A method and system for automatically testing a vehicle-mounted USB product at the end of a production line

Abstract

The application discloses a kind of vehicle-mounted USB product offline automatic test method and system, belong to the technical field of automatic test, this method includes: the station of the vehicle-mounted USB product to be tested is defined as to be scheduled station, when multiple to-be-scheduled stations initiate gear switching request, calculate power jump variable, each to-be-scheduled station is paired and is defined as first station and second station respectively, the power jump variable of first station, the average time consumption of gear switching and the current load power input shock attenuation function of second station, obtain the residual voltage deviation of first station to second station influence;Residual voltage deviation is used to establish conflict dependency graph;Determine the first test strategy of to-be-scheduled station based on conflict dependency graph, test vehicle-mounted USB product based on the first test strategy, and generate the second test strategy in the testing process, test the vehicle-mounted USB product not tested based on the second test strategy, the accuracy of voltage measurement can be improved by the application.

Description

Technical Field

[0001] This invention belongs to the field of automated testing technology, specifically relating to an automated testing method and system for vehicle-mounted USB products after they are taken off the production line. Background Technology

[0002] In power-on testing of automotive USB products, a multi-station parallel testing architecture is typically used to improve production efficiency. Each station shares a programmable power supply, a CC decoy control module, and data acquisition equipment. During the testing process, each station sequentially switches between different PD power levels and measures the product's output voltage, current, and other parameters at each level to verify whether the product's electrical performance meets specifications. However, because multiple stations share the same high-power DC power supply to simulate the constant power supply of a vehicle's KL30, when a product at one station switches power levels, its load power changes abruptly within a short time. This causes a drop or overshoot in the output voltage of the shared power bus. If other stations simultaneously or successively initiate power-switching requests, the superposition of power jumps from multiple stations creates a complex voltage fluctuation process on the power bus. This causes the product voltage value acquired by the station performing steady-state measurements to deviate from the actual output value, resulting in measurement errors.

[0003] To address the aforementioned issues, existing technologies increase the shared high-power DC power supply voltage to 14V. Then, an independent non-isolated or isolated high-power DC-DC regulator module is installed at the input of each workstation. The output voltage of all modules is uniformly set to simulate the typical voltage of a KL30 vehicle. This allows the shared 14V bus to function as a relatively high-voltage, low-impedance DC bus, meaning fluctuations in the load current at each workstation only cause voltage fluctuations on the 14V bus. Furthermore, since each workstation's DC-DC module has input voltage regulation and load regulation, its output 13.5V can remain stable as long as the input voltage is within its allowable range, thus achieving decoupling and isolation of the power bus. However, this solution requires configuring a DC-DC module for each workstation, increasing hardware costs and heat dissipation equipment requirements, and necessitates extensive debugging, increasing labor costs. Summary of the Invention

[0004] To address the aforementioned problems, this invention provides an automated testing method and system for automotive USB products after production line completion, thereby resolving the issues described in the background section.

[0005] To achieve the aforementioned objectives, this invention proposes an automated testing method for automotive USB products after they are removed from the production line, comprising: Construct an impact attenuation function, define the workstation where the vehicle USB product under test is located as the workstation to be scheduled, and when multiple workstations to be scheduled initiate a gear switching request, extract the target gear to be switched, the gear before switching, and the current load power of each workstation to be scheduled. Based on the power jump variable calculated from the gear position before switching and the target gear position to be switched, each work station to be scheduled is paired up and defined as the first work station and the second work station respectively. The power jump variable of the first work station, the average time of gear switching, and the current load power of the second work station are input into the impact attenuation function to obtain the residual voltage deviation of the first work station on the second work station. If the residual voltage deviation is greater than the preset voltage deviation threshold, then a directed edge between workstations is established based on the residual voltage deviation, and the residual voltage deviation is used as the edge weight to construct a conflict dependency graph. A first test strategy is determined based on the conflict dependency graph. A gear shift test is performed on at least one workstation based on the first test strategy. After the gear shift test is completed at the workstation, the conflict dependency graph is corrected to obtain a corrected dependency graph. A second test strategy for scheduling untested in-vehicle USB products is generated based on the corrected dependency graph. The untested in-vehicle USB products are tested based on the second test strategy.

[0006] This invention also provides an automated testing system for off-line testing of in-vehicle USB products. This system is used to implement the above-described method and includes: The function construction module constructs an impact attenuation function, defines the workstation where the vehicle USB product under test is located as a workstation to be scheduled, and extracts the target gear to be switched, the gear before switching, and the current load power of each workstation when multiple workstations to be scheduled initiate a gear switching request. The deviation calculation module calculates the power jump variable based on the gear position before switching and the target gear position to be switched. It pairs each work position to be scheduled and defines it as the first work position and the second work position respectively. It inputs the power jump variable of the first work position, the average time of gear switching, and the current load power of the second work position into the impact attenuation function to obtain the residual voltage deviation of the first work position on the second work position. The graph construction module, if the residual voltage deviation is greater than the preset voltage deviation threshold, establishes directed edges between workstations based on the residual voltage deviation and uses the residual voltage deviation as the edge weight to construct a conflict dependency graph. The strategy scheduling module determines a first test strategy for the workstations to be scheduled based on a conflict dependency graph, performs gear switching tests on at least one workstation to be scheduled based on the first test strategy, corrects the conflict dependency graph after the workstation to be scheduled completes the gear switching test to obtain a corrected dependency graph, generates a second test strategy for scheduling untested in-vehicle USB products based on the corrected dependency graph, and tests the untested in-vehicle USB products based on the second test strategy.

