A method for standardizing test data of a photonic quantum chip

Abstract

The application relates to the technical field of computer data processing, and discloses a photonic quantum chip test data standardization processing method; the method comprises the following steps: firstly, acquiring photonic quantum chip test data generated by different test instruments, establishing a time corresponding relationship and associating data source information; secondly, classifying and preprocessing the test data according to channel purposes and data characteristics, calling a locked reference configuration for mapping, and completing reference working condition conversion and long-term drift correction according to actual test wavelengths, temperatures and test durations; and finally, saving original data, processing results and quality states in association, outputting standardized data after checking, and enabling test results of different chips, test batches and test stations to be compared, reviewed and traced.

Description

Technical Field

[0001] This invention relates to the field of computer data processing technology, specifically to a method for standardizing test data processing of optical quantum chips. Background Technology

[0002] In the research and development, wafer-level testing, post-packaging testing, and system-level verification of quantum chips, it is typically necessary to use testing equipment such as single-photon detectors, time-to-digital converters, voltage drivers, ammeters, power meters, wavelength meters, and temperature sensors to collect data such as single-photon counts, coincidence counts, voltage, current, optical power, coupling loss, insertion loss, test wavelength, and chip temperature. For example, authorized Chinese invention patent application CN116402150B discloses a test calibration method for quantum computers, which calibrates quantum logic gates and joint readout criteria by testing the intrinsic parameters of qubits. Another example is authorized Chinese invention patent application CN113011592B, which discloses a crosstalk analysis method, device, computer equipment, and storage medium for qubits, which determines the crosstalk strength between qubits by performing spectral quantum process tomography separately and simultaneously. The aforementioned technologies primarily involve qubit parameter calibration or qubit crosstalk analysis, but do not establish a coordinated data processing method for the various test data generated by different testing equipment during the testing of optical quantum chips. Photon event data, continuous measurement data, and operating condition and time-base data differ in physical meaning, dimensions, value range, acquisition cycle, and error sources. If the same data scaling method is used for processing, high-count data may occupy a larger mapping range, making it difficult to distinguish data differences within the low-count interval. Abnormal spikes, zero-point drift, and range saturation data in continuous measurement data may also change the processing benchmark, compressing data within the normal operating range. Furthermore, optical... The test results of quantum chips are easily affected by the test wavelength, chip temperature, detector temperature, and long-term operational drift of the test equipment. Data obtained from different chips, different test batches, or different test stations are difficult to compare directly if they are not converted to the same test conditions. If existing test systems only save the final processed results without saving the original data, correction parameters, reference configurations, and acquisition conditions, it will be difficult to verify abnormal results and trace the source of data deviations. Therefore, how to adopt appropriate processing methods for test data of different types of optical quantum chips and form comparable, verifiable, and traceable test results under a unified reference condition has become a technical problem that needs to be solved. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a standardized processing method for optical quantum chip test data, which solves the problems mentioned in the background section.

[0004] To achieve the above objectives, the present invention provides the following technical solution: A method for standardizing test data processing of optical quantum chips includes: Acquire photonic quantum chip test data generated by different testing instruments, establish the time correspondence between the photonic quantum chip test data, and associate the data source information; Based on the instrument channel usage and data characteristics, the test data of the optical quantum chip is divided into photon event data, continuous measurement data, and operating condition and time base data; The photon event data, the continuous measurement data, and the operating condition and time base data are preprocessed respectively. Invoke the reference configuration locked when the test task starts, and map the preprocessed data according to the test parameter type; Based on the actual test wavelength, actual test temperature and test duration, the mapped data is converted to the reference operating conditions and long-term drift correction is performed to obtain standardized data; The system associates and saves the original test data, preprocessing results, mapping results, standardized data, processing parameters, and quality status. It also verifies the associated and saved data. If the verification passes, the standardized data is output. If the verification fails, the affected data is isolated and returned to the corresponding processing stage.

[0005] Furthermore, establishing the time correspondence between the test data of the optical quantum chip includes: Based on hardware trigger flags, common clock count values, or instrument-recorded acquisition times, the data generated by different testing instruments are divided into multiple acquisition segments; The original acquisition time of each data stream is retained, and the operating condition data corresponding to the performance data is determined based on the acquisition time of the performance data. The synchronization deviation between different test channels is compared with the corresponding allowable synchronization deviation. A time correspondence is established when the synchronization deviation does not exceed the allowable synchronization deviation. When the synchronization deviation exceeds the allowable synchronization deviation, the synchronization anomaly and the affected data range are recorded.

[0006] Further, determining the operating condition data corresponding to the performance data includes: When the acquisition time of the performance data is between the acquisition times of two adjacent operating condition data, the interpolation error is determined based on the rate of change and time interval of the two adjacent operating condition data. When the interpolation error does not exceed the allowable error of the corresponding operating condition parameter, and no wavelength switching, temperature adjustment, optical path switching or abnormal recovery occurs between the two adjacent operating condition data, the corresponding operating condition value is determined based on the two adjacent operating condition data. If the time of collection of the performance data is outside the time range of existing operating condition data, or if there is a state switch, data loss, or abnormal change between two adjacent operating condition data, the performance data will be marked as operating condition data missing.

[0007] Furthermore, the step of dividing the test data of the optical quantum chip into photon event data, continuous measurement data, and operating condition and time base data includes: Based on the instrument type, instrument channel, and channel purpose recorded in the test task, the test data of the optical quantum chip are preliminarily classified. The preliminary classification results are verified based on the physical unit, data format, value type, sampling interval, variation, and range status of the data. When the purpose of the instrument channel matches the data characteristics, the classification result is confirmed; when the purpose of the instrument channel does not match the data characteristics, a classification anomaly record is generated and the corresponding data segment is isolated.

[0008] Furthermore, the preprocessing of the photon event data includes: Dark count data is collected when the incident light path is closed and the detection channel is in a dark state. The dark count data is divided into multiple statistical windows, and the dark count baseline is determined according to the dark count rate of each statistical window. Subtract the dark count baseline from the observed count rate; When the observed count rate after deducting the dark count is within the correction range confirmed by the detector calibration, it is corrected according to the dead time correction relationship and the after-pulse calibration relationship corresponding to the detector. When the observation count rate exceeds the correction range, the detector is saturated, or the correction parameter is missing, the generation of the corrected count rate is stopped, and the processing status of the affected data is marked.

[0009] Further, the preprocessing of the continuous measurement data includes: Candidate data segments that did not experience scanning transitions, channel switching, drive state changes, temperature adjustments, instrument resets, or abnormal recovery were selected from the continuous measurement data. A stable data segment is determined based on the fluctuation range, rate of change, and duration of the measured values ​​within the candidate data segment, and the acquisition link baseline is determined based on the stable data segment; The baseline of the acquisition link at the corresponding time moment is subtracted from the continuous measurement data, and filtering is performed based on the response bandwidth of the device under test, the sampling rate of the acquisition device, and the noise spectrum of the acquisition link; Compare the response amplitude, inflection point, and settling time before and after filtering. If the comparison result does not meet the corresponding allowable conditions, mark the filtering result as abnormal. When the continuous measurement data reaches the range boundary or the acquisition device outputs an overload state, the corresponding data segment is marked as range saturation.

[0010] Furthermore, the mapping of preprocessed data according to test parameter type includes: For count data spanning multiple orders of magnitude, a logarithmic mapping is performed based on the lower bound of the count rate, the upper bound of the count rate, and the minimum positive quantity used to handle the uncounted state; For test parameters with non-zero reference values, mapping is performed based on the deviation between the current measured value and the reference value, as well as the corresponding allowable deviation. For test parameters whose target value is not suitable as a divisor, mapping is performed based on the degree of deviation between the current measured value and the target value; For test parameters expressed in decibels, a mapping is performed based on the difference between the current decibel value and the reference decibel value; For continuous measurement data used for comparisons between different chips, different test batches, or different test stations, a bounded mapping is performed based on the reference center and reference scale.

[0011] Furthermore, the reference center and the reference scale are determined based on reference data obtained under the same chip type, the same test parameters, the same instrument configuration, and the same test conditions; The reference data used to determine the reference center and the reference scale does not include range saturation data, operating condition over-limit data, time synchronization failure data, equipment abnormality data, and interpolation data; The reference center and the reference scale are written into the reference configuration before the test task starts and remain unchanged during this data processing.

[0012] Furthermore, the process of converting the mapped data to the reference operating condition includes: The response relationship of the test parameters with wavelength variation is determined based on multiple wavelength measurement points, and the response relationship that satisfies the residual allowable condition is selected based on the residual between the measured value of each wavelength and the corresponding result of the response relationship. Based on the physical meaning of the test parameters and the wavelength scanning calibration results, additive or proportional correction is used to convert the mapped data to the reference wavelength. The response relationship of the test parameters to temperature changes is determined based on multiple stable temperature points, and additive or proportional correction is used to convert the wavelength-corrected data to the reference temperature. When the measurement difference between the heating and cooling processes does not meet the allowable conditions for temperature measurement, heating response relationship and cooling response relationship are established respectively, and the corresponding response relationship is selected according to the current direction of temperature change.

[0013] Furthermore, the long-term drift correction includes: The current baseline is determined from continuous test data that has never experienced wavelength switching, temperature adjustment, drive state switching, instrument reset, or abnormal recovery. The reference baseline is determined from the stable data range after the start of the test task, the calibrated reference channel data, or the reference configuration. When the long-term drift exhibits an additive change, correction is made based on the difference between the current baseline and the reference baseline. When long-term drift manifests as a proportional change, and the current baseline and the reference baseline meet the corresponding applicable conditions, correction is performed based on the ratio between the current baseline and the reference baseline; Among them, range saturation data, synchronization failure data, unstable operating condition data, interpolation data, data during abnormal recovery, re-acquired transitional data, and data to be confirmed are not used to determine or update the current baseline.

[0014] Furthermore, the associated data storage includes original test data, preprocessing results, mapping results, standardized data, processing parameters, and quality status, and the associated data storage is verified, including: By using chip number, test parameter number, acquisition segment number, and data point number, the raw test data, preprocessing results, mapping results, standardized data, and processing information are linked together. When reprocessing the same original data segment using different reference configurations or processing parameters, a new version of the result is generated, the original processing result is retained, and each version of the result is associated with the corresponding original test data. Perform integrity checks, configuration version checks, processing record checks, abnormal status checks, and recalculation verification on the standardized data to be output; The recalculation and verification process starts with the original test data, re-executes the data processing according to the original processing program version, reference configuration version, processing parameters, and storage precision, and compares the recalculation results with the standardized data to be output.

[0015] Furthermore, the process of isolating the affected data and returning it to the corresponding processing stage includes: Based on the scope of their impact, anomalies are classified into channel-level anomalies, test bench-level anomalies, or batch-level anomalies. When a channel-level anomaly occurs, the affected channel is isolated, and the decision on whether to continue collecting data from unaffected channels is made based on the dependencies between test parameters. When a test bench-level anomaly occurs, the current test task is paused. After the time synchronization status, wavelength status, temperature status, instrument range, and channel configuration are restored, the data is reacquired from the acquisition segment where the data acquisition is complete and the quality status meets the test requirements. When multiple chips or multiple channels within the same test batch exhibit the same type of anomaly and meet the batch review criteria, the output of new results for that test batch is paused and batch review is performed. For continuous data that does not meet the completion criteria and for missing photon event data, the affected area should be re-acquired or the corresponding data should be isolated.

