A multi-output power supply safety system with overload protection and a testing method

Abstract

The application provides a multi-channel power output safety system with overload protection and a testing method. The method comprises the following steps: collecting real-time electrical parameters of a power supply loop of a device to be tested, acquiring output voltage, output current and power loss data of each channel, performing sliding window filtering and baseline calibration processing on the collected data, and generating a standardized electrical characteristic sequence of each channel; performing threshold comparison and trend slope analysis on the standardized electrical characteristic sequence, and generating a multi-channel electrical state feature set; through real-time collection and cross-channel correlation comparison of electrical parameters of each channel, when short circuit or leakage current of a certain channel increases sharply, the fault channel can be accurately locked within a microsecond time window and single-channel cutting-off is performed, so that the power supply quality of the remaining normal channels is effectively prevented from being affected by the backflow of fault current to the common bus.

Description

Technical Field

[0001] This invention proposes a multi-channel power output safety system and testing method with overload protection, belonging to the field of electronic component reliability verification and automated testing technology. Background Technology

[0002] In the mass R&D and production of semiconductor chips, power modules, and complex electronic systems, multi-channel parallel aging tests and functional verification are crucial for ensuring product reliability. With increasing test density, high-parallel power supply architectures with 16 or more channels have become the mainstream configuration. Currently, the industry commonly uses centralized power supply solutions with relay arrays for channel switching and control. While this architecture can meet basic needs in low-channel scenarios, it exposes significant safety vulnerabilities under high-density parallel testing conditions. Specifically, when a short circuit or leakage surge occurs in a channel of the device under test (DUT), the abnormal current rapidly flows back to the common power bus, causing a significant voltage drop. This leads to deterioration of the power supply quality in other normal channels, distorted test data, and in severe cases, even damage to the DUT. If a master switch is used for unified power-off protection, the entire batch of tests will be forced to stop, rendering completed test data invalid and severely impacting production efficiency. While existing technologies offer single-path fuse or electronic switch protection solutions, their response times are typically in the millisecond range, making it difficult to isolate faults within a microsecond time window. Furthermore, they lack the capability to assess the overall power supply stability. Therefore, there is an urgent need for a multi-path power supply safety testing method capable of accurately isolating faulty channels at the microsecond level while ensuring the continuous and stable operation of other channels. Summary of the Invention

[0003] This invention provides a multi-channel power output safety system and testing method with overload protection to solve the problems mentioned in the background art above:

[0004] This invention proposes a safety testing method for multi-channel power output with overload protection, the method comprising:

[0005] S1. Real-time electrical parameter acquisition of the power supply circuit of the device under test, obtaining the output voltage, output current and power loss data of each channel, performing sliding window filtering and baseline calibration on the acquired data, generating a standardized electrical feature sequence for each channel; threshold comparison and trend slope analysis are performed on the standardized electrical feature sequence to generate a multi-channel electrical state feature set;

[0006] S2. Perform cross-channel correlation comparison analysis on the multi-channel electrical status feature set, use the time-series anomaly detection algorithm to identify the degree of abrupt deviation of electrical parameters in each channel, map the degree of deviation to the preset fault level classification system, and generate fault level label data; conduct a propagation risk assessment on the fault level label data, determine whether the anomaly has the conditions to spread to adjacent channels, and generate fault propagation risk assessment data.

[0007] S3. Perform isolation strategy matching processing on the fault propagation risk assessment data, convert the risk assessment results in natural language form into machine-executable single-channel disconnection control commands, and generate single-channel power disconnection command data; perform execution timing arrangement on the single-channel power disconnection command data, and generate differentiated disconnection timing sequences by combining the load priority of each channel and the weight of the test phase, and generate time-sequential disconnection control data.

[0008] S4. Perform execution domain delineation processing on the time-sequential disconnection control data. Using the fault channel as the isolation core, combine the electrical coupling relationship of adjacent channels to perform boundary convergence of the disconnection influence domain and generate boundary data of the affected channels. Perform power supply continuity verification on the boundary data of the affected channels, mark the channels not included in the influence domain as continuous power supply channels, and generate a list of continuous power supply channels.

[0009] S5. Perform resource reallocation processing on the continuous power supply channel list data, allocate the power margin released by the faulty channel to the still running channels according to priority rules, and generate power dynamic rebalancing data; perform global stability verification on the power dynamic rebalancing data, and generate a multi-channel parallel test safe operation confirmation signal.

[0010] This invention proposes a multi-channel power output safety system with overload protection, the system comprising:

[0011] One or more processors;

[0012] Memory, used to store one or more programs;

[0013] Wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the test method described in any one of the above.

[0014] The beneficial effects of this invention are as follows: By real-time acquisition and cross-channel correlation comparison of electrical parameters of each channel, when a short circuit or leakage surge occurs in a certain channel, the faulty channel can be accurately located within a microsecond time window and a single-channel disconnection can be executed, effectively preventing the fault current from flowing back to the common bus and affecting the power supply quality of other normal channels. By narrowing the fault impact domain to the smallest range and verifying the power supply continuity of each unaffected channel, the faulty channel can be quickly isolated, and the testing process of other channels can be kept undisturbed, significantly reducing the probability of forced interruption of the entire batch of tests due to a single-channel anomaly. At the same time, the power margin released by the faulty channel can be dynamically allocated to the still-operating channels according to priority rules, enhancing the global power supply stability in multi-channel parallel testing scenarios and reducing the risk of test data distortion and device-under-test damage caused by bus voltage drops. The entire protection process does not rely on a master control switch for unified power outage, significantly improving the production efficiency of high-density parallel testing while ensuring equipment safety. Attached Figure Description

[0015] Figure 1 This is a diagram illustrating the steps of the method described in this invention;

[0016] Figure 2 This is a detailed flowchart of step S3 in this invention. Detailed Implementation

[0017] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0018] One embodiment of the present invention, such as Figure 1 As shown, a safety test method for multiple power outputs with overload protection is provided, the method comprising:

[0019] S1. Real-time electrical parameter acquisition is performed on the power supply circuit of the device under test with 16 DC input channels. Output voltage, output current and power loss data of each channel are obtained. The acquired data is processed by sliding window filtering and baseline calibration to generate a standardized electrical feature sequence for each channel. Threshold comparison and trend slope analysis are performed on the standardized electrical feature sequence to extract the overload precursor features and steady-state operation features of each channel and generate a multi-channel electrical state feature set.