[0007] The beneficial effects of this invention are as follows: This invention first constructs an impulse attenuation function, then calculates the power jump variable and pairs workstations together, using the impulse attenuation function to calculate the residual voltage deviation, thereby predicting the degree of mutual interference between workstations. By constructing a conflict dependency graph and establishing directed edge relationships between workstations based on the residual voltage deviation, the impulse impact between workstations is transformed into schedulable dependencies, thus ensuring both measurement accuracy and maximizing the efficiency of multi-workstation parallel testing. Through the above technical solution, this invention achieves voltage fluctuation prediction and scheduling optimization for multi-workstation shared power supply without increasing hardware costs, solving the voltage fluctuation measurement error problem caused by shared power supply, and avoiding the increased hardware costs associated with configuring a DC-DC module for each workstation. Attached Figure Description

[0008] Figure 1 This is a flowchart of the steps of an automated testing method for vehicle-mounted USB products after production line completion according to the present invention; Figure 2 This is a comparison chart of voltage measurement accuracy under multi-station concurrent conditions according to the present invention; Figure 3 This is a schematic diagram of the structure of an automated testing system for vehicle-mounted USB products after production line according to the present invention. Detailed Implementation

[0009] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0010] This application's multi-station power level switching scheduling method is applied to an automated testing system for automotive USB products. The system's hardware units include a programmable low-voltage DC power supply, a programmable low-voltage DC load, a programmable CC decoy, a barcode scanner, a PLC module, and a host computer. The programmable low-voltage DC power supply simulates the constant power supply of a car's KL30 engine to power the product under test. The programmable low-voltage DC load simulates electrical equipment as the output load. The programmable CC decoy automatically switches the power levels of the USB PD product. The barcode scanner is used to bind the product part number and serial number for MES traceability. The PLC module interacts with the host computer and controls the test bench. The host computer is responsible for executing the entire testing process and storing the test data.

[0011] The system's software units include an MES module, a logging module, a front-end interface, a test automation module, a CC control module, an OTG switching module, a power control module, a load control module, and a PLC control module. The MES module binds, stores, and traces product information and test results; the test automation module runs automated testing software and completes data acquisition and interpretation; the CC control module automatically switches PD output levels; the OTG switching module switches uplink and downlink for products supporting OTG functionality; the power control module and load control module automatically adjust voltage, current, and load power, respectively; and the PLC control module responds to commands from the host computer and controls the rack-mounted equipment.

[0012] The testing system comprises eight workstations, each sharing programmable power supply and load resources. Test fixtures are categorized into single-process and dual-process workstations based on product type. Single-process workstations are used for testing USB Chargers and USB PD products, while dual-process workstations are used for testing USB Hub products. The testing process includes product clamping, barcode scanning and binding, automated test execution, result interpretation and uploading, and product distribution. For USB Chargers and USB PD products, a single-process test is performed, automatically verifying product programming information, current and voltage parameters, and overload protection functions. PD products also require a power level switching test. For USB Hub products, a dual-process test is performed: the forward process completes basic information and electrical performance testing, while the reverse process completes OTG functionality and USB-C port switching tests. After testing, the system automatically determines the test results and uploads them to the MES system.

[0013] In actual production testing, the eight workstations typically operate concurrently. When multiple workstations simultaneously initiate PD (Power Distribution) level switching requests, since all workstations share the same programmable power bus, a level switching at one workstation can cause transient fluctuations in the power bus voltage. These fluctuations propagate through the power bus to other workstations under testing, affecting the accuracy of their steady-state measurements. To improve the accuracy of model fitting and voltage compensation, this application optimizes the design for scenarios where voltage drops are caused by level boosting and power increases.

[0014] like Figure 1 As shown, an automated testing method for vehicle-mounted USB products after production line completion includes: S1: Construct an impact attenuation function, define the workstation where the vehicle USB product under test is located as the workstation to be scheduled, and when multiple workstations to be scheduled initiate a gear switching request, extract the target gear to be switched, the gear before switching, and the current load power of each workstation to be scheduled.

[0015] In this embodiment, the impact attenuation function is constructed, including: An initial function is constructed, which includes a fundamental coefficient factor characterizing the initial voltage drop amplitude caused by a unit power jump, a time constant factor characterizing the characteristic time scale of the voltage recovery process, and a load sensitivity coefficient factor characterizing the amplification effect of the second station load power on the voltage drop. A training set is constructed, which includes the power jump variable of the first station, the current load power of the second station, the data acquisition timestamp, and the residual voltage deviation. The power jump variable is the difference in output power of the first station before and after the gear switching, and the residual voltage deviation is the absolute value of the difference between the target output voltage value and the voltage value at the current acquisition time.

[0016] The initial function is fitted based on the training set to obtain the basic coefficient factor, time constant factor, and load sensitivity coefficient factor, so as to construct the impact decay function.