[0016] Compared with existing technologies, this invention provides a method for standardizing test data processing of optical quantum chips, which has the following beneficial effects: 1. This invention establishes a time correspondence between the output data of different testing instruments and divides the test data into photon event data, continuous measurement data, and operating condition and time base data according to the instrument channel usage and data characteristics. Dark count subtraction, dead time and after-pulse correction, baseline correction, abnormal data processing, and operating condition status determination are performed on each of these data. Then, the corresponding reference configuration is called according to the test parameter type for mapping, and the mapping results are converted to reference wavelength, reference temperature, and reference time status. Simultaneously, the original data, processing parameters, configuration version, and quality status are associated and saved. Abnormal data is isolated, reprocessed, or reacquired, thereby reducing the impact of high count data, abnormal spikes, baseline drift, and range saturation on the data processing results. This enables the comparison, verification, and traceability of test results from different chips, different test batches, and different test stations.

[0017] 2. This invention performs integrity checks, configuration version consistency checks, processing record checks, and recalculation verification on the output data. Based on the scope of the anomaly's impact, it performs channel isolation, task pause, re-acquisition, or batch verification. This ensures that data corresponding to configuration anomalies, range saturation, time synchronization failures, abnormal operating conditions, and missing data are not included in the reference configuration and baseline updates, nor are they written to the valid results area. Simultaneously, it retains the abnormal data, the original processing results, and their version information, thereby preventing local anomalies from affecting subsequent test data and other test results in the same batch. Furthermore, it allows for verification and traceability of the anomaly's location, processing procedure, recovery starting point, and the correspondence between different result versions. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the structure of the optical quantum chip test data standardization processing system of the present invention; Figure 2 This is a flowchart of the method for standardizing test data processing of optical quantum chips according to the present invention; Figure 3 This is a schematic diagram illustrating the data acquisition and time alignment of multiple testing instruments according to the present invention; Figure 4 This is a flowchart of the preprocessing process for the three types of test data in this invention; Figure 5 This is a schematic diagram of the reference mapping for different test parameters of the present invention; Figure 6 This is a flowchart of the reference operating condition conversion and long-term drift correction process for this invention. Figure 7 This is a schematic diagram illustrating the association and storage of the original data and processing results in this invention; Figure 8 This is a standardized data quality verification flowchart for this invention; Figure 9 This is a flowchart of the test anomaly classification and recovery process of the present invention; Figure 10 This is a schematic diagram of the main control device for testing according to the present invention. Detailed Implementation

[0019] The technical solution of this application will be further described in detail below with reference to the accompanying drawings. The following embodiments are only used to illustrate the technical solution of this application and are not intended to limit the scope of protection of this application. While ensuring that the necessary sequential relationship and data correspondence between each processing step remain unchanged, and that the configuration version used in the same processing process is consistent, those skilled in the art can make adaptive adjustments to the steps that can be executed in parallel, the deployment location of each functional module, and the data storage method according to the structure of the chip under test, the interface form of the test equipment, the type of test parameters, and the on-site deployment conditions.

[0020] Please see Figure 1 - Figure 10 This invention provides a method for standardizing test data processing of optical quantum chips.

[0021] In the research and development, wafer-level testing, post-packaging testing, and system-level verification of quantum optical chips, it is typically necessary to use various testing instruments to collect data such as single-photon counts, coincidence counts, event timestamps, interference visibility, phase drive voltage, monitored photocurrent, optical power, coupling loss, insertion loss, center wavelength, chip temperature, and time synchronization status. These data can be generated by single-photon detectors, time-to-digital converters, coincidence counting units, voltage drive sources, ammeters, power meters, wavelength meters, temperature sensors, and time synchronization components, respectively.

[0022] Different test data vary in terms of physical source, units, value range, acquisition cycle, and sources of error. When test results need to be compared between different chips, different batches, or different test stations, applying the same linear scaling method to all types of data may compress or amplify the original differences, affecting the accurate comparison of test results.

[0023] Under low-light testing conditions, single-photon counts can span multiple orders of magnitude, and count variations close to the dark count level can still reflect changes in chip output, optical path state, or detector state. If linear scaling is applied based on the maximum and minimum values ​​of all count data, the high-count range will occupy a large mapping range, making it difficult to distinguish changes in the low-count range.

[0024] For continuous measurement data such as voltage, current, and optical power, abnormal spikes, zero-point drift, and range saturation may alter the reference values ​​and mapping ranges used in processing, compressing data within the normal operating range. If each chip or each test batch redetermines the processing reference based on its own data, it may also eliminate the overall deviations that actually exist between chips, which is not conducive to comparing test results from different chips or different batches.

[0025] The transmission, interference, modulation, and detection results of quantum optical chips are also affected by the test wavelength, chip temperature, detector temperature, and long-term drift of the test stage. Measurement results obtained from the same chip at different wavelengths, temperatures, or test periods should not be directly compared without being converted to the same reference wavelength, temperature, and time.

[0026] To ensure the verifiability and traceability of test results, the original data record must include at least the following fields: measured value, physical unit, acquisition time, instrument identifier, channel identifier, chip identifier, and instrument status. The processing record must include at least the following fields: dark count baseline, dead time parameter, continuous measurement baseline, reference operating condition, judgment conditions, and processing version. Dead time refers to the time interval after the detector completes a photon event response, during which it cannot normally record subsequent photon events. If the test system only saves the final processing result and not the aforementioned original data and processing records, it will be difficult to determine whether the deviation originates from the chip under test, the test instrument, the test environment, or the data processing configuration when anomalies occur in the test results.

[0027] Based on the above, this implementation method divides the test data into corresponding processing channels according to the physical source and data format of the test data; for photon event data, it processes the effects of dark count, dead time, afterpulse, and probe saturation; for continuous measurement data, it processes the effects of zero-point offset, baseline drift, abnormal spikes, and range saturation; for operating condition and time base data, it determines wavelength deviation, temperature deviation, test duration, and synchronization status; after completing the above processing, according to the physical meaning of each test parameter, it calls the corresponding reference value, allowable deviation, effective range, or reference curve for mapping, and converts the test results to the preset reference operating condition according to the actual test wavelength, test temperature, and test duration.

[0028] During the test data processing, the raw data record includes at least the fields of measured value, physical unit, acquisition time, instrument identifier, channel identifier, and chip identifier; the processing record includes at least the fields of preprocessing result, reference mapping result, operating condition calibration result, acquisition conditions, processing parameters, and configuration version; the above data establishes a correspondence through chip identifier, test parameter identifier, and acquisition time, so that the standardized result can be queried to the corresponding raw measurement data and processing basis, and can be recalculated according to the same processing parameters and configuration version.

[0029] like Figure 1 As shown, the photonic quantum chip test data standardization processing system 100 can be applied to R&D laboratory test benches, wafer-level test lines, chip test stations, or system-level verification platforms; the system 100 can be implemented by an industrial control computer, a test main control server, an edge computing device, or an embedded test controller.

[0030] System 100 can interact with tunable laser sources, single-photon detectors, time-to-digital converters, voltage drive sources, power meters, ammeters, wavelength meters, temperature control stations, and time synchronization devices to receive measurement data and instrument status. When an acquisition anomaly or an instrument status anomaly occurs, System 100 can also send control commands to the corresponding instrument, such as pausing acquisition, resuming acquisition, re-acquiring, adjusting the range, adjusting the integration time, switching to a backup channel, or reloading the configuration.

[0031] System 100 may include a data acquisition and time alignment module 110, a data classification module 120, a data preprocessing module 130, a reference mapping module 140, a working condition correction module 150, a data storage module 160, a quality verification module 170, an anomaly handling module 180, and a reference configuration library 190.

[0032] The data acquisition and time alignment module 110 is used to receive measurement data, instrument status, event timestamps and trigger information output by each test instrument, and to divide the acquisition segments according to the test task to establish the time correspondence between data from different instruments.

[0033] The data classification module 120 is used to associate data source information with data points or data segments, and to divide the test data into corresponding processing channels according to the instrument channel purpose, data unit and data format.

[0034] The data preprocessing module 130 is used to perform dark count subtraction, dead time correction, after-pulse correction and detection saturation check on photon event data, perform baseline correction, abnormal peak processing and range check on continuous measurement data, and determine the test wavelength, test temperature, test duration and synchronization status.

[0035] The reference mapping module 140 is used to read reference values, target values, permissible deviations, effective ranges or reference curves that match the current chip, test parameters and test equipment from the reference configuration library 190, and convert the preprocessed data into relative values ​​that are easy to compare according to the physical meaning of the test parameters.

[0036] The operating condition correction module 150 is used to perform reference operating condition conversion and long-term drift correction on the relative values ​​based on the actual test wavelength, actual test temperature and test duration, so as to obtain standardized data under the reference operating condition.

[0037] The data storage module 160 is used to associate and store raw data, preprocessed data, relative values, standardized data, processing parameters, configuration versions, and quality status.

[0038] The quality verification module 170 is used to check whether the data records are complete, whether the configuration versions are consistent, whether the processing process is continuous, and whether the processing results can be recalculated. Based on the verification results, it controls whether standardized data is written to the valid result area.

[0039] The anomaly handling module 180 is used to receive anomaly information generated by each module and perform channel isolation, task suspension, reprocessing, re-acquisition, instrument reset or batch verification according to the anomaly type and scope of impact.

[0040] Reference configuration library 190 is used to store reference values, target values, permissible deviations, reference curves, judgment thresholds, applicable scopes and configuration versions corresponding to different chip types, test parameters and test equipment; the reference configuration can be derived from equipment calibration records, device design documents, product test specifications, equipment technical documents, repeatability test results or confirmed historical valid test data.

[0041] Each threshold record must include at least the following fields: threshold name, physical unit, upper or lower limit, determination basis, applicable chip type, applicable test equipment, applicable channel, applicable operating conditions, configuration version, and effective time. Unconfirmed abnormal data, interpolated data, and temporary correction data are not used to generate the formal reference configuration.

[0042] When a test task starts, System 100 calls a reference configuration that matches the current chip, test equipment, and test parameters, and keeps the configuration version unchanged during this data processing. When it is necessary to adjust the reference value, judgment threshold, or reference curve, a new configuration version is generated, and the affected data is reprocessed according to the new configuration. Parameters that lack a definite basis, applicable scope, or version information are not used to generate valid test results.

[0043] Photon event data refers to discrete event data generated by single-photon detectors, time-to-digital converters, or coincidence counting units, as well as data such as count rate, coincidence rate, second-order coherence function, and interferometric visibility obtained from discrete event statistics. This type of data is mainly affected by dark count, dead time, afterpulse, and detector saturation. The correction parameters used should correspond to the detector model, operating bias, detector temperature, and acquisition configuration.

