exceeds the safety threshold after the warning, the interlock protection program is immediately activated, and the shutdown operation is strictly performed according to the gradient of current reduction, voltage reduction, and power cut-off. First, the output current is quickly reduced to a safe range, then the output voltage is gradually reduced to 0V, and finally the main power supply circuit is cut off, and the electrical connection between the load and the power supply is disconnected simultaneously.

[0031] See Figure 8 As shown, during the testing process, multi-source data was collected, and a hierarchical fusion algorithm was used for data processing. A dual verification mechanism was used to determine the validity of the data, and the specific details of obtaining a valid dataset include: All types of data from the entire testing process are aggregated, including power output parameter data, load feedback data, environmental monitoring data, disturbance record data, and safety protection data, and a layered fusion algorithm is used for data processing. Through the first step of data cleaning in the layered fusion process, outliers and duplicate records in the data are identified and removed, and invalid data interference is eliminated. By analyzing the correlation between different data types through the second step of hierarchical fusion, a corresponding mapping between power supply output parameters and environmental changes, load status, and disturbance factors is constructed. The effectiveness verification of the third step of the hierarchical integration verifies the consistency of the values ​​of voltage, current, and power, as well as the timing matching of the dynamic response curve with load changes. The key test points of rated values ​​and extreme values ​​are repeatedly tested, and the verified data are integrated to form a standardized and effective dataset.

[0032] Specifically, multi-dimensional data association and fusion are implemented to construct a hierarchical fusion system. The first layer performs basic data association, mapping power output data, load data, and environmental data at the same time stamp to form a basic parameter-load-environment association unit. The second layer conducts deep logical mapping, analyzing the causal relationship between output parameter changes and factors such as increased ambient temperature, sudden load changes, and grid fluctuations, establishing a logical association model for multi-dimensional data. For example, it clarifies the correspondence between output voltage fluctuations over a certain period and sudden drops in grid voltage during the same period. Through hierarchical fusion, scattered single-dimensional data are integrated into a comprehensive dataset with logical relationships, clearly presenting the impact patterns of various factors on power supply performance. The dual validity verification mechanism has two layers. The first layer is logical verification, which checks the logical consistency of the values ​​of voltage, current, and power to ensure that the actual collected data conforms to the basic working principle of the electrical equipment and has no obvious logical contradictions. It also verifies the timing matching of dynamic response data with load changes, confirming that the changing trend of power supply output parameters is synchronized with the load adjustment timing, with no obvious lag or lead deviation. The second layer is repeated verification, which retrieves the results of 3-5 repeated tests for key test points such as rated values ​​and extreme values, analyzes the degree of data variation, and determines that the test point data is valid if the dispersion of the test results is within an acceptable range. The formula for calculating data discreteness is: ; in, The relative standard deviation, The standard deviation is... The average of n repeated test data. This represents the actual value collected during the i-th test.

[0033] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0034] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0035] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A test control method of a direct current power supply, characterized by, The application relates to a test system and a test method for a direct-current power supply. A three-dimensional adaptive model is constructed based on rated parameters of the direct-current power supply to be tested, test requirements and test environment reference parameters, hardware configuration and software parameters of the test system are matched and adaptive calibration is performed; Dynamic load characteristics are simulated based on actual application scenarios of the direct-current power supply to be tested, actual absorption voltage and current data of the load are collected in real time, preset load parameters are compared, and load calibration is performed based on comparison results; The test is started through a three-stage starting mechanism, the first stage is self-checking of a control module, the second stage is no-load starting of the direct-current power supply to be tested, and the third stage is gradual connection of the load, abnormal conditions occurring in the starting process trigger shutdown protection and record abnormal nodes; A plurality of mode test sequences are generated based on preconfigured test requirements, parameter scanning tests are performed according to preset test point distribution in each mode, and power supply parameters are recorded in real time during the test process; A double-loop feedback regulation mechanism is constructed, an inner loop compares power supply parameters with preset target values to generate a regulation signal, and an outer loop monitors power grid parameters, establishes a disturbance influence model based on historical test data, predicts an influence trend of the disturbance on power supply output parameters, and outputs a compensation signal in advance; Safety boundary thresholds are generated based on rated parameters of the power supply to be tested, test stages and real-time environment data, real-time comparison is performed between power supply parameters and dynamic safety thresholds, and a prewarning mechanism and interlocking protection are triggered when monitored values approach and exceed the dynamic safety thresholds; Multi-source data in the test process are collected, a hierarchical fusion algorithm is used for data processing, a double-checking mechanism is used for validity determination of the data, and an effective data set is obtained.