[0017] When constructing the training set, two qualified standard reference products were selected and labeled as Standard Reference Product A and Standard Reference Product B, respectively. Standard Reference Product A was mounted at the first station, and Standard Reference Product B was mounted at the second station. The first station serves as the impact source station, responsible for performing the gear switching operation to generate a power jump, while the second station serves as the observation station, responsible for measuring the voltage fluctuation of its own output.

[0018] The standard reference product A at the first station is subjected to multiple switching operations at different power levels, including 5V, 9V, 12V, 15V, and 20V each time. For each power level switching combination, multiple different load power levels are set at the second station, such as 10W, 30W, 50W, and 70W. During each power level switching at the first station, transient voltage waveforms are acquired at the second station for 50ms. Through this method, multiple sets of transient voltage waveform data under different power jumps and different load power combinations are obtained.

[0019] After data acquisition, the residual voltage deviation at multiple time points is recorded in the transient voltage waveform. The residual voltage deviation is the absolute value of the difference between the target output voltage value and the current voltage value at the output terminal of the product at the second station. The target output voltage value is the voltage value that the output port of the USB product under test should output. The power jump variable for each switch is recorded. The power jump variable is the difference in output power of the first station before and after the gear switch.

[0020] Using power jump variables, load power of the second station, average switching time, and residual voltage deviation as experimental data, multiple sets of experimental data were obtained and integrated into a training set. Based on the training set, a pre-constructed initial function was fitted to obtain the impulse decay function. The impulse decay function describes the decay process of the impact of the first station's switching on the output voltage of the second station over time. Finally, the average switching time was statistically calculated based on the training set. Specifically, for each switching operation performed by the first station in the training set, the starting point is the moment the switching command is issued, and the ending point is the moment when the residual voltage deviation in the transient voltage waveform of the second station first enters and stabilizes within the preset steady-state allowable band (e.g., ±1% of the target voltage). This time period is the single switching time. Then, all switching records in the training set are traversed, and the switching time for each switch is identified from the recorded transient waveform data according to the above definition. The average value is taken as the average switching time.

[0021] The impact attenuation function is specifically as follows: ,in, This refers to the deviation of the output voltage of the product at the second workstation from the target output voltage value. The base coefficient represents the initial voltage drop caused by a unit power jump. The power jump variable of the first station. This represents the time interval from the completion of the switch from the first workstation to the current evaluation time. In subsequent actual use, the average shift time will be factored into the calculations. The time constant represents the characteristic time scale of the voltage recovery process; the larger the value, the slower the recovery. The load sensitivity coefficient represents the amplification effect of the load power at the second station on voltage drops. This is the load power of the second station. The rated power is the maximum allowable load power of a single station in the testing system.

[0022] During the fitting process, the power jump at the first station, the load power at the second station, and the residual voltage deviation were used as training data, and the least squares method was employed for fitting. The basic coefficients, time constant, and load sensitivity coefficient were obtained through iterative calculation of the mean square error between the predicted and actual measured values.

[0023] In the impulse decay function, the exponential decay term reflects the system's recovery from voltage disturbance to steady state. The time constant is determined by the power supply internal resistance, output capacitance, and the response characteristics of the product's internal voltage regulation circuit. The product of the fundamental coefficient and the power jump variable reflects the initial voltage disturbance caused by the sudden change in load current through the power supply internal resistance and bus impedance. This disturbance is transmitted from the product's input to the output. The load amplification term reflects the impedance coupling effect when workstations share the power bus. The larger the load at the observed workstation, the more significant the voltage drop across the common bus impedance caused by its branch current, leading to amplified perception of voltage fluctuations at that workstation. The load sensitivity coefficient characterizes the bus impedance distribution and the coupling strength between workstations.

[0024] S2: Calculate the power jump variable based on the gear position before switching and the target gear position to be switched. Pair each work position to be scheduled into two and define them as the first work position and the second work position respectively. Input the power jump variable of the first work position, the average time of gear switching, and the current load power of the second work position into the impact attenuation function to obtain the residual voltage deviation of the first work position on the second work position.

[0025] During the formal testing phase, when multiple workstations initiate gear switching requests, these workstations are added to the scheduling queue as workstations to be scheduled. Then, the current status parameters of each workstation in the scheduling queue are read, including the workstation number, the target gear to be switched to, the previous gear, and the current load power.

[0026] Substituting the current power jump of the first station, the average time of gear switching, and the load power of the second station into the impulse attenuation function, the residual voltage deviation of the second station is calculated. The residual voltage deviation represents the magnitude of interference that the output voltage of the product at the second station will still experience after the first station switches, following the average time of gear switching, and when the second station is ready to switch. Similarly, the residual voltage deviation of the second station switching on the first station is calculated in reverse.

[0027] S3: If the residual voltage deviation is greater than the preset voltage deviation threshold, then a directed edge between workstations is established based on the residual voltage deviation, and the residual voltage deviation is used as the edge weight to construct a conflict dependency graph.

[0028] The residual voltage deviations of the first and second stations are compared with a preset voltage deviation threshold, which is determined according to the voltage accuracy requirements in the test specification. When the residual voltage deviation of the first station is greater than the voltage deviation threshold, it indicates that switching the first station first will interfere with the steady-state measurement of the product at the second station. In this case, a directed edge from the first station to the second station is added to the conflict dependency graph, with the edge weight set to the value of the residual voltage deviation. Similarly, a directed edge from the second station to the first station can be established.