[0044] Continuous measurement data refers to data such as voltage, current, optical power, coupling loss, insertion loss, and demodulation results output in continuous numerical form. Even if this type of data is represented by discrete sampling points after analog-to-digital conversion, its physical quantities are still continuously changing quantities, mainly affected by zero-point offset, baseline drift, acquisition jitter, abnormal spikes, and range saturation.

[0045] Operating condition and time base data refer to data used to describe test conditions and the time relationship between different instruments, including at least test wavelength, chip temperature, detector temperature, test duration, synchronization deviation, and synchronization status.

[0046] Test duration represents the time interval between the current acquisition time and the start time of the test segment or the preset reference time, used to determine the drift caused by the long-term operation of the test equipment; synchronization deviation represents the time difference between the time bases of different instruments, used to determine whether the data from different channels can establish an effective time correspondence; test duration and synchronization deviation are used for drift correction and time alignment, respectively, and the two are not interchangeable.

[0047] Data source information should include at least the instrument identifier, channel identifier, acquisition time, configuration version, and synchronization status, and may also include fields such as chip number, wafer coordinates, batch number, task number, sampling rate, integration time, range, optical path configuration, and temperature control status.

[0048] Reference benchmarks are reference information set separately for different test parameters. They can take the form of reference values, target values, nominal values, permissible deviations, reference ranges, or reference curves that vary with wavelength and temperature. Different test parameters call upon the corresponding reference benchmarks according to their physical meanings, and all data are not processed using the same fixed value.

[0049] For test parameters with a target value of zero or close to zero, the target value is not used as a divisor; for test parameters expressed in decibels, the difference between the current decibel value and the reference decibel value is used, or the relative result is determined based on the allowable decibel deviation.

[0050] The standardized data described in this embodiment refers to the data obtained after the original test data has undergone data preprocessing, reference mapping, wavelength correction, temperature correction, and long-term drift correction. The standardized data is used for comparison between different chips, different test segments, or different test equipment. It does not replace the original measurement data, nor does it represent an absolute measurement result that is detached from specific test conditions.

[0051] like Figure 2 As shown, the standardized processing method for test data of optical quantum chips includes S101 to S107.

[0052] S101, acquire photonic quantum chip test data generated by different test instruments, establish the time correspondence between the data of each test instrument, and associate the test data with data source information.

[0053] S102, based on the purpose and data characteristics of the instrument channel, the test data is divided into photon event data, continuous measurement data, and operating condition and time base data.

[0054] S103, based on the generation method and main error sources of various test data, preprocess the photon event data, continuous measurement data, and operating condition and time base data respectively.

[0055] S104, based on the type of test parameters, calls the corresponding reference value, target value, allowable deviation, effective range, or reference curve to perform mapping processing on the preprocessed test data.

[0056] S105 converts the mapped data to a preset reference condition based on the actual test wavelength, actual test temperature, and test duration, and corrects for the drift caused by long-term operation of the test equipment to obtain standardized data.

[0057] S106, which associates and saves the original test data, preprocessing results, mapping results, standardized data, data source information, processing parameters, and quality status.

[0058] S107 performs integrity and consistency checks on the associated and stored data; if the check passes, it outputs valid standardized data; if the check fails, it isolates the affected data according to the cause of the anomaly and returns it to the corresponding processing stage for reprocessing or re-collection.

[0059] In S101, system 100 receives test tasks; the test tasks include at least the chip under test identifier, test parameters, instrument-channel correspondence, and reference configuration version.

[0060] For wafer-level testing, test tasks may also include fields such as wafer number, die coordinates, and probe card channels; for post-package testing, test tasks may also include fields such as package number, fiber array number, and coupling port; for system-level verification, test tasks may also include fields such as system connection relationships and the location of the device under test.

[0061] The acquisition and time alignment module 110 sends acquisition configuration to each test instrument according to the test task; the acquisition configuration includes at least parameters such as sampling rate, integration time, range, triggering mode, scanning range, scanning interval, temperature setpoint and temperature stability conditions.

[0062] The integration time is determined based on the expected count rate, the dark count rate, the allowable statistical error, and the test time requirements. For test channels with low count rates, the integration time can be extended or the number of repeated collections can be increased to increase the number of valid events, rather than shortening the collection time by relaxing the data validity requirements.

[0063] The sampling rate is determined based on the effective response bandwidth of the device under test, the bandwidth of the acquisition equipment, and the anti-aliasing filtering conditions, so that the acquired data can reflect the signal changes that need to be retained; the measurement range is determined based on the allowable input range of the equipment, the expected signal amplitude, and the overload protection requirements, so that the normal test signal is within the effective measurement range of the equipment.

[0064] For single-photon data and coincidence count data that require establishing event-level time correspondence, time alignment can be performed using a common reference clock, hardware trigger signal, or an internal time base shared by the test instrument.

[0065] For wavelength, temperature, and optical power monitoring data that change slowly, a time relationship can be established with the test data at the corresponding time based on the acquisition timestamp and acquisition period. The data network can be used for configuration distribution and measurement data transmission, but the time it takes for data packets to arrive at the system is affected by network transmission delay and is not used as the hardware trigger time or instrument acquisition time.

[0066] like Figure 3 As shown, the acquisition and time alignment module 110 divides the data generated by different test instruments into multiple acquisition segments according to the hardware trigger flag, the common clock count value, or the acquisition time recorded by the instrument; each acquisition segment corresponds to a continuous acquisition data segment of the same chip under test under the same test conditions, such as a stable operating condition, a wavelength scanning range, or a temperature stable range.

[0067] For data with different sampling rates, the acquisition and time alignment module 110 retains the original acquisition time of each data stream and uses the acquisition time of the performance data as a reference to find the corresponding wavelength, temperature and other operating condition data in time, without changing the sampling order and sampling time of the original data.

[0068] The allowable synchronization deviation is determined based on the accuracy requirements of the test parameters for time correspondence and the calibration results of the time synchronization device; the allowable synchronization deviation is not greater than the maximum allowable time error of the current test parameter and is within the effective range confirmed by the calibration of the time synchronization device; when the same test task includes multiple test parameters, the allowable synchronization deviation can be set separately; when a unified allowable synchronization deviation is used, the most stringent allowable value among the test parameters is taken.

[0069] When the actual synchronization deviation between two test channels is not greater than the corresponding allowable synchronization deviation, the acquisition and time alignment module 110 establishes the time correspondence between the corresponding data; when the actual synchronization deviation exceeds the allowable synchronization deviation, the relevant data is not used as valid test data at the same moment, and the synchronization anomaly and the range of affected data are recorded.

[0070] For continuously acquired operating condition data such as wavelength, temperature and optical power, the acquisition and time alignment module 110 determines the acquisition time, value difference and rate of change of two adjacent operating condition data. When the acquisition time of performance data is between the acquisition times of two adjacent operating condition data, the corresponding operating condition value can be determined by linear interpolation.

[0071] Before using linear interpolation, the possible error of the operating condition value is estimated based on the rate of change and time interval of adjacent operating condition data. When the estimation error is not greater than the allowable error of the operating condition parameter, and no wavelength switching, temperature setpoint adjustment, optical path switching or abnormal recovery occurs between adjacent acquisition times, the interpolated operating condition value can establish a correspondence with the performance data.

[0072] The permissible error of the operating condition parameters is determined based on the device test specifications, the sensitivity of the measured performance parameters to changes in operating conditions, and the repeatability of the operating condition measurement equipment. For example, when the performance parameters are more sensitive to temperature changes, a smaller permissible error is used for the temperature data; when the performance parameters change less within the corresponding temperature range, a larger permissible error can be used according to the test specifications. The operating condition data obtained by interpolation should record information such as the original adjacent data, the interpolation method, and the corresponding time.

[0073] When the time of performance data collection is outside the time range of existing operating condition data, or when there is a state switch, data missing or abnormal change between adjacent operating condition data, the operating condition value is not supplemented for the performance data. Instead, it is marked as operating condition data missing, and re-collection or isolation processing is performed according to the missing range.

[0074] The acquisition and time alignment module 110 associates data source information with each data point or data segment; the data source information includes at least fields such as instrument identifier, channel identifier, chip identifier, acquisition time, acquisition segment number, synchronization status and configuration version, and is sent to the data classification module 120 along with the original data.

[0075] The original measurement data remains unchanged in subsequent processing; when the same test parameter is re-acquired, the system generates a new acquisition segment number for the re-acquired data and saves the original acquisition data and the re-acquired data separately to avoid the re-acquired results from overwriting the original records.

[0076] When events occur such as timestamp reversal, inconsistency between the number of hardware triggers and the number of acquisition segments, loss of common clock lock, synchronization deviation exceeding the allowable range, or raw data failing to be associated with a valid test channel, the acquisition and time alignment module 110 generates a time alignment anomaly record. The anomaly record includes at least the anomaly type, instrument identifier, channel identifier, affected acquisition segment, time of anomaly occurrence, and the most recent complete valid acquisition segment, and is sent to the anomaly handling module 180.

[0077] In step S102, the data classification module 120 first performs preliminary classification of the test data according to the instrument type, instrument channel and channel purpose recorded in the test task.

[0078] The data output from the single-photon detector, time-to-digital converter, and coincidence counting unit are classified as photon event data; the data output from the voltage drive source, ammeter, power meter, and lock-in amplifier are classified as continuous measurement data; and the data output from the wavelength meter, temperature sensor, and time synchronization device are classified as operating condition and time base data.

[0079] To prevent data classification errors caused by channel reuse, configuration errors, or instrument replacement, the data classification module 120 also verifies the preliminary classification results based on the physical unit, data format, value type, sampling interval, variation, and range status of the data.

[0080] Photon event data is typically a non-negative integer count or a data sequence with event timestamps; for example, single photon counts or coincidence counts recorded within a fixed integration time, whose physical unit can be count or count rate, and the data value is usually not less than zero; the data classification module 120 can determine whether the data belongs to photon event data based on the data source, physical unit and value form.

[0081] Continuous measurement data are usually expressed as continuous values ​​such as voltage, current, optical power or loss, and change with the measured state within the effective range of the test instrument; the data classification module 120 also checks whether the data continuously reaches the upper or lower limit of the range, remains at a fixed value for a long time, or is overloaded or interrupted in the acquisition link; the above states are used to determine whether the data is valid, without changing the category of its continuous measurement data.

[0082] Operating conditions and time base data are used to represent test conditions such as test wavelength, chip temperature, detector temperature, acquisition time, synchronization deviation, and synchronization status. This type of data usually has a specific wavelength, temperature, or time unit, and its range and rate of change are limited by the working status of the laser source, temperature control equipment, and time synchronization equipment.

[0083] When the purpose of the instrument channel is consistent with the physical unit, value format and sampling characteristics of the data, the data classification module 120 confirms the classification result and sends the data to the corresponding data preprocessing channel.

[0084] When the purpose of the instrument channel is inconsistent with the data characteristics, the data classification module 120 does not directly adopt the preliminary classification result, but generates a classification anomaly record. The classification anomaly record includes at least the fields of instrument identifier, channel identifier, data unit, data format, acquisition segment number, preliminary classification result and anomaly reason, and is sent to the anomaly handling module 180 for processing.