[0020] S2. Perform cross-channel correlation comparison analysis on the multi-channel electrical status feature set, use the time-series anomaly detection algorithm to identify the degree of abrupt deviation of electrical parameters in each channel, map the degree of deviation to the preset fault level classification system, and generate fault level label data; conduct a propagation risk assessment on the fault level label data, determine whether the anomaly has the conditions to spread to adjacent channels, and generate fault propagation risk assessment data.

[0021] S3. Perform isolation strategy matching processing on the fault propagation risk assessment data, convert the risk assessment results in natural language form into machine-executable single-channel disconnection control commands, and generate single-channel power disconnection command data; perform execution timing arrangement on the single-channel power disconnection command data, and generate differentiated disconnection timing sequences by combining the load priority of each channel and the weight of the test phase, and generate time-sequential disconnection control data.

[0022] S4. Perform execution domain delineation processing on the time-sequential disconnection control data. Using the fault channel as the isolation core, combine the electrical coupling relationship of adjacent channels to perform boundary convergence of the disconnection influence domain and generate boundary data of the affected channels. Perform power supply continuity verification on the boundary data of the affected channels, mark the channels not included in the influence domain as continuous power supply channels, and generate a list of continuous power supply channels.

[0023] S5. Perform resource reallocation processing on the continuous power supply channel list data, allocate the power margin released by the faulty channel to the still running channels according to priority rules, and generate power dynamic rebalancing data; perform global stability verification on the power dynamic rebalancing data to confirm that the voltage drop of all continuous power supply channels is within the allowable range, and generate multi-channel parallel test safe operation confirmation signal.

[0024] The working principle and effects of the above technical solution are as follows: By acquiring and refining the characteristics of multi-channel electrical parameters in real time, it can accurately capture overload precursors and abnormal fluctuations in power supply circuits, effectively improving the accuracy and sensitivity of fault identification during multi-power supply testing. Through cross-channel correlation analysis and fault propagation risk prediction modes, it can identify fault propagation trends in advance, reducing the probability of synchronous faults in multiple power supplies. Through hierarchical isolation and time-sequential power-off control, it can selectively disconnect faulty channels, reducing unnecessary intervention on normal power supply channels and enhancing the overall operational stability of the testing process. By dynamically defining the power-off impact domain and power resource redistribution mechanism, it can avoid large-scale power supply fluctuations caused by local faults, ensure continuous and stable power supply to channels without abnormalities, and prevent local overload faults from interrupting the overall testing task, effectively improving the overall reliability and testing efficiency of parallel safety testing of multiple power supplies.

[0025] In one embodiment of the present invention, S1 includes:

[0026] S11. Deploy distributed multi-channel sampling and acquisition units along 16 DC input channels. The acquisition units synchronously capture raw sampling data of instantaneous output voltage, instantaneous output current and real-time power loss of each power supply circuit. The raw sampling data is uniformly merged into the front-end temporary storage block to complete the raw data storage and generate a raw electrical sampling dataset for the entire channel. The acquisition clock code value corresponding to each sampling is recorded synchronously and bound to the end of the corresponding single sampling data to generate raw sampling summary data with acquisition time sequence identifier.

[0027] S12. The original electrical sampling dataset of all channels is retrieved and sent to the filtering and operation unit. The filtering and operation unit calls the fixed-length sliding window algorithm to remove pulse clutter data caused by environmental electromagnetic interference layer by layer. After the clutter removal is completed, it is connected to the baseline calibration operation link. The operation link completes the zero-point correction of single-channel data based on the unloaded reference parameters of all channels. The corrected data is summarized and integrated to generate a standardized electrical feature sequence that is stored independently for each channel. The standardized electrical feature sequence is partitioned and stored in the feature cache storage space according to the channel number, generating partitioned storage of standardized sequence data.

[0028] S13. Retrieve partitioned standardized sequence storage data in batches from the feature cache storage space. The parameter comparison unit retrieves the factory-calibrated single-channel rated voltage threshold, rated current threshold and rated loss threshold. Perform numerical threshold comparison calculations for the standardized electrical feature sequence channel by channel. Simultaneously fit the change slope values ​​of the continuous sampling points of the sequence and generate single-channel threshold over-limit marker data and slope change statistics.

[0029] S14. The data of single-channel threshold over-limit markers and slope change statistics are sent to the feature extraction and operation module. The module filters the parameter points with sharp increases in short-cycle parameters to extract overload precursor features and filters the parameter points with stable fluctuations in long-term intervals to extract steady-state operation features. The two types of features are bound to the corresponding channel numbers to complete the classification and collection, and generate independent feature sub-items storage data for each channel.

[0030] S15. Integrate all independent feature items stored in each channel, remove duplicate and redundant feature items, and complete the cross-channel data aggregation. The aggregated data is then arranged in a structured manner according to the channel layout order to generate a complete multi-channel electrical status feature set. The multi-channel electrical status feature set is transferred to the feature database for storage and awaits subsequent calculations.

[0031] The working principle and effects of the above technical solution are as follows: By employing a distributed multi-channel synchronous sampling acquisition method, the time-series acquisition and storage of electrical parameters for multiple power supplies are completed, improving the comprehensiveness and synchronization of multi-channel power supply test data acquisition. Through sliding window filtering and baseline calibration, abnormal data deviations caused by environmental interference are eliminated, reducing the adverse effects of noise interference on parameter analysis and improving the purity and accuracy of the original electrical data. Through a dual analysis method of threshold comparison and slope fitting, the steady-state operating status and early overload change trends of channel electrical parameters can be accurately captured, allowing for early detection of abnormal equipment operation. By extracting various electrical features layer by layer and completing structured integration and storage, the independent analysis accuracy of single-channel operating features is ensured, while unified collection and management of multi-channel data is achieved, avoiding state analysis deviations caused by scattered data loss or redundancy, effectively improving the detail and overall accuracy of multi-channel power supply electrical status identification.