[0029] The system performs the above calculations on all workstations in the scheduling queue and finally completes the construction of the conflict dependency graph. Each node in the graph represents a workstation to be scheduled, and the directed edges represent the influence relationship between workstations.

[0030] S4: Determine the first test strategy for the workstation to be scheduled based on the conflict dependency graph, perform gear switching test on at least one workstation to be scheduled based on the first test strategy, modify the conflict dependency graph after the workstation to be scheduled completes the gear switching test to obtain the modified dependency graph, generate a second test strategy for scheduling untested vehicle USB products based on the modified dependency graph, and test the untested vehicle USB products based on the second test strategy.

[0031] In this embodiment, the first test strategy for determining the workstation to be scheduled based on the conflict dependency graph includes: The in-degree and out-degree of the workstation to be scheduled are calculated based on the conflict dependency graph. The in-degree is the number of directed edges in the conflict dependency graph pointing to the workstation to be scheduled, and the out-degree is the number of directed edges originating from the workstation to be scheduled in the conflict dependency graph. Workstations to be scheduled with an in-degree of zero are identified as candidate workstations. The out-degree of the candidate workstations is obtained, and the blocking relief component is calculated based on the out-degree and the preset relief weight coefficient.

[0032] All workstations with an in-degree of zero are designated as candidate workstations. A candidate workstation represents a workstation that can be safely switched without being affected by other workstations in the queue. For each candidate workstation, the number of directed edges originating from it is counted, defined as its out-degree. The out-degree count indicates how many other workstations awaiting scheduling are currently blocked by this candidate workstation. For example, if the current queue contains 5 workstations A, B, C, D, and E, and the conflict dependency graph contains edges A to B, A to C, and A to D, with workstations A and E both having an in-degree of zero, and workstation A having an out-degree of 3, then workstations B, C, and D will be affected by workstation A.

[0033] Next, the blocking relief component is calculated: Blocking relief component = 1 + Out-degree * Relief weight coefficient. The relief weight coefficient is a preset constant that remains unchanged during the test. The relief weight coefficient is a dimensionless scheduling parameter used to characterize the effect of the candidate workstation on unblocking other workstations to be scheduled. The larger the relief weight coefficient, the greater the influence of the candidate workstation's out-degree on its blocking relief component. The relief weight coefficient is greater than 0. The relief weight coefficient can be determined as follows: multiple candidate relief weight coefficients are set for the test system, and a comprehensive scheduling score is calculated based on each candidate relief weight coefficient to generate a test strategy. The calculation method of the comprehensive scheduling score will be introduced later. The test completion time corresponding to each test strategy is statistically analyzed. Among the candidate relief weight coefficients whose voltage deviation meets the voltage deviation threshold requirement, the candidate relief weight coefficient with the shortest corresponding test completion time is determined as the final relief weight coefficient used. The larger the out-degree, the greater the impact of the candidate workstation. When the workstation completes the switch, the system will remove it from the conflict dependency graph, and all directed edges originating from that workstation will be deleted simultaneously. The in-degree of the blocked workstation to be scheduled will decrease accordingly, possibly becoming zero, thus unlocking it and allowing it to be selected by the system to participate in subsequent tests. Therefore, prioritizing the switching of workstations with higher out-degrees can release more candidate workstations in the next round of scheduling, improving the production line's concurrency efficiency. In the example above, if workstation A is switched first, A is removed from the graph after the switch, edges A to B, A to C, and A to D disappear, and the in-degrees of workstations B, C, and D all become zero, becoming candidate workstations along with workstation E. A maximum of four workstations can be scheduled simultaneously in the next round. If workstation E is switched first, E is removed from the graph after the switch, but B, C, and D are still blocked by A, and in the next round, there is still only workstation A as a candidate workstation.

[0034] The average steady-state time of candidate workstations is obtained based on the training set. The average steady-state time is the time it takes for the voltage to return to steady state after the candidate workstation is switched. The ratio of the preset response benchmark value to the average steady-state time is calculated to obtain the response speed component. The blockage release component and the response speed component are weighted and summed to obtain the comprehensive scheduling score. Candidate workstations are selected in descending order of comprehensive scheduling score to obtain the first test strategy including the test order.

[0035] The average steady-state time is obtained from statistics in the test set. Specifically, it involves selecting multiple time intervals between switching from a gear position to voltage stability when testing the same type of automotive USB product at the same workstation in the past, and calculating the average of these multiple time intervals as the average steady-state time. The response speed component = response baseline value / average steady-state time. The response baseline value is a preset constant, for example, set to 100ms. The shorter the average steady-state time, the faster the workstation can reach steady state after switching when testing the current product, the faster it can be removed from the conflict dependency graph, and thus the faster it can unlock subsequent workstations.

[0036] The system calculates a comprehensive scheduling score, which is calculated as follows: Comprehensive Scheduling Score = Blockage Relief Component * First Weighting Coefficient + Response Speed ​​Component * Second Weighting Coefficient. The first and second weighting coefficients are preset constants.