[0085] For test lines with fixed equipment composition and channel usage, they can be mainly classified according to the pre-configured instrument channel relationships in the test task, and the channel configuration can be checked for abnormalities by using data units, data formats, and range status.

[0086] For R&D test benches that can collect different physical quantities in different test tasks using the same acquisition channel, the data classification module 120 jointly determines the data category based on the current test task, instrument configuration, physical unit, sampling rate, and data changes, avoiding classification based solely on channel number.

[0087] like Figure 4 As shown, in step S103, the data preprocessing module 130 preprocesses the photon event data, continuous measurement data, and operating condition and time base data respectively; only data with valid classification results can enter the corresponding data processing channel.

[0088] For photon event data, the data preprocessing module 130 sequentially performs dark count subtraction, dead time correction, afterpulse correction, and probe saturation check.

[0089] Dark counts are the counts generated by a single-photon detector in the absence of effective incident light. The dark count baseline should be obtained under the same detector operating bias, detector temperature, integration time, and triggering method as the formal test, in order to avoid the dark count baseline being inapplicable due to different detector operating states.

[0090] Before the formal test, the incident light path can be closed or blocked, and the detection channel can be confirmed to be in dark state by the optical power monitoring equipment. The upper limit of dark state is determined according to the noise floor, minimum resolution and dark state calibration results of the optical power monitoring equipment. When the monitored optical power is not higher than the upper limit of dark state, and the optical path control status indicates that the incident light path has been closed, the current detection channel is determined to be in dark state.

[0091] The dark counting acquisition process can be divided into multiple statistical windows of equal duration, and the dark count rate within each statistical window can be calculated separately. The length of the statistical window is determined based on the detector dark count rate, integration time, and the allowable statistical error of the dark count baseline.

[0092] When the dark count rate is not close to zero, the ratio of the absolute value of the difference between the dark count rates of adjacent statistical windows to the dark count reference value can be used as the dark count change rate; when a preset number of statistical windows all satisfy the dark count change rate not being greater than the dark count stability threshold, the dark count baseline is determined to have reached a stable state.

[0093] When the dark count reference value is close to zero and is not suitable for ratio calculation, the absolute value of the difference between the dark count rates of adjacent statistical windows can be used as a stability criterion, and this difference can be compared with the absolute change threshold of the dark count.

[0094] The statistical window length, the number of consecutive statistical windows, the dark count stability threshold, and the dark count absolute change threshold are determined based on the repeated dark state test results of the same detector under the same operating bias, the same temperature, and the same acquisition configuration, and are recorded in the calibration configuration of the corresponding detection channel. When the dark count rate is low or the number of events in a single statistical window is insufficient, the statistical window can be extended or the number of statistical windows can be increased without shortening the dark count acquisition time by relaxing the stability conditions.

[0095] The data preprocessing module 130 can use the average or median value of the dark count rate within multiple valid statistical windows as the dark count baseline. When there are communication interruptions, detector resets or other abnormal values ​​in the dark count data, the abnormal data is excluded first, and then the dark count baseline is determined. The dark count baseline record includes at least the fields of detector identifier, channel identifier, acquisition time, working offset, detector temperature, integration time and configuration version.

[0096] For test tasks that last for a long time, a dark count acquisition interval can be set during the period when the optical path is closed; when the optical path is in a dark state, the detector's working state has not changed and the dark count meets the stability condition, the data of this acquisition interval can be used to update the dark count baseline.

[0097] The testing system can also be configured to have a reference dark channel that does not receive valid light signals. The reference dark channel and the measurement channel use the same type of detector and have corresponding operating bias, operating temperature and acquisition configuration. When the reference dark channel meets the calibration requirements, the dark count change for the corresponding time period can be determined based on its output.

[0098] The data preprocessing module 130 subtracts the corresponding dark count baseline from the observed count rate to obtain the count rate after dark count deduction. When the dark count deduction result is less than zero, the result can be set to zero and the state below the dark count baseline can be recorded. Alternatively, the corresponding integration interval can be marked as invalid data according to the test task requirements.

[0099] Dead time is the time during which a detector cannot respond normally to subsequent photon events after responding to a photon event. The dead time parameters, dead time correction relationships, and effective correction range are determined based on the detector calibration records or equipment technical documents, and correspond to the detector model, operating bias, and detector temperature.

[0100] For a detection channel that has been calibrated and confirmed to use a non-extended dead-time model, when the observed count rate after dark count subtraction is within the effective correction range, the dead-time correction count rate can be determined according to the following formula: in, The observation count rate is represented by the dark count subtraction, and τ represents the detector dead time. This represents the dead-time correction count rate. When using the above calculation relationship, the product of the observation count rate and the dead time should be less than one, and the observation count rate must not exceed the effective correction upper limit confirmed by the detector calibration.

[0101] The effective correction upper limit is determined based on the test results of the detector's correction error at different count rates; when the dead time correction error exceeds the allowable error of the corresponding test parameter, the corresponding count rate is no longer considered as data within the effective correction range.

[0102] When the observation count rate exceeds the effective correction limit, the detector output is saturated, or the applicable conditions of the correction relationship are not met, the data preprocessing module 130 will no longer calculate the corrected count rate and will generate a dead time correction over-limit record; the record includes at least the fields of detector identifier, channel identifier, observation count rate, dead time parameter, effective correction limit and acquisition segment number.

[0103] When the detector uses other dead-time models or the equipment technical documents provide dedicated correction curves, corrections can be made according to the corresponding calibration relationships; the correction relationships, parameter versions, and applicable scopes used should be saved along with the processing results.

[0104] The afterpulse is the additional count generated by the detector after a valid response; the afterpulse parameters can be derived from the detector calibration curve, repeated test results or equipment technical documents, and correspond to the detector's operating bias, operating temperature and count rate range.

[0105] When the afterpulse parameter represents the proportion of the afterpulse count to the total count before correction, the dead time correction count rate can be multiplied by one and the afterpulse proportion subtracted to obtain the count rate after afterpulse correction.

[0106] When the afterpulse parameter represents the probability of a main event generating an afterpulse, or is represented by an afterpulse response curve that varies with time, it can be corrected by looking up a table according to the calibration relationship of the corresponding detector; the processing record should include at least the definition of the afterpulse parameter, parameter value, calibration conditions, correction method and configuration version, etc.

[0107] When the dark count baseline, dead time parameter, or afterpulse parameter is missing, parameters from other detector models should not be used to replace them; the affected data should be marked as data to be confirmed, and preprocessing should be performed after the corresponding calibration parameters are loaded or recalibration is completed.

[0108] For continuous measurement data, the data preprocessing module 130 first determines the stable data segment that can reflect the baseline of the acquisition link; the stable data segment does not include data during wavelength scanning transitions, channel switching, sudden changes in drive voltage, temperature setpoint adjustments, instrument resets, and abnormal recovery periods.

[0109] The data preprocessing module 130 can calculate the difference between the maximum and minimum values ​​of the measured values ​​within the candidate data segment, and calculate the rate of change of the measured values ​​over time. When the difference between the maximum and minimum values ​​is not greater than the baseline fluctuation threshold, the absolute value of the rate of change is not greater than the baseline drift rate threshold, and the above state continues to reach the baseline stability duration, the candidate data segment is determined as a stable data segment.

[0110] The baseline fluctuation threshold is determined based on the peak-to-peak noise and device repeatability of the acquisition link when the input remains constant; the baseline drift rate threshold is determined based on the long-term stability test results of the test bench; the baseline stabilization time is determined based on the response time of the device under test and the time required for the acquisition device to reach a stable output.

[0111] Voltage, current, optical power, and demodulation results are judged using their respective baseline conditions, and the same fixed threshold is not used; each threshold should be associated with and saved with the test parameters, acquisition equipment, range, sampling rate, and configuration version.

[0112] For stable data segments with a few abnormal spikes, the sliding median can be used to determine the baseline; for data with gradual changes and few abnormal spikes, the sliding average can be used to determine the baseline; the sliding window length is determined based on the effective response time of the device under test, the sampling rate, and the baseline change rate, and ensures that the window length is greater than the duration of the short-term fluctuations that need to be suppressed, while being less than the effective response duration that needs to be retained.

[0113] When the baseline changes continuously and slowly over time, a first-order relationship, a second-order relationship, or a piecewise continuous relationship can be established based on stable data segments, and the baseline value at each acquisition time can be determined according to the corresponding relationship. The order of the relationship used is determined based on the fitting residual and the stability of the acquisition link, and extrapolation results that exceed the actual measurement range are not used.

[0114] When the test bench is equipped with a reference channel that does not contain a valid response from the chip under test, the drift of the acquisition link shared by multiple measurement channels can be determined based on the reference channel. The range, sampling rate and acquisition link of the reference channel and the measurement channel should be matched, and the reference channel itself should not have range saturation or acquisition abnormality.

[0115] The data preprocessing module 130 subtracts the baseline at the corresponding time from the continuous measurement data to obtain the baseline correction data; then, it determines the filtering conditions based on the effective response bandwidth of the device under test, the sampling rate of the acquisition device, and the noise spectrum of the acquisition link.

[0116] The filter passband should cover the frequency band of the device response that needs to be retained and suppress the frequency band where the main noise is located; the filtering conditions can be determined based on the device step response test, the acquisition link frequency response test, the equipment technical documents, or the results of on-site calibration.

[0117] For high-frequency random jitter, low-pass filtering can be used; for a small number of isolated spikes, median smoothing or abnormal data isolation can be used; after filtering, the scanning inflection point, the phase zero-crossing point, and the effective fast response of the device should be preserved.

[0118] The data preprocessing module 130 compares the effective response amplitude, scan inflection point, and settling time before and after filtering; the allowable variation in effective response amplitude is determined based on the repeatability of the test instrument and the product test specifications; the allowable variation in scan inflection point is determined based on the scan interval and parameter extraction resolution; and the allowable variation in settling time is determined based on the device step response test results.

[0119] When any change exceeds the corresponding allowable value, the current filtering result is not used to generate a valid test result, and a filtering anomaly record is generated; the filtering anomaly record includes at least the following fields: channel identifier, filtering method, filtering parameters, response amplitude before and after processing, inflection point, stabilization time, and anomaly cause.

[0120] When continuous measurement data reaches the upper or lower limit of the device's range, multiple consecutive sampling points maintain the same extreme output value, or the acquisition device reports an overload state, the corresponding data segment is marked as range saturation; range saturation data is not used to determine or update the baseline, nor is it used as normal data in subsequent reference mapping.

[0121] For operating conditions and time base data, the data preprocessing module 130 reads the reference wavelength, reference temperature and test segment start time corresponding to the current test task; the reference wavelength and reference temperature are determined according to the device design operating point, product test specifications, equipment calibration conditions or test task configuration, and are associated with chip type, test parameters and configuration version.

[0122] The data preprocessing module 130 calculates the signed difference between the current test wavelength and the reference wavelength, the signed difference between the current test temperature and the reference temperature, and the time interval between the current acquisition time and the start time of the test segment.