[0032] In one embodiment of the present invention, S13 includes:

[0033] The parameter comparison unit traverses the feature cache storage space, reads the partitioned storage-style standardized sequence storage data in sequence according to the channel number, performs time-series alignment and sorting on the batch retrieved sequence data, removes duplicate data entries remaining in the cache, and generates ordered standardized sequence sorting data.

[0034] The parameter comparison unit reads the electrical parameter standard system calibrated by the device at the factory, extracts the critical values ​​of rated voltage, rated current and rated power loss of a single channel, and summarizes them to form a standard dataset of channel electrical thresholds.

[0035] The parameter comparison unit performs frame-by-frame and channel-by-channel comparison operations on the ordered standardized sequence data and the channel electrical threshold standard dataset to determine the state of the real-time operating parameters exceeding the rated parameters and generate threshold comparison result data for each channel's single time-series node.

[0036] The parameter comparison unit performs linear fitting calculations on the sequence data of continuously sampled time nodes, statistically analyzes the rate of increase and decrease of electrical parameters within a fixed time period, calculates the dynamic change slope value of each channel parameter, and generates statistical data on the slope change of the entire channel.

[0037] The parameter comparison unit collects the abnormal threshold comparison results and normal operation results of each channel, uniformly marks the over-limit status of each channel, and integrates the information with the slope change statistics of the whole channel to generate complete single-channel threshold over-limit marking data and slope change statistics.

[0038] The working principle and effects of the above technical solution are as follows: By performing time-series alignment and redundancy cleanup on standardized sequence data within the cache, the regularity of electrical parameter data is improved, reducing interference from invalid data to subsequent analysis. Parameter comparison is performed against the factory-set electrical thresholds of the equipment, enabling accurate determination of over-limit conditions for real-time operating parameters of each channel, thus improving the accuracy of electrical anomaly detection. A detailed frame-by-frame, channel-by-channel comparison mode refines the parameter operating status at different time-series nodes, avoiding the problem of local time-series data anomalies being overwritten by the overall data. By calculating the slope of parameter changes through linear fitting, the dynamic changing trends of electrical parameters can be captured, compensating for the shortcoming of fixed threshold discrimination in identifying slow abnormal changes. This solution achieves both accurate determination of static over-limit states of electrical parameters and effective capture of dynamic operating trends, significantly improving the comprehensiveness and reliability of electrical status detection for multiple power supply channels.

[0039] In one embodiment of the present invention, S2 includes:

[0040] S21. Read the multi-channel electrical state feature set completely from the feature database. The cross-channel comparison unit uses the same sampling clock as the alignment reference to horizontally compare the voltage, current and loss feature values ​​of adjacent numbered channels at the same time node, record the result of the time sequence parameter difference one by one, and generate cross-channel parameter difference statistical ledger data after the full channel comparison is completed.

[0041] S22. The timing operation module loads the timing anomaly detection algorithm program, analyzes the continuous timing parameter change pattern in the cross-channel parameter difference statistical ledger data line by line, and quantifies the specific value of parameter mutation deviation in each channel by accumulating the parameter offset in multiple consecutive sampling periods. After all the channel deviation values ​​are calculated, a full-channel parameter mutation deviation quantification dataset is generated.

[0042] S23. The hierarchical mapping unit retrieves the pre-divided multi-level fault level classification scale, completes hierarchical matching according to the deviation value range of each channel within the full-channel parameter mutation deviation quantification dataset, binds each matching result to the corresponding channel number to generate a single-channel fault level record, and generates structured fault level label data after summarizing and organizing all records; the fault level label data is sorted according to the severity of the fault and stored in the hierarchical label temporary cache.

[0043] S24. The risk assessment unit reads the fault level label data in the temporary cache of the graded label, retrieves the inter-channel line coupling impedance parameters and bus common load parameters recorded in the hardware layout of the device under test, and performs an inter-channel abnormal conduction probability calculation in combination with the fault severity level marked by the label. It calculates the quantitative probability value of a single faulty channel conducting faults to adjacent channels one by one and generates detailed data of single-channel conduction probability calculation.

[0044] S25. Collect all detailed data on single-channel transmission probability calculations, perform correlation merging calculations according to the arrangement of adjacent channels, divide the fault propagation intervals into three categories: high, medium, and low, bind and integrate the interval division results with the corresponding fault channel information, generate the final fault propagation risk assessment data, and transfer the fault propagation risk assessment data to the risk database for storage and backup.

[0045] The working principle and effects of the above technical solution are as follows: By simultaneously comparing the electrical parameters of multiple channels using a synchronous reference, the differences in parameters between channels can be accurately captured, improving the detail of troubleshooting anomalies in multiple power supply channels. By quantifying parameter mutation states through multi-cycle parameter offset accumulation calculations, the judgment error caused by instantaneous sampling fluctuations is weakened, making the judgment results of electrical anomaly deviations more consistent with the actual operating conditions of the equipment. By classifying fault levels through a standardized hierarchical matching method, the judgment criteria for various abnormal states are unified, improving the standardization and accuracy of fault level classification. By combining the equipment hardware coupling parameters to calculate the fault propagation probability, the fault propagation pattern between channels can be realistically restored, improving the accuracy of risk prediction. It can accurately locate electrical anomalies in a single channel and predict the scope of fault spread in advance, reducing the testing risks caused by the spread of sudden faults and effectively enhancing the risk control capabilities of multi-power supply testing processes.

[0046] In one embodiment of the present invention, step S22 includes:

[0047] The timing calculation module retrieves cross-channel parameter difference statistical ledger data, performs timing integrity verification and invalid data removal on the ledger data, clears abnormal entries with missing data and timing misalignment, and generates normalized timing difference statistical data.

[0048] The timing calculation module is equipped with a timing anomaly detection algorithm program, which traverses and analyzes the normalized timing difference statistics item by item, captures the dynamic fluctuation characteristics and numerical change trajectory of the electrical parameters of each channel within a continuous sampling period, and generates dynamic change characteristic data of timing parameters.

[0049] The timing calculation module performs multi-cycle offset statistical calculations on the dynamic change characteristic data of timing parameters, accumulates and corrects the parameter offset values ​​generated in different sampling periods, and generates single-cycle parameter offset quantization data for each channel.