2. The test control method of a DC power supply according to claim 1, characterized by, The three-dimensional adaptive model is constructed based on rated parameters of the direct-current power supply to be tested, test requirements and test environment reference parameters, hardware configuration and software parameters of the test system are matched and adaptive calibration is performed, and the three-dimensional adaptive model specifically comprises the following steps: The rated parameters of the direct-current power supply to be tested include output voltage, current range, power and ripple rejection ratio design values; The test requirements include steady-state accuracy, dynamic response, overload capacity and environmental adaptability test; Environment reference parameters are collected in real time by an integrated temperature and humidity sensor, including temperature, humidity and power grid input voltage fluctuation range; A three-dimensional adaptive model is built based on collected power supply rated parameters, test requirements and environment reference parameters, a mapping relationship between parameters is established, hardware and software parameters of the test system are matched through the three-dimensional adaptive model, and pre-calibration is performed on the matched sensors and load modules.

3. The method of claim 1, wherein the test control method of a DC power supply is characterized by, Dynamic load characteristics are simulated based on specific application scenarios of the direct-current power supply to be tested, a plurality of dynamic loads are generated, including pulse load, nonlinear load, step load and random fluctuation load; Actual absorption voltage and current data of the load are continuously collected, preset load parameters are compared, if the deviation exceeds + / - 1%, the amplitude and phase of the load driving signal are adjusted, nonlinear errors and temperature drifts of the load module are compensated, and load simulation deviation is corrected. ​ 4. The method of claim 1, wherein the test control method of a DC power supply is characterized by, The test is started through a three-stage starting mechanism, the first stage is a control module self-check, the second stage is a no-load start of the to-be-tested direct current power supply, and the third stage is a gradual access of a load, based on an abnormal situation occurring in a starting process, triggering a shutdown protection and recording an abnormal node specifically includes: Performing a first stage control module self-check, verifying whether a connectivity between modules is normal, checking a working state of each module, and ensuring no hardware failure or communication interruption problem; Performing a second stage no-load start and monitoring, controlling the to-be-tested direct current power supply to enter a no-load start mode, controlling an output voltage of the power supply to rise from 0V to a test starting voltage according to a preset slope, monitoring a ripple voltage and a noise voltage of a no-load output of the power supply in real time, and determining whether there is a no-load current leakage; Performing a third stage graded loading, after the no-load start is abnormal, gradually accessing the load to the power supply according to a load ratio of 10%, 30%, 50%, 80% and 100%, and continuously monitoring a fluctuation amplitude of an output parameter of the power supply; If the ripple voltage exceeds 50% of a design threshold value, there is a current leakage, or the parameter fluctuation exceeds ±3% in the graded loading, immediately triggering a shutdown protection mechanism, and recording node data of an abnormal occurrence.

5. The method of claim 1, wherein the test control method of a DC power supply is characterized by, The multi-mode test sequence is generated based on a preconfigured test requirement, each mode performs parameter scanning test according to a preset test point distribution, and during the test process, the power supply parameters are recorded in real time, specifically including: Generating a test sequence containing a constant voltage mode, a constant current mode, a constant power mode and a mixed mode based on a preconfigured test requirement; Setting a test point for each test mode, including a rated value, a limit value, a critical value and a random sampling value, wherein the limit value test adopts a step-by-step approximation method, and each time the increase / decrease amplitude is 5% of the rated value; Performing parameter scanning test based on the set test sequence and test point, and monitoring the output voltage accuracy, the output current accuracy, the load regulation rate, the source effect regulation rate, the dynamic response time and the output ripple peak-to-peak value of the power supply in real time during the test process.