[0123] The wavelength difference and temperature difference are used for subsequent reference condition conversion. The time interval is used to determine the drift caused by long-term operation of the test equipment. The synchronization deviation is recorded separately as a time alignment quality parameter and is not combined with the test duration.

[0124] When the wavelength meter is unlocked, the temperature control station does not meet the temperature stability conditions, the temperature sensor output is abnormal, or the time synchronization status fails, the data preprocessing module 130 will not use the corresponding operating condition data for the formal test results and will generate an operating condition anomaly record. The operating condition anomaly record includes at least the instrument identifier, anomaly type, anomaly occurrence time, affected acquisition segment, actual operating condition value, reference operating condition value, and configuration version, and will be sent to the anomaly processing module 180.

[0125] like Figure 5 As shown, in step S104, the reference mapping module 140 calls the reference configuration determined when the test task starts, and maps the preprocessed test data according to the physical meaning, data unit and value characteristics of the test parameters; the reference configuration keeps the version consistent in the same processing process.

[0126] For data such as count rate and coincidence count rate, which are non-negative and may span multiple orders of magnitude, a logarithmic mapping method can be used to reduce the order of magnitude difference between the high and low count intervals; the mapping result of the count data can be determined according to the following formula: Where r represents the count rate after dark count subtraction, dead time correction, and after-pulse correction. Indicates the lower limit of the effective count rate. This represents the upper limit of the effective count rate, and ε represents the minimum positive number used to handle zero count rate. This represents the mapping result of the count data; clip means that when the calculation result is less than 0, the value is 0, when the calculation result is greater than 1, the value is 1, and the original calculation result is retained in other cases.

[0127] The lower limit of the effective count rate is determined based on the detector's dark count level, integration time, and the minimum count rate that can be reliably distinguished; the upper limit of the effective count rate is determined based on the detector's effective range, dead time correction range, and chip testing specifications; ε can be the count rate corresponding to a counting event within the current integration time, or it can be determined based on the data recording accuracy; the above parameters should be stored in association with the detector model, operating bias, detector temperature, integration time, and configuration version.

[0128] By taking the value of 0 to 1 for the mapping result, data in different counting ranges can be represented by the same numerical range. When the mapping result reaches 0 or 1, the reference mapping module 140 simultaneously records the state below the effective lower limit or above the effective upper limit, avoiding the loss of the out-of-limit direction of the original data after boundary truncation.

[0129] For parameters with non-zero reference values, such as detection efficiency and interferometric visibility, mapping can be performed based on the deviation between the current measured value and the reference value; the corresponding mapping result can be determined according to the following formula: Where x represents the current measurement value, Indicates a reference value. This indicates the permissible deviation relative to the reference value. The mapping result is represented as follows: when the current measured value equals the reference value, the mapping result is 0.5; when the current measured value reaches the reference value minus the allowable deviation and the reference value plus the allowable deviation, the mapping results are 0 and 1, respectively.

[0130] The allowable deviation is determined based on the device design specifications, product testing specifications, test equipment repeatability, or confirmed historical valid test results; for parameters whose larger values ​​indicate worse performance, the mapping relationship can be set in the opposite direction, and the evaluation direction of the parameter can be recorded in the reference configuration.

[0131] For parameters such as the second-order coherence function value, whose target value is zero or close to zero, without using the target value as a divisor, the degree of deviation can be determined according to the following formula: Where x represents the current measurement value, Indicates the target value. Indicates the allowable deviation. This indicates the degree to which the current measured value deviates from the target value; A value of 0 indicates that the current measured value is equal to the target value; A value of 1 indicates that the current measurement value has reached or exceeded the allowable deviation.

[0132] For parameters such as coupling loss, insertion loss, and extinction ratio expressed in decibels, instead of directly dividing the decibel values ​​proportionally, the difference between the current decibel value and the reference decibel value is first calculated, and then mapped according to the allowable deviation.

[0133] For parameters such as coupling loss and insertion loss, where smaller values ​​indicate better performance, the signed difference can be obtained by subtracting the current decibel value from the reference decibel value. For parameters such as extinction ratio, where larger values ​​indicate better performance, the signed difference can be obtained by subtracting the reference decibel value from the current decibel value. The evaluation direction, reference decibel value, and allowable decibel deviation for each parameter are recorded in the corresponding reference configuration.

[0134] For continuous measurement data that needs to be compared between different chips, different test batches, or different test stations, the reference mapping module 140 reads the reference center and reference scale from the reference configuration library 190 and does not re-determine the comparison benchmark based on the data of the current single chip or the current single batch.

[0135] The reference center and reference scale are determined based on valid reference data obtained under the same chip type, the same test parameters, the same instrument configuration, and the same test conditions. The data used to generate the reference center and reference scale do not include data from range saturation, operating conditions exceeding limits, time synchronization failures, equipment malfunctions, or manual interpolation.

[0136] The reference center can be the median of valid reference data, and the reference scale can be determined according to the following formula: in, This represents the first quartile of the valid reference data. This represents the third quartile of the valid reference data. This represents the minimum allowable reference scale, and s represents the actual reference scale.

[0137] The minimum reference scale is determined based on the analog-to-digital conversion resolution, the noise floor of the acquisition link, and the results of equipment repeatability tests. It is used to prevent the acquisition noise from being amplified when the reference data distribution is too narrow. The reference configuration record should include at least the following fields: reference data source, applicable chip type, applicable test equipment, sample range, abnormal data exclusion conditions, reference center, reference scale, and configuration version.

[0138] The centering and mapping results of continuous measurement data can be expressed by the following formula: as well as Determine, where x represents the continuous measurement value after baseline correction. s represents the reference center, z represents the reference scale, and z represents the degree of deviation of the current measurement value relative to the reference center. This represents the mapping result of continuous measurement data.

[0139] When the current measurement value equals the reference center, z takes the value of 0. The value is 0.5; when the current measured value is higher than the reference center, the mapping result is greater than 0.5; when the current measured value is lower than the reference center, the mapping result is less than 0.5; the 0.5 in the formula represents the middle position of the value range from 0 to 1, and is not a judgment threshold.

[0140] By using the reference center and reference scale determined before the test task is started, the actual overall differences between different chips and different batches can be preserved, avoiding the elimination of these differences by recalculating the center and scale using their respective data.

[0141] When the processing purpose is only to analyze the data changes within the same acquisition segment, the center and scale of the acquisition segment can be determined based on the median and interquartile range of the current stable data segment. The mapping result formed using the data of the current acquisition segment should be marked as an intra-segment comparison result and should not be directly used for comparison between different chips, different batches, or different test stations.

[0142] When the same test parameter has significantly different value ranges under different working conditions, the reference center and reference scale can be set separately according to the driving state, wavelength range, temperature range or other determined test states. The division conditions, boundary assignment methods and corresponding reference configurations of each test state are determined when the test task is started. Data at the boundary position are assigned to the corresponding state according to the same pre-set rules.

[0143] For operating conditions and time base data, the reference mapping module 140 reads the wavelength allowable deviation, temperature allowable deviation and synchronization allowable deviation from the reference configuration library 190, and compares the actual wavelength, actual temperature and actual synchronization deviation with the corresponding allowable values ​​to form the wavelength status, temperature status and time synchronization status.

[0144] When the reference configuration is missing, the reference configuration does not match the current chip type or test equipment, or an unrecorded configuration version change occurs during data processing, the reference mapping module 140 stops processing the affected data and generates a configuration anomaly record. The configuration anomaly record includes at least the chip identifier, test parameters, instrument identifier, current configuration version, expected configuration version, cause of anomaly, and affected acquisition segment, and is sent to the anomaly handling module 180.

[0145] like Figure 6 As shown, in step S105, the operating condition correction module 150 establishes a correlation between the mapped test data and the corresponding test wavelength, test temperature and test duration, and converts the test results to the preset reference wavelength, reference temperature and reference time.

[0146] The processing order for operating condition correction can be wavelength correction, temperature correction, and long-term drift correction in sequence. The processing order used is saved along with the configuration version. When reprocessing the same batch of data, the same processing order is used to avoid different results due to changes in the correction order.

[0147] For test parameters that require wavelength correction, the operating condition correction module 150 determines the relationship between the test parameter and wavelength based on the actual measurement data of multiple wavelength points; the wavelength scanning range is determined based on the design operating band of the chip under test, the device's free spectral region, the tuning range of the tunable laser source, and the effective response range of the detector.

[0148] The interval between wavelength scanning points is determined based on the rate of change of the device spectrum, the width of the feature to be identified, the wavelength resolution of the laser source, and the allowable test time. When the device response changes rapidly or when narrowband features need to be identified, a smaller scanning interval is used; when the device response changes gradually, a larger scanning interval can be used.

[0149] For test parameters that vary gradually with wavelength, a first-order relationship can be established based on the actual measurement points. If the first-order relationship cannot meet the accuracy requirements, a second-order relationship or a piecewise continuous relationship can be established. When using a piecewise continuous relationship, the boundaries of each segment are determined based on the device response inflection point or the changes in the actual measurement data, and adjacent segments remain continuous at the boundaries. After establishing the wavelength response relationship, the difference between the measured value and the corresponding calculated value for each wavelength is calculated, and the accuracy of the established relationship is evaluated by the maximum absolute residual or the root mean square residual.

[0150] The allowable residual value is determined based on the repeated wavelength scan results under the same test equipment and the same acquisition configuration, the repeatability of the test instrument, and the product test specifications; the allowable residual value shall not exceed the allowable measurement error of the test parameter and shall be within the range that the existing test equipment can stably achieve.

[0151] When the residual of the first-order relation is not greater than the allowable residual value, the first-order relation is used for wavelength correction; when the first-order relation does not meet the requirements but the second-order relation or the piecewise continuous relation does, the relation that meets the requirements is used; when multiple relations meet the requirements, the relation with the simpler form is preferred; when none of the preset relations meet the requirements, the corresponding data is not corrected for wavelength, and the data is marked as needing to be collected or manually verified.

[0152] When the test parameters change with wavelength mainly by increasing or decreasing a fixed value, additive correction is used. Specifically, the influence caused by the wavelength change is determined based on the difference between the reference value corresponding to the current wavelength and the reference value corresponding to the reference wavelength, and this influence is subtracted from the current test result.

[0153] When the test parameters change proportionally with wavelength, and both the test parameters and the reference value are positive, proportional correction can be used. For example, when correcting optical power, count rate, or detection efficiency, the current test result can be converted to the reference wavelength based on the ratio between the corresponding value of the reference wavelength and the corresponding value of the current wavelength.

[0154] Whether to use additive or proportional correction depends on the physical meaning of the test parameters and the wavelength scan calibration results, and is recorded in the reference configuration. Proportional correction is not used when the reference value corresponding to the current wavelength is zero, exceeds the effective range, or the wavelength response relationship does not meet the residual requirements.

[0155] After wavelength correction is completed, the operating condition correction module 150 determines the relationship between test parameters and temperature changes based on actual measurement data from multiple temperature points. The temperature test range is the intersection of the allowable test temperature range of the chip under test and the effective control range of the temperature control equipment. The reference temperature is determined based on the device design operating point, product test specifications, or equipment calibration conditions.