[0050] The timing calculation module combines the normal operating parameter benchmark of the channel to perform abruptness identification calculation on the single-cycle parameter offset quantization data of each channel, distinguishes between normal fluctuation offset and abnormal abrupt offset, and generates the basic data of parameter abruptness deviation quantization for each channel.

[0051] The timing operation module summarizes and integrates all channel parameter mutation deviation quantization basic data, completes the unified regularization and structured collection of all channel data, and generates a standardized full-channel parameter mutation deviation quantization dataset.

[0052] The working principle and effects of the above technical solution are as follows: By verifying the integrity of time-series statistical data and cleaning up invalid entries, various abnormal data can be filtered out, reducing the interference of incomplete and misaligned data on parameter analysis results and ensuring the quality of basic data for subsequent calculations. By traversing and analyzing the change characteristics of time-series data item by item, the dynamic fluctuation trajectory of electrical parameters of each channel is comprehensively captured, improving the comprehensiveness of parameter change feature capture. Through multi-cycle offset accumulation calculation and deviation correction, the calculation error caused by single sampling fluctuations is weakened, making the parameter offset statistical results more consistent with the actual operating state of the equipment. By identifying the degree of data mutation through normal parameter benchmarks, the fluctuation of normal operation and abnormal parameter mutations are effectively distinguished, avoiding the misjudgment of normal parameter fluctuations as equipment failures. This not only ensures the accuracy of data calculation but also improves the effectiveness of electrical parameter anomaly identification, effectively improving the overall quality of quantitative analysis of multi-channel power supply parameter anomalies.

[0053] One embodiment of the present invention, such as Figure 2 As shown, S3 includes:

[0054] S31. Extract complete fault propagation risk assessment data from the risk database. The strategy retrieval unit retrieves the pre-stored hierarchical isolation and disposal rule base. It compares the propagation interval level marked in the risk assessment data with the isolation and disposal entries in the rule base one by one. It completes the automatic retrieval and matching of the corresponding isolation and disposal plan. The matching results are summarized to generate a risk and isolation plan matching list data.

[0055] S32, The instruction conversion unit reads the risk and isolation scheme matching list data. The isolation treatment content recorded in natural language in the list is compiled into binary control code through the built-in encoding conversion logic. The compiled control code is bound to the target channel that needs to perform power-off action. A single-channel independent power-off control code is generated one by one. All codes are summarized to generate batch single-channel power cut-off instruction data.

[0056] S33. The priority retrieval module loads the preset full-channel load priority file and the weight parameters of the current test stage of the device under test. The load priority file is divided into multiple load weight levels. The test stage weight is distinguished into three test conditions: no load, full load, and overload. The two types of parameters are uniformly stored in the weight temporary data table to generate the channel weight collection data table.

[0057] S34. The timing orchestration module synchronously retrieves the single-channel power cut-off command data and the channel weight collection data table. The fault-related channels with low weight values ​​are given priority in arranging the power cut-off execution points, while the related channels with high weight values ​​are arranged with delayed power cut-off execution points. The execution clock node corresponding to each cut-off command is marked one by one, and the initial timing draft data of the single command binding execution time is generated.

[0058] S35. Verify whether there is a power outage conflict between multiple channels of the same clock in the internal instruction execution clock of the initial timing data. Fine-tune the execution interval for instructions with overlapping times. After the interval is corrected, complete the solidification of the full instruction timing and generate the final timing cut-off control data. The timing cut-off control data is sent to the control instruction buffer.

[0059] The working principle and effects of the above technical solution are as follows: By automatically matching risk levels with isolation and handling rules, it can quickly adapt to isolation methods for different fault propagation scenarios, improving the fit of fault handling solutions. By converting risk handling content into standardized equipment control codes, it eliminates execution deviations caused by natural language descriptions, improving the execution accuracy of power outage control. By integrating channel load levels and test condition weight parameters, it completes differentiated timing arrangements of power outage commands, changing the coarse handling mode of uniform power outage. By investigating and correcting time conflicts in command execution, it eliminates the potential electrical disturbance hazards caused by simultaneous power outages of multiple channels, ensuring the orderly and stable execution of various power outage commands. It can quickly block the abnormal propagation of fault channels, reduce the impact of faults on the overall test circuit, and maximize the stable operation of important load channels, effectively enhancing the safety and rationality of multi-power supply overload fault isolation.

[0060] In one embodiment of the present invention, S32 includes:

[0061] The instruction conversion unit reads the risk and isolation scheme matching list data in the cache area, decomposes and splits the content and fields of various isolation disposal contents in the list, removes redundant descriptive information, retains the core disposal contents of channel isolation and power outage control, and generates standardized isolation disposal text data.

[0062] The instruction conversion unit calls the built-in encoding conversion logic to parse and convert the standardized isolation processing text data line by line, and compile the processing content expressed in natural language into binary control code elements that the device can recognize in batches, generating the original power-off control code metadata.

[0063] The instruction conversion unit performs field regularization and format calibration on the original power-off control code metadata, unifies the code compilation length and instruction format, removes invalid code content with compilation errors and format disorder, and generates basic data for compliant control coding.

[0064] The instruction conversion unit binds and associates the basic data of the compliance control code with the corresponding target power supply channels within the list, completes the precise matching of control codes and execution channels, and generates exclusive power outage control code data for a single channel.

[0065] The instruction conversion unit collects all single-channel dedicated power-off control code data, integrates and archives the data according to the channel arrangement order and handling priority, completes the unified summary and integration of batch instructions, and generates complete batch single-channel power cut-off instruction data.

[0066] The working principle and effects of the above technical solution are as follows: By disassembling the isolation and handling content and eliminating redundant information, invalid text content is simplified, making the core handling information more focused and improving the overall efficiency of subsequent command conversion. Through standardized encoding conversion methods, natural language handling content is converted into control codes that the device can recognize, eliminating errors from manual command input and allowing the device to automatically execute corresponding power-off operations. By unifying the format and standard of control codes and eliminating abnormal codes, the problem of command execution failure caused by encoding errors is reduced, improving the overall compliance of control commands. Through precise binding of control codes to target channels, command mismatch and erroneous execution are prevented, ensuring the targeted nature of single-channel power-off operations. This solution can achieve automated and accurate conversion of fault isolation commands and ensure stable output of multi-channel batch power-off commands, effectively improving the reliability and accuracy of multi-power supply fault isolation control.