6. The method of claim 1, wherein The double closed-loop feedback regulation mechanism is constructed, the inner loop compares the collected power supply parameters with the preset target value to generate a regulation signal, and the outer loop monitors the power grid parameters, establishes a disturbance influence model based on historical test data, predicts the influence trend of the disturbance on the power supply output parameters, and outputs a compensation signal in advance, specifically including: The double closed-loop feedback regulation mechanism is constructed, the inner loop is an output parameter feedback, and the outer loop is a disturbance prediction compensation; The inner loop compares the collected actual data with the preset target value to generate a regulation signal, and controls the PWM drive module of the power supply to correct the output deviation through the regulation signal; The outer loop collects power grid input voltage fluctuation, environmental temperature change and load mutation data, establishes a disturbance influence model based on an LSTM neural network model according to historical test data, predicts the influence trend of the disturbance on the power supply output parameters, and outputs a compensation signal in advance according to the prediction result; When a sudden disturbance of a power grid voltage drop or a load mutation amplitude exceeding a threshold value is detected, an emergency regulation mode is started, the feedback regulation period is shortened, and a pre-stored disturbance compensation strategy library is called to offset the influence of the sudden disturbance.

7. The method of claim 1, wherein the method further comprises: The safety boundary threshold is generated based on the rated parameters of the power supply to be tested, the test stage and real-time environmental data, and real-time comparison is performed between the power supply parameters and the dynamic safety threshold. When the monitoring value approaches and exceeds the dynamic safety threshold, a pre-warning mechanism is triggered and interlocking protection is started, specifically including: The safety boundary threshold is generated based on the rated parameters of the power supply to be tested, the current test stage and real-time environmental data, including the maximum allowed output voltage, the maximum allowed output current, the maximum allowed temperature rise and the minimum allowed insulation resistance. The output end parameters, shell temperature, internal component temperature and insulation resistance value of the power supply are continuously collected, and real-time comparison is performed between the collected real-time data and the dynamic safety boundary threshold. If the monitoring value reaches 90% of the threshold, a pre-warning mechanism is triggered, and the test load is reduced and the test parameters are adjusted. If the monitoring value exceeds the safety boundary threshold, the shutdown operation is performed in the gradient of current reduction, voltage reduction and power-off, and the electrical connection between the load and the power supply is cut off.

8. The method of claim 1, wherein the method further comprises: The multi-source data in the test process is collected, the data is processed by a hierarchical fusion algorithm, the effectiveness of the data is determined by a double verification mechanism, and the effective data set is obtained, specifically including: All kinds of data in the test process are summarized, including power supply output parameter data, load feedback data, environmental monitoring data, disturbance record data and safety protection data, and a hierarchical fusion algorithm is used for data processing; Through data cleaning in the first step of hierarchical fusion, abnormal values and repeated records in the data are identified and removed, and invalid data interference is removed; Through data association analysis of the correlation between different data types in the second step of hierarchical fusion, the corresponding mapping between the power supply output parameters and the environmental changes, load state and disturbance factors is constructed; Through the effectiveness verification in the third step of hierarchical fusion, the consistency of voltage, current and power values and the time sequence matching of dynamic response curve and load change are verified, and the rated value and limit value key test points are repeatedly tested. The data that passes the verification is integrated to form a standardized effective data set.

9. A test control system of a DC power supply for implementing the test control method of a DC power supply according to any one of claims 1 to 8, characterized by including: Three-dimensional adaptation and calibration module: based on the power supply specifications, test requirements and environmental parameters, a model is constructed to complete the automatic matching and initial calibration of the test system software and hardware; Dynamic load simulation module: simulates various dynamic load characteristics in real application scenarios, and performs online calibration of load simulation accuracy through real-time data feedback; Three-level safety start module: controls the test start process in the order of system self-checking, no-load starting and graded loading, and executes protection and recording in abnormal conditions; Multi-mode test sequence execution module: automatically generates and executes a parameter scanning test sequence containing multiple working modes and test points according to the preset test requirements; Double closed-loop regulation module: corrects output deviation through inner loop and predicts and compensates external disturbance through outer loop to maintain stability and accuracy of the test process; Safety boundary protection module: real-time calculation and update of safety threshold, multi-level pre-warning and gradient interlocking protection mechanism to ensure test safety; Multi-source data fusion processing module: collects test process data, performs cleaning, correlation analysis and double verification, and finally outputs a standardized effective data set; Processor: the processor is used for processing calculation processes of respective formulas and construction calculation processes of respective models.

Images