[0156] After the temperature control device reaches the set temperature, the system continuously monitors the difference between the actual temperature and the set temperature, as well as the temperature change over time. When the temperature difference is not greater than the allowable temperature deviation, the temperature change rate is not greater than the allowable temperature stabilization rate, and this state continues until the temperature stabilizes for a certain period of time, the system determines that the current temperature has reached a stable state.

[0157] The allowable temperature deviation is determined based on the product testing specifications and the steady-state control error of the temperature control equipment; the allowable temperature stabilization rate is determined based on the stability test results of the temperature control equipment; the temperature stabilization time is determined based on the time required for the chip, packaging structure, test fixture, and temperature sensor to reach thermal equilibrium; the above parameters are respectively associated with the chip type, temperature control equipment, test fixture, and configuration version.

[0158] For test parameters that change gradually with temperature, a first-order relationship can be established based on multiple stable temperature points; if the first-order relationship cannot meet the accuracy requirements, a second-order relationship or a piecewise continuous relationship can be established; the selection method for temperature response relationship and residual judgment method can be the same as the selection method for wavelength response relationship.

[0159] When the test parameter changes with temperature mainly by increasing or decreasing a fixed value, additive correction is performed based on the difference between the current temperature value and the reference temperature value. When the test parameter changes with temperature mainly by increasing proportionally, and the reference value is valid and not zero, correction is performed based on the ratio between the reference temperature value and the current temperature value.

[0160] When there are significant differences in the measurement results of the heating and cooling processes, heating response relationships and cooling response relationships are established separately, and the appropriate relationship is selected according to the current direction of temperature change. Whether there are significant differences is judged based on the measurement difference at the same temperature point during the heating and cooling processes and the allowable error of temperature measurement.

[0161] When a single chip is not suitable for completing a full wavelength scan or temperature scan, the measurement curve of a reference chip from the same batch or a confirmed historical valid curve can be used; the reference curve should have the same or similar device structure, material system, test parameters and test equipment configuration as the current chip.

[0162] Reference curve records must include at least the following fields: curve source, applicable chip type, applicable batch, device structure, test equipment, test conditions, generation time, verification results, and configuration version; reference curves that lack applicable scope or are unverified will not be used to generate valid calibration results.

[0163] After completing wavelength and temperature correction, the operating condition correction module 150 corrects the drift caused by long-term operation of the test equipment. Long-term drift can be caused by slow changes in laser source output, detector temperature changes, voltage or current acquisition link drift, and slow changes in the mechanical position of the test bench.

[0164] The operating condition correction module 150 selects the data range from the continuous test data where no wavelength switching, temperature adjustment, drive state switching, instrument reset, or abnormal recovery has occurred, and determines the current baseline based on the data range.

[0165] The length of the data interval used to determine the current baseline is determined based on the sampling period, test bench stability records, and the normal rate of change of the measured parameter. This data interval should contain sufficient valid data to reflect slow drift, but should not span moments when test conditions change, nor should it treat normal response changes of the measured chip as test equipment drift.

[0166] The reference baseline can be determined using the first stable data range after the start of the test task, or it can be determined using calibrated reference channel data or baseline data stored in the reference configuration library 190; the source and applicable conditions of the reference baseline are saved along with the processing results.

[0167] When long-term drift is mainly manifested as a fixed increase or decrease, additive correction is performed based on the difference between the current baseline and the reference baseline; when long-term drift is mainly manifested as a proportional change, and both the current baseline and the reference baseline are valid positive values, correction is performed based on the ratio between the two.

[0168] The drift correction method used for each test parameter is determined based on the long-term stability test results of the test bench. If the drift type cannot be determined, the current baseline is outside the effective range, or the reference baseline does not match the current test conditions, long-term drift correction will not be performed automatically, and the corresponding data will be marked as needing to be reviewed.

[0169] Only data that is in a valid synchronization state, operating stably, and has not experienced range saturation is used to determine and update the current baseline; interpolated data, data during anomaly recovery, re-acquired transitional data, and data awaiting confirmation are not used for formal baseline updates.

[0170] If the long-term stability records and repeated test results of the test bench show that the change of a certain test parameter within the test period does not exceed the allowable measurement error of that parameter, long-term drift correction is not required; the corresponding records should include at least the fields of test duration, test parameter, allowable measurement error, stability test results, judgment conclusion and configuration version.

[0171] When the test task can obtain measurement data under multiple wavelengths and multiple temperature combinations, the corresponding relationship between the test parameters and the wavelength and temperature can be established, and the data can be converted to the reference wavelength and reference temperature accordingly. The combination relationship used is only applicable to the actual measurement range, and no unfounded extrapolation is made to wavelengths or temperatures outside the measurement range.

[0172] like Figure 7 As shown, in step S106, the data encapsulation module 160 associates and saves the original data, preprocessing results, mapping results, standardization results and processing information, so that each standardization result can be queried to find the corresponding original measurement data, processing parameters and reference configuration.

[0173] The raw data record should include at least the following fields: raw measurement value, physical unit, acquisition time, event timestamp, instrument identifier, channel identifier, instrument status, and acquisition segment number. The raw measurement value remains unchanged in subsequent processing and is not overwritten by preprocessing results, standardized results, or reacquisition results.

[0174] Preprocessing records, depending on the type of test data, include at least the following fields: dark count baseline, dark count subtraction result, dead time correction result, after-pulse correction result, continuous measurement baseline, baseline correction result, filtering result, wavelength deviation, temperature deviation, and mapping result. If a certain processing item is not applicable to the current test data, the corresponding field will be recorded as inapplicable and will not be replaced with a zero value.

[0175] The standardized results record should include at least the following fields: standardized value, reference wavelength, reference temperature, reference baseline, operating condition correction status, boundary status, and quality status. The standardized results should correspond to the reference values, permissible deviations, reference curves, and their configuration versions used.

[0176] The processing information record should include at least the following fields: chip number, test parameter number, wafer coordinates, batch number, test task number, reference configuration version, processing program version, processing parameters, exception records, and output status.

[0177] The data encapsulation module 160 establishes a connection between raw data, preprocessing results, standardized results, and processing information through chip number, test parameter number, acquisition segment number, and data point number; among them, the acquisition segment number is used to distinguish different continuous acquisition processes in the same test task, and the data point number is used to distinguish different measurement data within the same acquisition segment.

[0178] When the same original data segment is reprocessed using different reference configurations or processing parameters, the system generates a new result version and records the reference configuration version, processing program version, and generation time of the new result; the original processing result is retained and is not overwritten by the new processing result, and each result version remains associated with the corresponding original data.

[0179] The data can be stored using relational tables, columnar files, or object storage. The specific storage method can be determined based on the amount of data and the testing platform, but the meaning of each field, the relationships between data, and the correspondence between result versions should remain consistent.

[0180] Data records may also include integrity check values, which are used to check whether data has been missing or unexpectedly changed during transmission or storage. Integrity check results are only used to determine whether data records are complete and are not used as a basis for judging the accuracy of measurement values ​​or the consistency of measurement results from different testing devices.

[0181] Data records can also be linked to information such as wafer location, process batch, device unit identifier, and system test indicators. When the standardization results are abnormal, the affected chip, wafer area, test channel, or process batch can be determined based on the above-mentioned linked information to narrow down the scope of retesting and cause investigation.

[0182] like Figure 8 As shown, in step S107, the quality verification module 170 performs integrity checks, configuration version checks, processing record checks, abnormal status checks, and recalculation verification on the standardized data to be output.

[0183] The integrity check is used to confirm that each standardized result can be associated with the corresponding raw data and processing records; the information associated with each standardized result includes at least the following fields: chip identifier, test parameter identifier, instrument identifier, channel identifier, acquisition time, acquisition segment number, reference configuration version, and processing program version; data lacking necessary associated information will not be written to the valid results area.

[0184] Configuration version checks are used to confirm that the reference values, allowable deviations, judgment thresholds, reference curves, and processing parameters used for the same result belong to the same reference configuration version. When the reference configuration is changed during data processing, a new configuration version should be generated. If configuration changes are not recorded or different version parameters are used in combination, the corresponding result will be marked as a configuration anomaly.

[0185] The processing record check is used to confirm that the raw data has completed data classification, preprocessing, reference mapping and operating condition correction, and that the data association between each processing step is complete; if classification anomalies have not been resolved, necessary processing parameters are missing, or the processing that should be performed has not been completed, the corresponding data will not be output as a valid result.

[0186] Abnormal status checks are used to confirm that states such as range saturation, count rate exceeding limits, operating conditions exceeding limits, synchronization abnormalities, mapping results reaching boundaries, and exceeding the calibration range have been recorded. Data with abnormal statuses can be saved or output with attached status tags according to the test task requirements, or it can enter the re-acquisition process, but it must not be processed as data without abnormalities.

[0187] Recalculation and verification refers to reading the data to be verified from the original data record, reclassifying, preprocessing, referencing, and correcting the data according to the original processing program version, reference configuration version, processing parameters, and numerical precision, and then comparing the recalculation results with the output results.

[0188] For processes such as logarithmic mapping, range constraints, and bounded mappings where the original data cannot be uniquely recovered from the standardization results, the original data is not calculated backwards from the standardization results. Instead, the saved original data is used as the starting point for recalculation.

[0189] When the same original data is recalculated using the same processing program, reference configuration, processing parameters, and storage precision, the recalculation result should be consistent with the original result. When minor differences arise due to processor numerical operations or intermediate result rounding, the allowable difference can be taken as the smallest storage unit of the standardized result. The smallest storage unit is determined according to the data format, significant digits, and storage precision of the result field.

[0190] Recalculation and verification are used to check whether the data processing results can be obtained repeatedly, not to determine whether the remeasurement results are consistent. When re-acquiring data for the same chip, the allowable difference in the remeasurement results should be determined based on the repeatability of the test instrument, the repositioning error of the probe or fiber optic cable, and the temperature recovery error. The corresponding allowable difference can come from the equipment technical documents, repeatability test results, or product test specifications.

[0191] When the quality verification passes, the quality verification module 170 marks the corresponding data as valid and writes it to the valid result area; the verification record includes at least the fields of verification time, verification item, verification result, reference configuration version, processing program version and output status.

[0192] If the quality verification fails, the corresponding data remains isolated and does not overwrite the already saved valid results; if the verification fails due to incorrect data classification, return to step S102 for reclassification; if the verification fails due to incorrect preprocessing parameters, return to step S103 for reprocessing; if the verification fails due to incorrect reference configuration, return to step S104 for remapping; if the verification fails due to incorrect operating condition correction, return to step S105 for correction; if the verification fails due to incomplete data association, return to step S106 to re-establish the data association.

[0193] After reprocessing, a new version of the result is generated, and the reason for reprocessing, the reference configuration version used, the processor version, and the generation time are recorded; the original processing result and its failure reason are retained and are not overwritten by the new result.