[0067] In one embodiment of the present invention, S33 includes:

[0068] The priority retrieval module reads the full-channel load priority file in the device's local storage area, traverses the rated load capacity and long-term operating load level parameters of each power supply channel in the file, completes the level division according to the difference in load carrying capacity, and generates multi-level load weight classification data.

[0069] The priority retrieval module collects real-time test status information of the device under test, identifies the current test process stage, matches the corresponding operating condition attributes of different operating stages, classifies and distinguishes three test states: no-load operation, full-load operation, and overload operation, and generates test condition classification data.

[0070] The priority retrieval module retrieves the preset working condition weight coefficient system, matches the corresponding standard weight coefficients to the running status corresponding to the test working condition classification data, completes the assignment of weight parameters for different test stages, and generates test stage weight configuration data.

[0071] The priority retrieval module integrates the multi-level load weight classification data with the weight configuration data during the testing phase, completes the corresponding binding of channel fixed load parameters and dynamic operating condition weight parameters, and generates integrated channel weight raw data.

[0072] The priority retrieval module organizes and stores the raw data of integrated channel weights, unifies the data storage format and sorting rules, centrally enters all channel weight parameters into the temporary data storage area, and generates a standardized channel weight aggregation data table.

[0073] The working principle and effects of the above technical solution are as follows: By assigning weight levels to each power supply channel based on its load-bearing capacity, the importance of the load on different channels can be accurately distinguished, providing a reliable basis for evaluating power outage sequence arrangements. By classifying and differentiating real-time test conditions of the equipment, weight standards can be adapted to different operating states, aligning with real-world test scenarios. By matching corresponding weight coefficients to various operating conditions, reasonable dynamic weight parameters can be assigned, improving the adaptability of weight determination. By integrating fixed load parameters and dynamic operating condition parameters, a more comprehensive basis for channel weight evaluation is formed, avoiding the one-sidedness caused by single-parameter evaluation. This approach balances the channel hardware load-bearing characteristics and real-time test status, while providing accurate data support for subsequent differentiated power outage sequence arrangement, effectively improving the rationality and scientific nature of multi-power supply fault handling sequence scheduling.

[0074] In one embodiment of the present invention, S35 includes:

[0075] The timing verification module extracts the initial timing data, splits the execution time information attached to all instructions according to a unified clock scale, groups and classifies all power-off instructions according to their time sequence, and generates statistical data of instruction groups arranged according to timing.

[0076] The timing verification module traverses each time group within the instruction group statistics data, counts the number of power-off instructions stacked at the same clock point, filters out abnormal group entries with multiple instructions stacked at a single clock point, and generates timing conflict point labeling data.

[0077] The timing verification module allocates idle clock segments sequentially along the original timing arrangement for conflicting instructions corresponding to the timing conflict point marking data, gradually increasing the execution interval of adjacent instructions, and generating intermediate timing data after interval correction.

[0078] The timing verification module repeatedly checks the clock arrangement status of the intermediate timing data and repeatedly checks for the problem of stacking of remaining implicit time until all instructions occupy their own non-overlapping execution periods, generating instruction data to be solidified with compliant timing arrangement.

[0079] The timing verification module locks the clock parameters and the correspondence between the instruction data to be solidified, integrates and encapsulates all instruction information to obtain timing-based cutoff control data, and then transfers the timing-based cutoff control data in batches to the control instruction buffer area.

[0080] The working principle and effects of the above technical solution are as follows: By using a unified clock to break down and categorize various power-off commands, the execution arrangement patterns of all commands can be quickly sorted out, improving the speed of finding timing conflict points. By checking each command item clustered at the same time, the positions of various timing conflicts are fully marked, reducing the probability of overlooking hidden timing vulnerabilities. By extending idle periods to increase the interval between command execution, the potential for instantaneous current surges caused by simultaneous power outages of multiple channels is eliminated, reducing the probability of short-term overloads on the power bus. Through multiple re-inspections to check for hidden arrangement problems, all potential hidden timing stacking risks are eliminated, avoiding command errors during subsequent on-site execution. It can stagger the execution nodes of power outages of faulty channels, weaken the electrical disturbances caused by power-off actions, and ensure that all disconnection commands are executed in an orderly manner, effectively improving the operational stability of multi-power supply fault isolation operations.

[0081] In one embodiment of the present invention, step S4 includes:

[0082] S41. Extract the timed disconnection control data from the control command buffer area. The domain division unit locks all the fault channels marked in the data as the center point of the area division. Retrieve the bus connection parameters, capacitor parallel connection parameters, and common ground connection parameters between the channels. Based on various coupling parameters, delineate the initial channel range that the power disturbance can be affected by the power outage action after the power outage action is triggered, and generate the initial power outage impact data of the entire domain.

[0083] S42. The boundary convergence calculation unit analyzes the withstand voltage parameters and isolation resistance values ​​of the isolation devices of each peripheral channel within the data of the initial power failure affecting the whole domain. Peripheral channels with parameter values ​​higher than the critical isolation standard are removed from the affected whole domain. The disturbance range is compressed layer by layer inward. After the range is compressed, fixed boundary points are defined, and the affected channel boundary data with marked boundary numbers and corresponding channel ranges are generated.

[0084] S43. The continuous simulation module loads the boundary data of the affected channel, and simulates the bus voltage fluctuation and instantaneous surge current change at the moment of power failure of the fault channel based on the circuit simulation model of the device under test. It calculates the power supply operation stability of each channel outside the boundary under the fluctuating conditions and generates single-channel power supply stability simulation calculation data.

[0085] S44. The classification and screening unit filters channel entries whose simulation results meet the continuous power supply standard from the single-channel power supply stability simulation calculation data. These channels are uniformly classified into the continuous power supply category, and the remaining channels that fall within the affected boundary are classified into the power outage control category. The information of the two types of channels is recorded in separate columns to generate a classification channel temporary record table.

[0086] S45. Organize all continuous power supply channel numbers, rated operating parameters, and current real-time load information in the temporary record table of classified channels. After removing duplicate channel entries, complete the orderly arrangement of entries and generate standardized continuous power supply channel list data. The list data is then transferred to the load dispatching front-end database.