[0037] In this embodiment, a second test strategy is generated during the testing process, including: Delete the nodes of the corresponding candidate workstations and their associated edges in the conflict dependency graph to obtain the corrected dependency graph. Based on the corrected dependency graph, recalculate the in-degree of the remaining workstations to be scheduled. If there are workstations to be scheduled with an in-degree of zero, substitute the power jump variable of the workstations that have completed the switching, the elapsed time from the switching completion time of the workstations that have completed the switching to the current time, and the current load power of the workstations with an in-degree of zero into the impact attenuation function to calculate the current residual voltage deviation.

[0038] When a candidate workstation completes its gear switching, the node corresponding to that workstation is removed from the conflict dependency graph, along with all directed edges originating from and pointing to that node, thus obtaining a corrected dependency graph. After deletion, the system recalculates the in-degree of the remaining workstations. Some workstations that originally had non-zero in-degrees have their in-degrees reduced to zero because the edges pointing to them were deleted. However, the workstations that have completed the switching generated power surges at the time of switching. Although these workstations have been removed from the graph, the resulting power bus voltage fluctuations may not have fully decayed to a safe level. Therefore, the system obtains the power jump variable of the workstations that have completed the switching and the elapsed time from the time of switching completion to the current time. Substituting the power jump variable, the elapsed time, and the current load power of the workstation to be scheduled into the surge decay function, the system calculates the current residual voltage deviation of the workstation to be scheduled.

[0039] If the current residual voltage deviation is less than or equal to the voltage deviation threshold, the workstation with an in-degree of zero will be added to the next round of candidate switching workstations for scheduling. Otherwise, the calculation will be performed again after a preset time until the residual voltage deviation is less than or equal to the voltage deviation threshold, so as to generate the second test strategy.

[0040] When the residual voltage deviation is less than or equal to the voltage deviation threshold, it indicates that the power surge caused by the candidate workstation has attenuated to a level that does not affect the measurement accuracy of the workstation to be scheduled, and the workstation to be scheduled can participate in the next round of comprehensive scheduling score calculation. If the residual voltage deviation is greater than the voltage deviation threshold, the calculation will be performed again after a preset time until the residual voltage deviation is less than or equal to the voltage deviation threshold, and then the workstation to be scheduled will be included as a candidate workstation in the comprehensive scheduling score calculation.

[0041] In this embodiment, if there are no workstations to be scheduled with an in-degree of zero in the conflict dependency graph or the modified dependency graph, the immunity score is calculated for all workstations to be scheduled in the conflict dependency graph or the modified dependency graph, and an ordered candidate sequence is obtained by sorting the immunity scores from high to low.

[0042] In this embodiment, a robustness score is calculated for all workstations to be scheduled in the conflict dependency graph or the modified dependency graph, including: Summing the weights of the directed edges from the workstation to be scheduled and from other workstations to itself, we obtain the sum of the weights of the two-way edges. We locate the minimum value of the sum of the weights of the two-way edges of all workstations, as well as the minimum value of the power jump variable of all workstations. Based on the minimum value of the sum of the weights of the two-way edges, the weights of the two-way edges of the workstation to be scheduled, the minimum value of the power jump variable, and the power jump variable of the workstation to be scheduled, we calculate the immunity score.

[0043] When the in-degree of all workstations in the conflict dependency graph or modified dependency graph is greater than zero, it indicates that there are dependencies between all workstations. In this case, for each workstation to be scheduled, the sum of the bidirectional edge weights with all other workstations is calculated. For example, for workstation i to be scheduled, the sum of the bidirectional edge weights is the sum of the weights of all directed edges from workstation i to other workstations in the conflict dependency graph, plus the sum of the weights of all directed edges from other workstations to workstation i. The minimum value is taken among the sums of the bidirectional edge weights of all workstations in the current queue, and denoted as the minimum bidirectional sum. The minimum value is also taken among the minimum power jump variables of all workstations in the current queue, denoted as the minimum power jump variable. Then, the immunity score is calculated based on the following formula: .in, Representative workstation The immunity score and These are the first and second sorting weight coefficients, respectively, and are preset constants. This represents the minimum sum of the bidirectional edge weights of all workstations in the current scheduling queue. For workstations The sum of the weights of the two-way edges, This represents the minimum value of the power jump variable. For workstations The power jump variable. This formula normalizes the power by dividing the minimum value by the current value, and calculates the power jump variable for each station. Then, they are sorted in descending order of scores, eventually forming an ordered candidate sequence.

[0044] The significance of the above calculations is that a workstation with a smaller sum of bidirectional edge weights indicates a weaker mutual impact with other workstations, making it easier to combine with other workstations to form a compatible parallel batch. A workstation with a smaller power jump variable indicates a smaller disturbance to the power supply during switching. The system sorts all workstations to be scheduled from largest to smallest based on their immunity score, forming an ordered candidate sequence.

[0045] A search algorithm based on an ordered candidate sequence is used to divide the workstations to be scheduled into test batches, and synchronous gear switching instructions are issued to each workstation in the test batch.

[0046] In this embodiment, a search algorithm is executed based on an ordered candidate sequence to divide the workstations to be scheduled into test batches, including: Construct empty first set, second set and third set. Fill the second set with all the workstations to be scheduled in the ordered candidate sequence. Take out the workstations to be scheduled from the second set in turn and add the taken workstations to the first set.