[0194] like Figure 9 As shown, the exception handling module 180 is used to handle the exceptions generated in steps S101 to S107; according to the scope of the impact of the exception, the exception can be divided into channel-level exceptions, test bench-level exceptions and batch-level exceptions.

[0195] The anomaly record includes at least the following fields: anomaly type, scope of impact, instrument identifier, channel identifier, time of anomaly occurrence, affected acquisition segment, processing method, recovery conditions, processing result, and configuration version. The anomaly processing module 180 determines the range of data that needs to be isolated based on the anomaly record and prevents the abnormal data from continuing to enter subsequent processing.

[0196] Channel-level anomalies refer to anomalies that affect only one test channel or one data stream, such as dark count anomalies in single-photon detection channels, range saturation in continuous measurement channels, temperature sensor anomalies, missing single data streams, or incorrect channel classification.

[0197] When a channel-level anomaly occurs, the anomaly handling module 180 isolates the affected channel; whether unaffected channels continue to collect data depends on the correspondence between the test parameters and the test task configuration; if the data from the affected channel is a necessary input for other test parameters, the relevant test is paused; if the affected channel is independent of other test parameters, data from unaffected channels can continue to be collected.

[0198] When the test bench is equipped with a backup channel, the switch to the backup channel can only be made if the backup channel has been calibrated, the correspondence between the backup channel and the original channel is correct, and the current wavelength, temperature and optical path status still meet the test requirements. After the switch, a new acquisition segment number is generated, and information such as the original channel abnormality, backup channel identification, switching time and acquisition configuration after the switch are recorded.

[0199] If the observed count rate exceeds the effective correction range of the detector dead time, the incident light power can be reduced, the integration time shortened or adjusted, the detector range switched, or the data reacquired, depending on the adjustment capability of the test equipment. The adjusted acquisition conditions should still meet the requirements of the counting statistics accuracy and test specifications.

[0200] If the continuous measurement channel experiences range saturation, the measurement range can be adjusted, the input signal reduced, or the data reacquired. If the filtering result does not meet the requirements for response amplitude, inflection point, or settling time, the filtering parameters can be adjusted, or the original data can be retained and reprocessed using new processing parameters without directly modifying the original measurement data.

[0201] Test bench level anomalies refer to anomalies that simultaneously affect multiple test channels or the current test task, such as loss of common clock lock, loss of laser light source wavelength lock, instability of temperature control equipment, simultaneous anomalies of multiple channels, or continuous interruption of communication links.

[0202] When a test bench-level anomaly occurs, the anomaly handling module 180 suspends the current test task and sends a recovery, relock, or reset command to the relevant equipment. After recovery, the time synchronization status, wavelength status, temperature stability status, instrument range, and channel configuration should be rechecked. When all statuses meet the test requirements again, the acquisition should resume from the most recent complete acquisition segment with valid quality status.

[0203] Data generated between the occurrence of an anomaly and the recovery of the equipment is retained as abnormal data and is not used to update the reference configuration, dark count baseline, continuous measurement baseline, and long-term drift baseline, nor is it written to the valid results area.

[0204] Batch-level anomalies refer to multiple chips or multiple test channels within the same test batch exhibiting the same type of anomaly consecutively, reaching the preset batch review conditions.

[0205] Batch review criteria can be determined based on parameters such as the number of abnormal chips, the number of abnormal channels, the number of consecutive occurrences, the duration of the abnormality, and the affected proportion. The values ​​of each parameter are based on the normal operation records of the test line, confirmed equipment failure records, process abnormality records, equipment repeatability test results, and product test specifications.

[0206] Batch review conditions should be associated with and saved as an anomaly type, applicable equipment, applicable chip type, and configuration version, and determined when the test task is started; batch review conditions must not be changed during the test without being recorded.

[0207] When a common clock, temperature control device, or laser source malfunctions simultaneously affect multiple chips, batch review can be triggered directly based on the number of affected channels, duration, and device status. For cases where only a single device parameter is abnormal, batch review can be triggered after the number of abnormal chips and the number of consecutive occurrences both reach the corresponding conditions.

[0208] Once the batch review is triggered, the valid output of new results for that batch will be suspended, and the test equipment, test configuration, optical path connection and related process status will be checked. Before the review is completed, the collected data will be retained, but will not be used to update the formal reference configuration.

[0209] For delays, duplicates, or out-of-order data transmission, the acquisition and time alignment module 110 sorts and deduplicates the data based on the data sequence number, event timestamp, and acquisition segment number. The data record can also save the receiving time to determine the transmission delay, but it does not use the receiving time to replace the instrument acquisition time.

[0210] The waiting time limit is determined based on normal data transmission latency, buffer capacity, acquisition cycle, and test cycle time. Data that is not acquired after the waiting time limit is marked as missing data and will not be merged with subsequent acquisition segments to avoid incorrect correspondence between data from different acquisition processes.

[0211] For continuously recorded data such as wavelength, temperature, and optical power, as well as continuously measured data, before completing missing data, determine the duration of the missing data, the rate of change of the data before and after the missing data, and whether wavelength adjustment, temperature adjustment, drive state switching, or abnormal recovery occurred during the missing period.

[0212] The upper limit of the missing duration is determined based on the allowable completion error of the corresponding parameters and the maximum rate of change under normal conditions. When the possible change amount estimated based on the missing duration and the maximum rate of change is not greater than the allowable completion error, and no test condition switching occurs during the missing period, interpolation completion can be performed based on the valid data before and after the missing.

[0213] The data completion process should record information such as the completion method, the start and end times of the missing data, the valid data used before and after the missing data, the estimation error, and the quality status. The allowable error for completion should be determined based on the product testing specifications, the sensitivity of the test parameters to changes in operating conditions, and the repeatability of the measuring equipment.

[0214] Data that does not meet the above completion conditions will not be automatically completed. Instead, the affected area will be re-acquired or the corresponding data will be isolated. Completed data can be used to maintain the continuity of the data sequence, but it will not be used to generate or update the formal reference configuration, dark count baseline, and long-term drift baseline.

[0215] For single-photon event data and coincidence count data, missing events are not supplemented using ordinary linear interpolation. If missing events affect the count rate, coincidence rate, or other statistical results, the corresponding integration interval is re-acquired. If re-acquisition is not possible, the corresponding statistical results are marked as invalid.

[0216] When the dark count, dead time, or afterpulse state of the detector exceeds the calibration range, stop using the data from the affected channel to update the dark count baseline, and re-stabilize the detector temperature, recalibrate, adjust the operating bias, or switch to a calibrated standby detector according to the equipment status; the restored detector status should meet the applicable conditions of the corresponding calibration configuration again.

[0217] When the wavelength meter is not locked or the temperature control device does not meet the stability conditions, the relevant operating condition data is marked as invalid, and the affected test data will not be formally corrected for wavelength or temperature. After the equipment is restored, the affected range will be re-acquired, and the restored wavelength or temperature data will not be used to replace the actual operating conditions when the abnormality occurred.

[0218] When the reference configuration version is updated, the original data, the original preprocessing results, and the original standardization results are retained, and the affected data is reprocessed according to the new reference configuration. The reprocessing generates a new result version, and the reference configuration version, processing program version, and generation time are recorded for the old and new results, respectively. The original results are not overwritten.

[0219] For example, in one specific application, the object under test is a photonic quantum chip with a phase modulation unit, an interference unit, an optical coupling port, and a single-photon output channel; the test task is to obtain test parameters such as single-photon count rate, interference visibility, phase driving voltage, coupling loss, center wavelength, and chip temperature.

[0220] When the test task is started, the system 100 reads and fixes the chip number, instrument and channel correspondence, reference wavelength, reference temperature, reference configuration version and processing program version used in this test; before formal acquisition, the incident light path is closed or blocked, and dark count data is acquired under the condition that the detector working bias, detector temperature and integration time meet the test requirements; after the dark count data reaches the aforementioned stable conditions, the dark count baseline is determined and the formal test begins.

[0221] System 100 divides the data generated by different test instruments into corresponding acquisition segments based on hardware trigger signals, common clock information, and instrument acquisition time; single photon counts and coincidence counts are classified into photon event data, driving voltage, optical power, and coupling loss are classified into continuous measurement data, and wavelength, temperature, and time synchronization status are classified into operating condition and time base data.

[0222] For photon event data, system 100 sequentially performs dark count subtraction, dead time correction, after-pulse correction, and probe saturation check; for continuous measurement data, system 100 performs baseline correction, abnormal data processing, filtering, and range check; for operating condition and time base data, system 100 determines the difference between the actual wavelength and the reference wavelength, the difference between the actual temperature and the reference temperature, the test duration, and the synchronization status.

[0223] After preprocessing, System 100 calls the corresponding reference configuration according to the physical meaning of the test parameters; for count data spanning multiple orders of magnitude, logarithmic mapping is used; for continuous measurement data such as voltage, current and optical power, mapping is performed using a pre-determined reference center and reference scale; for decibel parameters such as coupling loss and insertion loss, the difference between the current decibel value and the reference decibel value is used; for parameters with a target value of zero or close to zero, the deviation of the current measurement value from the target value is used.

[0224] System 100 converts the mapped test results to reference wavelength, reference temperature, and reference time based on the actual test wavelength, actual test temperature, and test duration. The wavelength correction relationship, temperature correction relationship, and long-term drift correction method used are all derived from the reference configuration used in this test and are saved along with the processing results.

[0225] The system 100 associates and saves raw measurement data, preprocessing results, mapping results, standardization results, and processing records. The raw data records include at least the fields of measured value, physical unit, acquisition time, instrument identifier, channel identifier, and acquisition segment number. The processing records include at least the fields of dark count baseline, correction parameters, reference configuration version, processing program version, operating condition calibration status, and quality status.

[0226] After the output results have undergone integrity checks, configuration version checks, abnormal status checks, and recalculation verification, they are written to the valid results area. The recalculation verification starts from the original measurement data and uses the same reference configuration, processing parameters, and processing procedures to recalculate, without reverse-engineering the original data based on the standardized results.

[0227] When the wavelength meter loses lock during wavelength scanning, the system 100 pauses the acquisition of the affected wavelength point and subsequent wavelength points, and records the time of the lock loss and the affected range. After the wavelength meter relocks and the wavelength status reaches a stable condition, it re-acquires the wavelength point that was last fully acquired and has a valid quality status before the lock loss. The data acquired during the lock loss period is retained, but is not used for wavelength correction, long-term baseline updates, or valid result output.

[0228] A complete test task can simultaneously acquire photon event data, continuous measurement data, and operating condition and time base data; if a test task does not generate one of the performance data, only the actual acquired data is processed accordingly, but each performance data should still be associated with wavelength, temperature, acquisition time and instrument status that meet the test requirements.

[0229] This method can be used for photonic quantum chips formed from silicon-based, lithium niobate, and other integrated photonic materials. When applied to different material systems or different device structures, the reference wavelength, reference temperature, effective range, allowable deviation, and correction relationship should be re-determined according to the operating band, test parameters, and test equipment of the corresponding device. Reference configurations with inconsistent applicable conditions should not be directly adopted.