[0087] The working principle and effects of the above technical solution are as follows: By combining the channel electrical coupling parameters to delineate the initial impact range of power outage disturbances, the system comprehensively covers the power fluctuation area caused by power outages, improving the completeness of the impact range determination. By relying on the isolation device parameters to progressively narrow the impact boundaries, peripheral channels unaffected by power outages are eliminated, reducing the over-amplification of the fault isolation range and avoiding unnecessary channel power outage operations. Circuit simulation mode simulates the electrical fluctuations at the moment of power outage, accurately verifying the power supply stability of each channel and improving the accuracy of continuous power supply channel determination. By classifying, sorting, and summarizing the channel operating states, the system clearly distinguishes between power outage control channels and continuous power supply channels. This approach can accurately pinpoint the actual impact area of ​​power outages, narrowing the coverage of fault isolation, while maximizing the preservation of the continuous operating state of normal channels, reducing unnecessary power interruptions during multi-power supply testing, and effectively improving the continuity and stability of the overall testing operation.

[0088] In one embodiment of the present invention, step S5 includes:

[0089] S51. Retrieve the continuous power supply channel list data from the load dispatching front-end database. The power calculation unit reads the real-time output power value and reserved redundant power value of the fault channel before the power failure. The two types of values ​​are added together to calculate the total idle power margin of the fault channel. The margin value is bound to the fault number and stored in the idle power storage entry to generate the summary data of idle power of the fault channel.

[0090] S52. The power distribution unit retrieves the channel load priority parameters attached to the continuous power supply channel list. The operating channels with higher priority are given priority to apply for idle power quotas, and the channels with lower priority are allocated the remaining idle power. The upper limit of the power allocated to a single channel does not exceed its own rated peak power. After the quotas of all channels are allocated, the supplementary details of single channel power are generated.

[0091] S53. Summarize all single-channel power supplementation details, add the original rated output power and the supplemented allocated power, update the new real-time operating power parameters of each continuous power supply channel, and archive the parameters in order of channel number to generate complete power dynamic rebalancing data.

[0092] S54. The global verification unit loads the power dynamic rebalancing data, simulates the bus power distribution conditions after the power redistribution in the full circuit simulation link, continuously collects the instantaneous output voltage change values ​​of each continuous power supply channel, calculates the specific amplitude of the voltage drop before and after the power distribution, and generates detailed statistical data of voltage drop in the entire channel.

[0093] S55. Compare the voltage drop amplitude of each channel with the allowable voltage drop limit value specified by the product in the full channel voltage drop statistics details. When the amplitude of all channels has not exceeded the limit value, a unified multi-channel parallel test safe operation confirmation signal is generated. The confirmation signal is pushed to the main control test terminal to complete the closed loop record of this overload protection test.

[0094] The working principle and effects of the above technical solution are as follows: By statistically analyzing the total power margin freed up after the faulty channel is shut down, the base of adjustable power is accurately calculated, providing reliable data support for power redistribution. By allocating spare power according to load priority and simultaneously constraining the power limit of a single channel, overload of channels is prevented, reducing device losses caused by power overload. By updating the operating power of each channel to form rebalancing data, idle power resources are rationally utilized, improving the actual utilization rate of the overall power resources. By simulating and reproducing the bus operating conditions after power redistribution, the voltage drop is statistically analyzed item by item, identifying potential power supply anomalies early and avoiding large-scale voltage instability caused by power redistribution. It can both absorb the surplus power released by the faulty channel and make up for the power gap of the normal channel, and stabilize the power supply indicators of the entire circuit, effectively improving the operational stability and test reliability of multiple power supplies during the fault handling phase.

[0095] According to one embodiment of the present invention, a multi-channel power output safety system with overload protection is provided, the system comprising:

[0096] One or more processors;

[0097] Memory, used to store one or more programs;

[0098] Wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the test method described in any one of the above.

[0099] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A safety test method for multiple power outputs with overload protection, characterized in that, The method includes: S1. Real-time electrical parameter acquisition of the power supply circuit of the device under test, obtaining the output voltage, output current and power loss data of each channel, performing sliding window filtering and baseline calibration on the acquired data, generating a standardized electrical feature sequence for each channel; threshold comparison and trend slope analysis are performed on the standardized electrical feature sequence to generate a multi-channel electrical state feature set; S2. Perform cross-channel correlation comparison analysis on the multi-channel electrical status feature set, use the time-series anomaly detection algorithm to identify the degree of abrupt deviation of electrical parameters in each channel, map the degree of deviation to the preset fault level classification system, and generate fault level label data; conduct a propagation risk assessment on the fault level label data, determine whether the anomaly has the conditions to spread to adjacent channels, and generate fault propagation risk assessment data. S3. Perform isolation strategy matching processing on the fault propagation risk assessment data, convert the risk assessment results in natural language form into machine-executable single-channel disconnection control commands, and generate single-channel power disconnection command data; perform execution timing arrangement on the single-channel power disconnection command data, and generate differentiated disconnection timing sequences by combining the load priority of each channel and the weight of the test phase, and generate time-sequential disconnection control data. S4. Perform execution domain delineation processing on the time-sequential disconnection control data. Using the fault channel as the isolation core, combine the electrical coupling relationship of adjacent channels to perform boundary convergence of the disconnection influence domain and generate boundary data of the affected channels. Perform power supply continuity verification on the boundary data of the affected channels, mark the channels not included in the influence domain as continuous power supply channels, and generate a list of continuous power supply channels. S5. Perform resource reallocation processing on the continuous power supply channel list data, allocate the power margin released by the faulty channel to the still running channels according to priority rules, and generate power dynamic rebalancing data; perform global stability verification on the power dynamic rebalancing data, and generate a multi-channel parallel test safe operation confirmation signal.