[0047] Calculate the update superimposed impact on each original workstation in the first set after it is added, and the cumulative impact on the workstation to be scheduled. Calculate the compensation limit threshold based on the training set. If both the update superimposed impact and the cumulative impact are less than the compensation limit threshold, the workstation to be scheduled is retained in the first set; otherwise, the workstation to be scheduled is moved to the third set.

[0048] Initially, the first set is empty, the second set includes all workstations to be scheduled, and the third set is empty. Workstations to be scheduled are sequentially selected from the second set according to the ordered candidate sequence. Each workstation to be scheduled is added to the first set, and the update-cumulative impact on each existing workstation in the first set after addition is calculated. The update-cumulative impact is the sum of the edge weights pointing to itself from all existing workstations to the newly added workstation in the first set (excluding itself). The cumulative impact of the newly added workstation is the sum of the edge weights pointing to the newly added workstation from all existing workstations in the first set. If both the update-cumulative impact and the cumulative impact are less than the compensation limit threshold, the newly added workstation is retained in the first set; otherwise, it is moved back to the third set, and workstations to be scheduled are continuously drawn from the second set for addition testing.

[0049] The compensation limit threshold is determined using the following method: During the calibration of the impulse attenuation function, the power jump at the switching station is gradually increased, causing the residual voltage deviation to increase from small to large. For each power jump, the predicted residual between the predicted value of the impulse attenuation function and the actual measured voltage waveform is recorded. When the standard deviation of the predicted residual exceeds 1 / 10 of the voltage deviation threshold, it indicates that the impulse amplitude is too large, the system has entered the nonlinear response region, or the internal voltage regulation circuit or current limiting protection circuit of the product under test has intervened, causing the exponential decay model of the attenuation function to become inaccurate. The corresponding residual voltage deviation value at this time is calibrated as the compensation limit threshold. As long as the superimposed impulse between stations does not exceed the compensation limit threshold, subsequent voltage compensation is considered to be within the reliable range.

[0050] To improve search efficiency, this embodiment identifies the workstation with the largest cumulative impact after being added to the first set, defining it as the target workstation. The cumulative impact is defined as the impact accumulation value. Simultaneously, the edge weights of each remaining workstation in the second set relative to the target workstation are recorded, and the smallest edge weight is designated as the target edge weight. When the sum of the cumulative impact value and the target edge weight is greater than or equal to the compensation limit threshold, it indicates that adding any remaining workstation to the second set would result in a cumulative impact exceeding the compensation limit threshold. At this point, the search is terminated, and all workstations in the second set are moved to the third set.

[0051] When the second set is empty, configure the first set as a test batch, delete the workstations to be scheduled contained in the test batch, move the workstations to be scheduled in the third set to the second set, and repeat the above steps until all workstations to be scheduled are sorted.

[0052] When the second set is empty, the first set is defined as a test batch. The workstations to be scheduled in the test batch are deleted, and the workstations to be scheduled in the third set are moved to the second set. Then, from the second set, the workstations to be scheduled are added back to the first set in the order of the ordered candidate sequence. This process is repeated until all batches of workstations to be scheduled are divided.

[0053] Finally, the test batches are sorted according to their generation order, and then measured in batches according to the sorting results. For the workstations to be scheduled within a test batch, a gear switching command is simultaneously issued. After each workstation completes the switching and enters a steady state, the voltage sequence at the output terminal of each workstation is collected. For any test workstation j within a batch, the voltage measurement of workstation j is affected by the power jump generated during the switching of other workstations k within the same batch. Therefore, the voltage value collected by workstation j needs to be compensated and corrected. During correction, the superimposed predicted impact deviation experienced by workstation j at each sampling time t is calculated using the impact attenuation function. The predicted impact deviation generated by a single workstation k on workstation j is calculated by substituting the power jump of workstation k, the elapsed time from the completion of workstation k's switching to sampling time t, and the current load power of workstation j into the impact attenuation function. This value represents the residual deviation caused by the switching behavior of workstation k on the output voltage of workstation j at sampling time t.

[0054] The superimposed predicted impact deviation is equal to the sum of the predicted impact deviations of all stations k (excluding station j) within the batch, i.e., the total voltage interference experienced by station j from all other stations at that moment. Then, the original acquired voltage value of station j at sampling time t is added to the superimposed predicted impact deviation to correct the test results of station j. In this application, the superimposed predicted impact deviation is the predicted value of the voltage drop deviation caused by voltage increases or power increases at other stations, and its value is positive. Since this voltage drop causes the original acquired voltage value of station j to be lower than the actual output voltage value when there is no interference, compensation is performed by increasing the acquired voltage value; that is, the compensated voltage value is equal to the sum of the original acquired voltage value and the superimposed predicted impact deviation. The voltage compensation described in this application only applies to voltage drop situations caused by voltage increases or power increases, and does not compensate for voltage overshoot situations that may be caused by voltage decreases or power reductions.

[0055] To verify the effectiveness of the technical solution proposed in this application, a multi-station concurrent test comparison experiment was designed. The experiment used eight stations to simultaneously conduct offline testing on the in-vehicle USB product, employing both the existing multi-station direct concurrent switching method and the scheduling compensation scheme based on the impulse attenuation function proposed in this application. During the experiment, the number of stations simultaneously initiating gear shifting requests was gradually increased, starting with a single station switching and progressing to eight stations switching simultaneously. For each concurrency level, the deviation between the measured output voltage value of the product at each station and the standard reference voltage value was recorded, and the measurement accuracy was statistically analyzed. Figure 2 As shown, with the existing technology, the measurement accuracy gradually decreases as the number of concurrent workstations increases. However, with the technical solution of this application, the measurement accuracy remains high even in the extreme case where all 8 workstations switch simultaneously.