[0230] By establishing a time correspondence between data from different testing instruments and preprocessing the data according to their physical sources, we can avoid incorrect combinations of data from different acquisition times, different testing conditions, or different configuration versions. By adopting appropriate mapping methods for count data, continuous measurement data, decibel parameters, and zero target parameters, we can preserve changes in the low count range, the direction of parameter deviations, and the actual differences between different chips and different batches.

[0231] By linking and saving raw data, processing parameters, reference configurations, standardized results, and quality status, and setting up mechanisms for anomaly isolation, re-collection, recalculation, and version retention, standardized results can be reviewed and traced, and abnormal data can be prevented from directly entering the valid results area.

[0232] This method can be derived from... Figure 10 The test master control device shown executes the test; the test master control device includes at least a processor, a memory, an instrument acquisition interface, a communication interface and a clock interface; the memory is used to store test tasks, reference configurations, processing programs and test data, and the processor is used to call the processing program to execute steps S101 to S107.

[0233] Each processing function can be executed by the same main test control device, or by the test station controller and the main test control server working together. When multiple devices work together, the data transmitted between the devices must include at least the fields of acquisition segment number, data source information, reference configuration version, processing program version and quality status, in order to maintain the correspondence between the processing results of each device.

[0234] The modules in the attached diagram are used to illustrate the relationships between various processing functions. It is not required that each module be implemented by independent hardware; multiple functions can be executed by the same processor, and the same function can also be completed by multiple communicating devices.

[0235] The specific parameters in this embodiment are determined according to the chip under test, the test equipment, and the test task. Each parameter record includes at least the following fields: parameter name, parameter value, physical unit, determination basis, applicable chip type, applicable test equipment, applicable operating conditions, and configuration version. The basis for parameter values ​​can be device design documents, equipment technical documents, equipment calibration records, repeatability test results, or product test specifications. Parameters lacking a basis for value selection or applicable scope are not used to generate valid test results.

[0236] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0237] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for standardizing test data processing of optical quantum chips, characterized in that, include: Acquire photonic quantum chip test data generated by different testing instruments, establish the time correspondence between the photonic quantum chip test data, and associate the data source information; Based on the instrument channel usage and data characteristics, the test data of the optical quantum chip is divided into photon event data, continuous measurement data, and operating condition and time base data; The photon event data, the continuous measurement data, and the operating condition and time base data are preprocessed respectively. Invoke the reference configuration locked when the test task starts, and map the preprocessed data according to the test parameter type; Based on the actual test wavelength, actual test temperature and test duration, the mapped data is converted to the reference operating conditions and long-term drift correction is performed to obtain standardized data; The system associates and saves the original test data, preprocessing results, mapping results, standardized data, processing parameters, and quality status. It also verifies the associated and saved data. If the verification passes, the standardized data is output. If the verification fails, the affected data is isolated and returned to the corresponding processing stage.

2. The method for standardizing test data processing of optical quantum chips according to claim 1, characterized in that, Establishing the time correspondence between the test data of the optical quantum chip includes: Based on hardware trigger flags, common clock count values, or instrument-recorded acquisition times, the data generated by different testing instruments are divided into multiple acquisition segments; The original acquisition time of each data stream is retained, and the operating condition data corresponding to the performance data is determined based on the acquisition time of the performance data. The synchronization deviation between different test channels is compared with the corresponding allowable synchronization deviation. A time correspondence is established when the synchronization deviation does not exceed the allowable synchronization deviation. When the synchronization deviation exceeds the allowable synchronization deviation, the synchronization anomaly and the affected data range are recorded.

3. The method for standardizing test data processing of optical quantum chips according to claim 2, characterized in that, The determination of the operating condition data corresponding to the performance data includes: When the acquisition time of the performance data is between the acquisition times of two adjacent operating condition data, the interpolation error is determined based on the rate of change and time interval of the two adjacent operating condition data. When the interpolation error does not exceed the allowable error of the corresponding operating condition parameter, and no wavelength switching, temperature adjustment, optical path switching or abnormal recovery occurs between the two adjacent operating condition data, the corresponding operating condition value is determined based on the two adjacent operating condition data. If the time of collection of the performance data is outside the time range of existing operating condition data, or if there is a state switch, data loss, or abnormal change between two adjacent operating condition data, the performance data will be marked as operating condition data missing.

4. The method for standardizing test data processing of optical quantum chips according to claim 1, characterized in that, The process of dividing the test data of the optical quantum chip into photon event data, continuous measurement data, and operating condition and time base data includes: Based on the instrument type, instrument channel, and channel purpose recorded in the test task, the test data of the optical quantum chip are preliminarily classified. The preliminary classification results are verified based on the physical unit, data format, value type, sampling interval, variation, and range status of the data. When the purpose of the instrument channel matches the data characteristics, the classification result is confirmed; when the purpose of the instrument channel does not match the data characteristics, a classification anomaly record is generated and the corresponding data segment is isolated.

5. The method for standardizing test data processing of optical quantum chips according to claim 1, characterized in that, The preprocessing of the photon event data includes: Dark count data is collected when the incident light path is closed and the detection channel is in a dark state. The dark count data is divided into multiple statistical windows, and the dark count baseline is determined according to the dark count rate of each statistical window. Subtract the dark count baseline from the observed count rate; When the observed count rate after deducting the dark count is within the correction range confirmed by the detector calibration, it is corrected according to the dead time correction relationship and the after-pulse calibration relationship corresponding to the detector. When the observation count rate exceeds the correction range, the detector is saturated, or the correction parameter is missing, the generation of the corrected count rate is stopped, and the processing status of the affected data is marked.

6. The method for standardizing test data of a quantum chip according to claim 1, characterized in that, The preprocessing of the continuous measurement data includes: Candidate data segments that did not experience scanning transitions, channel switching, drive state changes, temperature adjustments, instrument resets, or abnormal recovery were selected from the continuous measurement data. A stable data segment is determined based on the fluctuation range, rate of change, and duration of the measured values ​​within the candidate data segment, and the acquisition link baseline is determined based on the stable data segment; The baseline of the acquisition link at the corresponding time moment is subtracted from the continuous measurement data, and filtering is performed based on the response bandwidth of the device under test, the sampling rate of the acquisition device, and the noise spectrum of the acquisition link; Compare the response amplitude, inflection point, and settling time before and after filtering. If the comparison result does not meet the corresponding allowable conditions, mark the filtering result as abnormal. When the continuous measurement data reaches the range boundary or the acquisition device outputs an overload state, the corresponding data segment is marked as range saturation.

7. The method for standardizing test data processing of optical quantum chips according to claim 1, characterized in that, The mapping of preprocessed data according to test parameter type includes: For count data spanning multiple orders of magnitude, a logarithmic mapping is performed based on the lower bound of the count rate, the upper bound of the count rate, and the minimum positive quantity used to handle the uncounted state; For test parameters with non-zero reference values, mapping is performed based on the deviation between the current measured value and the reference value, as well as the corresponding allowable deviation. For test parameters whose target value is not suitable as a divisor, mapping is performed based on the degree of deviation between the current measured value and the target value; For test parameters expressed in decibels, a mapping is performed based on the difference between the current decibel value and the reference decibel value; For continuous measurement data used for comparisons between different chips, different test batches, or different test stations, a bounded mapping is performed based on the reference center and reference scale.

8. The method for standardizing test data processing of optical quantum chips according to claim 7, characterized in that, The reference center and the reference scale are determined based on reference data obtained under the same chip type, the same test parameters, the same instrument configuration, and the same test conditions. The reference data used to determine the reference center and the reference scale does not include range saturation data, operating condition over-limit data, time synchronization failure data, equipment abnormality data, and interpolation data; The reference center and the reference scale are written into the reference configuration before the test task starts and remain unchanged during this data processing.

9. The method for standardizing test data of a quantum chip according to claim 1, characterized in that, The process of converting the mapped data to the reference operating condition includes: The response relationship of the test parameters with wavelength variation is determined based on multiple wavelength measurement points, and the response relationship that satisfies the residual allowable condition is selected based on the residual between the measured value of each wavelength and the corresponding result of the response relationship. Based on the physical meaning of the test parameters and the wavelength scanning calibration results, additive or proportional correction is used to convert the mapped data to the reference wavelength. The response relationship of the test parameters to temperature changes is determined based on multiple stable temperature points, and additive or proportional correction is used to convert the wavelength-corrected data to the reference temperature. When the measurement difference between the heating and cooling processes does not meet the allowable conditions for temperature measurement, heating response relationship and cooling response relationship are established respectively, and the corresponding response relationship is selected according to the current direction of temperature change.

10. The method for standardizing test data processing of optical quantum chips according to claim 1, characterized in that, The long-term drift correction includes: The current baseline is determined from continuous test data that has never experienced wavelength switching, temperature adjustment, drive state switching, instrument reset, or abnormal recovery. The reference baseline is determined from the stable data range after the start of the test task, the calibrated reference channel data, or the reference configuration. When long-term drift exhibits additive changes, correction is made based on the difference between the current baseline and the reference baseline. When long-term drift manifests as a proportional change, and the current baseline and the reference baseline meet the corresponding applicable conditions, correction is performed based on the ratio between the current baseline and the reference baseline; Among them, range saturation data, synchronization failure data, unstable operating condition data, interpolation data, data during abnormal recovery, re-acquired transitional data, and data to be confirmed are not used to determine or update the current baseline.

11. The method for standardizing test data processing of optical quantum chips according to claim 1, characterized in that, The associated data storage includes original test data, preprocessing results, mapping results, standardized data, processing parameters, and quality status. The associated data storage is also verified, including: By using chip number, test parameter number, acquisition segment number, and data point number, the raw test data, preprocessing results, mapping results, standardized data, and processing information are linked together. When reprocessing the same original data segment using different reference configurations or processing parameters, a new version of the result is generated, the original processing result is retained, and each version of the result is associated with the corresponding original test data. Perform integrity checks, configuration version checks, processing record checks, abnormal status checks, and recalculation verification on the standardized data to be output; The recalculation and verification process starts with the original test data, re-executes the data processing according to the original processing program version, reference configuration version, processing parameters, and storage precision, and compares the recalculation results with the standardized data to be output.

12. The method for standardizing test data processing of optical quantum chips according to claim 1, characterized in that, The process of isolating the affected data and returning it to the corresponding processing stage includes: Based on the scope of their impact, anomalies are classified into channel-level anomalies, test bench-level anomalies, or batch-level anomalies. When a channel-level anomaly occurs, the affected channel is isolated, and the decision on whether to continue collecting data from unaffected channels is made based on the dependencies between test parameters. When a test bench-level anomaly occurs, the current test task is paused. After the time synchronization status, wavelength status, temperature status, instrument range, and channel configuration are restored, the data is reacquired from the acquisition segment where the data acquisition is complete and the quality status meets the test requirements. When multiple chips or multiple channels within the same test batch exhibit the same type of anomaly and meet the batch review criteria, the output of new results for that test batch is paused and batch review is performed. For continuous data that does not meet the completion criteria and for missing photon event data, the affected area should be re-acquired or the corresponding data should be isolated.

Images