2. The method for testing the safety of multiple power outputs with overload protection according to claim 1, characterized in that, S1 includes: S11. Deploy distributed multi-channel sampling and acquisition units along 16 DC input channels. The acquisition units synchronously capture the raw sampling data of instantaneous output voltage, instantaneous output current and real-time power loss of each power supply circuit to generate a raw electrical sampling dataset for the entire channel. Simultaneously record the acquisition clock code value corresponding to each sampling and bind it to the end of the corresponding single sampling data to generate raw sampling summary data with acquisition timing identifier. S12. Retrieve the original electrical sampling dataset of the entire channel and send it to the filtering operation unit. Call the fixed-length sliding window algorithm to remove pulse clutter data layer by layer. After the removal is completed, connect to the baseline calibration operation link. The operation link completes the zero-point correction of the single-channel data based on the unloaded reference parameters of the entire channel and generates a standardized electrical feature sequence that is stored independently for each channel. Store the data in the feature cache storage space according to the channel number and generate a partitioned storage type standardized sequence storage data. S13. Retrieve partitioned standardized sequence storage data in batches from the feature cache storage space. The parameter comparison unit retrieves the factory-calibrated single-channel rated voltage threshold, rated current threshold and rated loss threshold. Perform numerical threshold comparison calculations for the standardized electrical feature sequence channel by channel. Simultaneously fit the change slope values ​​of the continuous sampling points of the sequence and generate single-channel threshold over-limit marker data and slope change statistics. S14. The data of single-channel threshold over-limit markers and slope change statistics are sent to the feature extraction and operation module. The module filters the parameter points with sharp increases in short-cycle parameters to extract overload precursor features and filters the parameter points with stable fluctuations in long-term intervals to extract steady-state operation features. The two types of features are bound to the corresponding channel numbers to complete the classification and collection, and generate independent feature sub-items storage data for each channel. S15. Integrate all independent feature items stored in each channel, remove duplicate and redundant feature items, and complete the cross-channel data aggregation. The aggregated data is then arranged in a structured manner according to the channel layout order to generate a complete multi-channel electrical status feature set. The multi-channel electrical status feature set is transferred to the feature database for storage and awaits subsequent calculations.

3. The method for testing the safety of multiple power outputs with overload protection according to claim 2, characterized in that, S13 includes: The parameter comparison unit traverses the feature cache storage space, reads the partitioned storage-style standardized sequence storage data in sequence according to the channel number, performs time-series alignment and organization on the batch retrieved sequence data, and generates ordered standardized sequence organization data; Read the electrical parameter standard system calibrated by the equipment at the factory, extract the critical values ​​of rated voltage, rated current and rated power loss of a single channel, and summarize them to form a standard dataset of channel electrical thresholds; The ordered standardized sequence data is compared frame by frame and channel by channel with the channel electrical threshold standard dataset to determine the state of the real-time operating parameters exceeding the rated parameters, and generate threshold comparison results data for each channel's single time node. Linear fitting is performed on the sequence data of continuously sampled time nodes to statistically analyze the rate of increase or decrease of electrical parameters within a fixed time period, calculate the dynamic change slope value of each channel parameter, and generate statistical data on the slope change of the entire channel. Collect abnormal threshold comparison results and normal operation results for each channel, uniformly mark the over-limit status of each channel, and integrate the information with the slope change statistics of the whole channel to generate complete single-channel threshold over-limit marking data and slope change statistics.

4. The method for testing the safety of multiple power outputs with overload protection according to claim 1, characterized in that, The S2 includes: S21. Read the multi-channel electrical state feature set completely from the feature database. Using the same sampling clock as the alignment reference, compare the voltage, current and loss feature values ​​of adjacent numbered channels at the same time node, record the result of the time sequence parameter difference one by one, and generate cross-channel parameter difference statistical ledger data after the full channel comparison is completed. S22. The timing operation module loads the timing anomaly detection algorithm program, analyzes the continuous timing parameter change pattern in the cross-channel parameter difference statistical ledger data line by line, and quantifies the specific value of parameter mutation deviation in each channel by accumulating the parameter offset in multiple consecutive sampling periods. After all the channel deviation values ​​are calculated, a full-channel parameter mutation deviation quantification dataset is generated. S23. The hierarchical mapping unit retrieves the pre-divided multi-level fault level classification scale, completes hierarchical matching according to the deviation value range of each channel within the full-channel parameter mutation deviation quantification dataset, and generates structured fault level label data after summarizing and organizing all records; the fault level label data is sorted by fault severity and stored in the hierarchical label temporary cache. S24. The risk assessment unit reads the fault level label data in the temporary cache of the graded label, retrieves the inter-channel line coupling impedance parameters and bus common load parameters recorded in the hardware layout of the device under test, and performs an inter-channel abnormal conduction probability calculation in combination with the fault severity level marked by the label, and generates detailed data of single-channel conduction probability calculation. S25. Collect all detailed data on single-channel transmission probability calculations, perform correlation merging calculations according to the arrangement of adjacent channels, divide the fault propagation intervals into three categories: high, medium, and low, bind and integrate the interval division results with the corresponding fault channel information, generate the final fault propagation risk assessment data, and transfer the fault propagation risk assessment data to the risk database for storage and backup.

5. The method for testing the safety of multiple power outputs with overload protection according to claim 1, characterized in that, The S3 includes: S31. Extract complete fault propagation risk assessment data from the risk database. The strategy retrieval unit retrieves the pre-stored hierarchical isolation and disposal rule base. It compares the propagation interval level marked in the risk assessment data with the isolation and disposal entries in the rule base one by one. It completes the automatic retrieval and matching of the corresponding isolation and disposal plan. The matching results are summarized to generate a risk and isolation plan matching list data. S32, The instruction conversion unit reads the risk and isolation scheme matching list data. The isolation treatment content recorded in natural language in the list is compiled into binary control code through the built-in encoding conversion logic. The compiled control code is bound to the target channel that needs to perform power-off action. A single-channel independent power-off control code is generated one by one. All codes are summarized to generate batch single-channel power cut-off instruction data. S33. The priority retrieval module loads the preset full-channel load priority file and the weight parameters of the current test stage of the device under test. The load priority file is divided into multiple load weight levels. The test stage weight is distinguished into three test conditions: no load, full load, and overload. The two types of parameters are uniformly stored in the weight temporary data table to generate the channel weight collection data table. S34. The timing orchestration module synchronously retrieves the single-channel power cut-off command data and the channel weight collection data table. The fault-related channels with low weight values ​​are given priority in arranging the power cut-off execution points, while the related channels with high weight values ​​are arranged with delayed power cut-off execution points. The execution clock node corresponding to each cut-off command is marked one by one, and the initial timing draft data of the single command binding execution time is generated. S35. Verify whether there is a power outage conflict between multiple channels of the same clock in the internal instruction execution clock of the initial timing data. Fine-tune the execution interval for instructions with overlapping times. After the interval is corrected, complete the solidification of the full instruction timing and generate the final timing cut-off control data. The timing cut-off control data is sent to the control instruction buffer.