[0056] like Figure 3 As shown, the present invention also provides an automated testing system for vehicle-mounted USB products after production line completion. This system is used to implement the above-described method and includes: The function construction module constructs an impact attenuation function, defines the workstation where the vehicle USB product under test is located as the workstation to be scheduled, and extracts the target gear to be switched, the gear before switching, and the current load power of each workstation when multiple workstations to be scheduled initiate gear switching requests.

[0057] The deviation calculation module calculates the power jump variable based on the gear position before switching and the target gear position to be switched. It pairs each work position to be scheduled and defines it as the first work position and the second work position respectively. It inputs the power jump variable of the first work position, the average time of gear switching, and the current load power of the second work position into the impact attenuation function to obtain the residual voltage deviation of the first work position on the second work position.

[0058] The graph construction module establishes directed edges between workstations based on the residual voltage deviation if the residual voltage deviation is greater than the preset voltage deviation threshold, and uses the residual voltage deviation as the edge weight to construct a conflict dependency graph.

[0059] The strategy scheduling module determines a first test strategy for the workstations to be scheduled based on a conflict dependency graph. Based on the first test strategy, it performs gear switching tests on at least one workstation to be scheduled. After the gear switching tests are completed at the workstations to be scheduled, it corrects the conflict dependency graph to obtain a corrected dependency graph. Based on the corrected dependency graph, it generates a second test strategy for scheduling untested in-vehicle USB products. Based on the second test strategy, it tests the untested in-vehicle USB products.

[0060] It should be noted that the various threshold parameters, weighting coefficients, and normalization methods involved in this invention are all technical means that can be determined by those skilled in the art through conventional experimental calibration and parameter tuning based on the hardware configuration, product specifications, and test accuracy requirements of the actual test system. There is no issue of insufficient disclosure. For example, the voltage deviation threshold can be set according to the voltage measurement accuracy requirements in the test specification; the compensation limit threshold is calibrated by gradually increasing the power jump variable and monitoring the standard deviation of the predicted residual; the first weighting coefficient and the second weighting coefficient can be selected after statistical analysis of multiple sets of experimental data to achieve the optimal match between the comprehensive scheduling score and the actual test efficiency. None of these affect the technical effect of this invention. Furthermore, the formula expressions involved in this invention are simplified representations for ease of understanding of the technical solutions. In actual engineering implementation, those skilled in the art will use conventional boundary condition judgment and outlier handling methods to avoid possible division by zero situations. These are conventional technical means in the field and do not need to be detailed one by one in this invention.

Claims

1. An automated testing method for off-line testing of in-vehicle USB products, characterized in that, include: Construct an impact attenuation function, define the workstation where the vehicle USB product under test is located as the workstation to be scheduled, and when multiple workstations to be scheduled initiate a gear switching request, extract the target gear to be switched, the gear before switching, and the current load power of each workstation to be scheduled. Based on the power jump variable calculated from the gear position before switching and the target gear position to be switched, each work station to be scheduled is paired up and defined as the first work station and the second work station respectively. The power jump variable of the first work station, the average time of gear switching, and the current load power of the second work station are input into the impact attenuation function to obtain the residual voltage deviation of the first work station on the second work station. If the residual voltage deviation is greater than the preset voltage deviation threshold, then a directed edge between workstations is established based on the residual voltage deviation, and the residual voltage deviation is used as the edge weight to construct a conflict dependency graph. A first test strategy is determined based on the conflict dependency graph. A gear shift test is performed on at least one workstation based on the first test strategy. After the gear shift test is completed at the workstation, the conflict dependency graph is corrected to obtain a corrected dependency graph. A second test strategy for scheduling untested in-vehicle USB products is generated based on the corrected dependency graph. The untested in-vehicle USB products are tested based on the second test strategy.

2. The method according to claim 1, characterized in that, Constructing the impact attenuation function includes: An initial function is constructed, which includes a basic coefficient factor characterizing the initial voltage drop amplitude caused by a unit power jump, a time constant factor characterizing the characteristic time scale of the voltage recovery process, and a load sensitivity coefficient factor characterizing the amplification effect of the load power of the second station on the voltage drop. A training set is constructed, which includes the power jump variable of the first station, the current load power of the second station, the data acquisition timestamp, and the residual voltage deviation. The power jump variable is the difference in output power of the first station before and after the gear switching, and the residual voltage deviation is the absolute value of the difference between the target output voltage value and the voltage value at the current acquisition time. The initial function is fitted based on the training set to obtain the basic coefficient factor, time constant factor, and load sensitivity coefficient factor, so as to construct the impact decay function.