6. The method for testing the safety of multiple power outputs with overload protection according to claim 5, characterized in that, S32 includes: Read the risk and isolation scheme matching list data in the cache area, decompose the content and split the fields of various isolation treatment contents in the list, and generate standardized isolation treatment text data; The built-in encoding conversion logic is invoked to parse and convert the standardized isolation treatment text data line by line, and compile the treatment content expressed in natural language into binary control code elements that the device can recognize in batches, generating the original power-off control code metadata. The original power outage control code metadata is standardized in terms of field regularization and format calibration, the code compilation length and instruction format are unified, invalid code content with compilation errors and format disorder is removed, and basic data for compliance control coding is generated. The basic data of compliance control codes are bound one by one with the corresponding target power supply channels within the list, so as to achieve accurate matching between control codes and execution channels and generate exclusive power outage control code data for each channel. Collect all single-channel dedicated power-off control code data, integrate and archive the data according to the channel arrangement order and handling priority, complete the unified summary and integration of batch commands, and generate complete batch single-channel power cut-off command data.

7. The method for testing the safety of multiple power outputs with overload protection according to claim 5, characterized in that, The S35 includes: Extract the initial timing data, split the execution time information of all instructions according to a unified clock scale, group and classify all power-off instructions according to time sequence, and generate statistical data of instruction groups arranged in time sequence. Iterate through the instruction group statistics for each time group, count the number of power failure instructions stacked at the same clock point, filter out abnormal group entries with multiple instructions stacked at a single clock point, and generate timing conflict point labeling data. For conflicting instructions corresponding to the timing conflict point marking data, idle clock segments are allocated sequentially along the original timing arrangement, and the execution interval of adjacent instructions is gradually increased to generate intermediate timing data after interval correction. Repeatedly verify the clock arrangement status of intermediate timing data, repeatedly check for the problem of stacking of remaining implicit time, until all instructions occupy their own non-overlapping execution periods, and generate instruction data to be solidified with compliant timing arrangement. The clock parameters and the correspondence between the fixed instruction data are uniformly locked, and the full instruction information is integrated and encapsulated to obtain the timing cut-off control data. The timing cut-off control data is then transferred in batches and stored in the control instruction buffer area.

8. The method for testing the safety of multiple power outputs with overload protection according to claim 1, characterized in that, The S4 includes: S41. Extract the timed disconnection control data from the control command buffer area. The domain division unit locks all the fault channels marked in the data as the center point of the area division. Retrieve the bus connection parameters, capacitor parallel connection parameters, and common ground connection parameters between the channels. Based on various coupling parameters, delineate the initial channel range that the power disturbance can be affected by the power outage action after the power outage action is triggered, and generate the initial power outage impact data of the entire domain. S42. Analyze the withstand voltage parameters and isolation resistance values ​​of the isolation devices of each peripheral channel within the data of the initial power outage affecting the entire domain. Peripheral channels with parameter values ​​higher than the critical isolation standard are removed from the affected domain. The disturbance range is compressed layer by layer inward. After the range is compressed, fixed boundary points are defined, and the affected channel boundary data with marked boundary numbers and corresponding channel ranges are generated. S43. Load the boundary data of the affected channel, and simulate the bus voltage fluctuation and instantaneous surge current change at the moment of power failure of the fault channel based on the circuit simulation model of the device under test. Calculate the power supply stability of each channel outside the boundary under the fluctuating conditions, and generate single-channel power supply stability simulation calculation data. S44. Select channel entries whose simulation results meet the continuous power supply standard from the single-channel power supply stability simulation calculation data, and generate a temporary record table of classified channels. S45. Organize all continuous power supply channel numbers, rated operating parameters, and current real-time load information in the classification channel temporary record table, generate standardized continuous power supply channel list data, and transfer the list data to the load dispatching front-end database.

9. The method for testing the safety of multiple power outputs with overload protection according to claim 1, characterized in that, The S5 includes: S51. Retrieve the continuous power supply channel list data from the load dispatching front-end database. The power calculation unit reads the real-time output power value and reserved redundant power value of the fault channel before the power failure. The two types of values ​​are added together to calculate the total idle power margin of the fault channel. The margin value is bound to the fault number and stored in the idle power storage entry to generate the summary data of idle power of the fault channel. S52. The power distribution unit retrieves the channel load priority parameters attached to the continuous power supply channel list. The operating channels with higher priority are given priority to apply for idle power quotas, and the channels with lower priority are allocated the remaining idle power. The upper limit of the power allocated to a single channel does not exceed its own rated peak power. After the quotas of all channels are allocated, the supplementary details of single channel power are generated. S53. Summarize all single-channel power supplementation details, add the original rated output power and the supplemented allocated power, update the new real-time operating power parameters of each continuous power supply channel, and archive the parameters in order of channel number to generate complete power dynamic rebalancing data. S54. The global verification unit loads the power dynamic rebalancing data, simulates the bus power distribution conditions after the power redistribution in the full circuit simulation link, continuously collects the instantaneous output voltage change values ​​of each continuous power supply channel, calculates the specific amplitude of the voltage drop before and after the power distribution, and generates detailed statistical data of voltage drop in the entire channel. S55. Compare the voltage drop amplitude of each channel with the allowable voltage drop limit value specified by the product in the full channel voltage drop statistics details. When the amplitude of all channels has not exceeded the limit value, a unified multi-channel parallel test safe operation confirmation signal is generated. The confirmation signal is pushed to the main control test terminal to complete the closed loop record of this overload protection test.

10. A multi-channel power output safety system with overload protection, characterized in that, The system includes: One or more processors; Memory, used to store one or more programs; Wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the test method according to any one of claims 1 to 9.

Images