3. The method according to claim 2, characterized in that, The first test strategy for determining the workstations to be scheduled based on the conflict dependency graph includes: The in-degree and out-degree of the workstation to be scheduled are calculated based on the conflict dependency graph. The in-degree is the number of directed edges in the conflict dependency graph pointing to the workstation to be scheduled, and the out-degree is the number of directed edges originating from the workstation to be scheduled in the conflict dependency graph. The workstation to be scheduled with an in-degree of zero is located as a candidate workstation. The out-degree of the candidate workstation is obtained. The blocking relief component is calculated based on the out-degree and the preset relief weight coefficient. The average steady-state time of candidate workstations is obtained based on the training set. The average steady-state time is the time it takes for the voltage to return to steady state after the candidate workstation is switched. The ratio of the preset response benchmark value to the average steady-state time is calculated to obtain the response speed component. The blockage release component and the response speed component are weighted and summed to obtain the comprehensive scheduling score. Candidate workstations are selected in descending order of comprehensive scheduling score to obtain the first test strategy including the test order.

4. The method according to claim 3, characterized in that, A second test strategy is generated during the testing process, including: The nodes corresponding to the candidate workstations and their associated edges in the conflict dependency graph are deleted to obtain the corrected dependency graph. Based on the corrected dependency graph, the in-degree of the remaining workstations to be scheduled is recalculated. If there are any workstations to be scheduled with an in-degree of zero, the power jump variable of the workstations that have completed the switching, the elapsed time from the switching completion time of the workstations that have completed the switching to the current time, and the current load power of the workstations with an in-degree of zero are substituted into the impact attenuation function to calculate the current residual voltage deviation. If the current residual voltage deviation is less than or equal to the voltage deviation threshold, the workstations with an in-degree of zero are added to the next round of candidate switching workstations for scheduling. Otherwise, the calculation is performed again after a preset time until the residual voltage deviation is less than or equal to the voltage deviation threshold, so as to generate the second test strategy.

5. The method according to claim 4, characterized in that, If there are no workstations to be scheduled with an in-degree of zero in the conflict dependency graph or the modified dependency graph, calculate the robustness score for all workstations to be scheduled in the conflict dependency graph or the modified dependency graph, and obtain an ordered candidate sequence based on the sorting of robustness scores from high to low. A search algorithm is executed based on an ordered candidate sequence to divide the workstations to be scheduled into test batches, and synchronous gear switching instructions are issued to each workstation in the test batch. The predicted impact deviation of each workstation to be scheduled is calculated based on the impact attenuation function. The collected voltage value of each workstation is summed with the predicted impact deviation to compensate for the collected voltage value.

6. The method according to claim 5, characterized in that, Calculate the immunity score for all workstations to be scheduled in the conflict dependency graph or the modified dependency graph, including: Summing the weights of the directed edges from the workstation to be scheduled and from other workstations to itself, we obtain the sum of the weights of the two-way edges. We locate the minimum value of the sum of the weights of the two-way edges of all workstations, as well as the minimum value of the power jump variable of all workstations. Based on the minimum value of the sum of the weights of the two-way edges, the weights of the two-way edges of the workstation to be scheduled, the minimum value of the power jump variable, and the power jump variable of the workstation to be scheduled, we calculate the immunity score.

7. The method according to claim 6, characterized in that, A search algorithm is executed based on an ordered candidate sequence to divide the workstations to be scheduled into test batches, including: Construct empty first set, second set and third set. Fill all the workstations to be scheduled in the ordered candidate sequence into the second set in turn. Take out the workstations to be scheduled from the second set in turn and add the taken workstations to the first set. Calculate the update superimposed impact on each original workstation in the first set after it is added, and the cumulative impact on the workstation to be scheduled. Calculate the compensation limit threshold based on the training set. If both the update superimposed impact and the cumulative impact are less than the compensation limit threshold, the workstation to be scheduled is retained in the first set; otherwise, the workstation to be scheduled is moved to the third set. When the second set is empty, configure the first set as a test batch, delete the workstations to be scheduled contained in the test batch, move the workstations to be scheduled in the third set to the second set, and repeat the above steps until all workstations to be scheduled are sorted.

8. An automated testing system for off-line testing of in-vehicle USB products, used to implement the method as described in any one of claims 1-7, characterized in that, The system includes: The function construction module constructs an impact attenuation function, defines the workstation where the vehicle USB product under test is located as a workstation to be scheduled, and extracts the target gear to be switched, the gear before switching, and the current load power of each workstation when multiple workstations to be scheduled initiate a gear switching request. The deviation calculation module calculates the power jump variable based on the gear position before switching and the target gear position to be switched. It pairs each work position to be scheduled and defines it as the first work position and the second work position respectively. It inputs the power jump variable of the first work position, the average time of gear switching, and the current load power of the second work position into the impact attenuation function to obtain the residual voltage deviation of the first work position on the second work position. The graph construction module, if the residual voltage deviation is greater than the preset voltage deviation threshold, establishes directed edges between workstations based on the residual voltage deviation and uses the residual voltage deviation as the edge weight to construct a conflict dependency graph. The strategy scheduling module determines a first test strategy for the workstations to be scheduled based on a conflict dependency graph, performs gear switching tests on at least one workstation to be scheduled based on the first test strategy, corrects the conflict dependency graph after the workstation to be scheduled completes the gear switching test to obtain a corrected dependency graph, generates a second test strategy for scheduling untested in-vehicle USB products based on the corrected dependency graph, and tests the untested in-vehicle USB products based on the second test strategy.

Images