A method for detecting power supply voltage tolerance of a QKD device and an electronic device

Abstract

The application provides a QKD device power voltage tolerance detection method and an electronic device, including: controlling a programmable power supply to provide a power voltage for a to-be-tested quantum key distribution (QKD) device, and collecting multiple input voltages of the to-be-tested QKD device; constructing multiple groups of input variables according to the multiple input voltages, wherein each group of input variables includes at least one input voltage; performing multiple-output prediction on the multiple groups of input variables by using a preset multiple-output Gaussian process model to obtain multiple groups of output variables, each group of output variables including multiple-dimension performance prediction parameters corresponding to one input voltage; performing power voltage tolerance analysis on the to-be-tested QKD device according to the multiple input voltages and the multiple-dimension performance prediction parameters to obtain a power voltage tolerance range of the to-be-tested QKD device. The multiple input voltages and the multiple-dimension performance prediction parameters are used to accurately locate a device voltage safe working interval, and the accuracy of abnormality detection is improved.

Description

Technical Field

[0001] This invention relates to the field of QKD equipment technology, and more specifically, to a power supply voltage tolerance detection method and electronic equipment for QKD equipment. Background Technology

[0002] The testing technology for QKD devices is a crucial bridge connecting theoretical security with practical applications. Its development path is characterized by "from device-dependent to device-independent, from single-link to network collaboration, and from laboratory to commercialization." Existing testing technologies have several shortcomings: limitations of static testing (traditional tests only cover the nominal voltage range and cannot simulate the impact of extreme power grid fluctuations on quantum devices, leading to key synchronization failures due to voltage anomalies in practical applications); isolated testing of single devices (failing to utilize multi-device collaborative data), failing to distinguish between voltage fluctuations and individual device differences, resulting in a high false positive rate; and low manual efficiency (relying on engineers to manually adjust test points, with testing a single device taking more than 2 hours and unable to dynamically adapt to batch differences). Summary of the Invention

[0003] The purpose of this invention is to address the shortcomings of the prior art by providing a power supply voltage tolerance detection method and electronic device for QKD devices. This method collects the input voltage of the QKD device, constructs multiple sets of input variables, and uses a multi-output Gaussian process model to predict multi-dimensional performance parameters. This allows for accurate positioning of the device's safe operating voltage range, avoids misjudgment based on a single parameter, and significantly improves the accuracy of anomaly detection.

[0004] To achieve the above objectives, the technical solutions adopted in the embodiments of this application are as follows:

[0005] In a first aspect, embodiments of this application provide a method for detecting the power supply voltage tolerance of a QKD device, the method comprising:

[0006] The programmable power supply is controlled to provide power voltage to the QKD device under test, and multiple input voltages of the QKD device under test are acquired.

[0007] Based on the multiple input voltages, multiple sets of input variables are constructed, wherein each set of input variables includes at least one input voltage;

[0008] A preset multi-output Gaussian process model is used to perform multi-output prediction on the multiple sets of input variables to obtain multiple sets of output variables. Each set of output variables includes: multiple dimensions of performance prediction parameters corresponding to the input voltage.

[0009] Based on the multiple input voltages and the corresponding performance prediction parameters of the multiple dimensions, a power supply voltage tolerance analysis is performed on the QKD device under test to obtain the power supply voltage tolerance range of the QKD device under test.

[0010] In an optional implementation, the method further includes:

[0011] Collect multiple ambient temperatures of the QKD device under test;

[0012] The step of constructing multiple sets of input variables based on the multiple input voltages includes:

[0013] Based on the multiple input voltages and the multiple ambient temperatures, the multiple sets of input variables are constructed, and each set of input variables further includes: the ambient temperature corresponding to one of the input voltages.

[0014] In an optional implementation, the performance prediction parameters of the multiple dimensions include: multiple basic performance prediction parameters and extended performance prediction parameters;

[0015] The multiple basic performance prediction parameters include: average key generation rate prediction parameter and quantum channel error rate prediction parameter; the extended performance prediction parameters include: system line loss margin prediction parameter.

[0016] In an optional implementation, the step of performing power supply voltage tolerance analysis on the QKD device under test based on the plurality of input voltages and the corresponding performance prediction parameters of the plurality of dimensions to obtain the power supply voltage tolerance range of the QKD device under test includes:

[0017] Based on the multiple input voltages and the corresponding performance prediction parameters of the multiple dimensions, determine the voltage range in which all multiple dimensions satisfy the corresponding performance thresholds from the multiple input voltages;

[0018] The power supply voltage tolerance range is determined based on the voltage range that meets the corresponding performance threshold.

[0019] In an optional implementation, determining the power supply voltage tolerance range based on the voltage range that satisfies the corresponding performance threshold includes:

[0020] Based on the multiple input voltages and the corresponding performance prediction parameters of the multiple dimensions, determine the abnormal voltage ranges that do not meet the corresponding performance thresholds from the multiple input voltages;

[0021] The voltage range that meets the corresponding performance threshold is corrected based on the abnormal voltage range, and the corrected voltage range is determined as the power supply voltage tolerance range.

[0022] In an optional implementation, the method further includes:

[0023] Based on the multiple input voltages and the corresponding performance prediction parameters of the multiple dimensions, multiple performance curves of the QKD device under test are generated, and abnormal voltage points are marked in each performance curve. Each performance curve includes: an average key generation rate prediction curve and a quantum channel bit error rate prediction curve.

[0024] In an optional implementation, each set of output variables further includes: operating current prediction parameters; the method further includes:

[0025] Based on the multiple input voltages and the predicted operating current parameters, a curve showing the relationship between input voltage and operating current is generated.

[0026] In an optional implementation, before using a preset multi-output Gaussian process model to perform multi-output prediction on the multiple sets of input variables to obtain multiple sets of output variables, the method further includes:

[0027] Multiple sets of historical operating parameters are obtained for the QKD device under test and the accompanying QKD device. Each set of historical operating parameters includes: a historical input voltage, the corresponding performance detection parameters of the multiple dimensions, and the corresponding historical operating current.

[0028] Using a preset initial multi-output Gaussian process model, multi-output prediction is performed on the historical input voltage of the QKD device under test to obtain the performance prediction parameters and current prediction parameters of the QKD device under test under the corresponding historical input voltage.

[0029] Using the preset initial multi-output Gaussian process model, multi-output prediction is performed on the historical input voltage of the QKD device under test to obtain the performance prediction parameters and current prediction parameters of the QKD device under the corresponding historical input voltage.

[0030] Based on the performance prediction parameters, current prediction parameters, performance detection parameters, and historical operating current of the QKD device under test at the corresponding historical input voltage, and the performance prediction parameters, current prediction parameters, performance detection parameters, and historical operating current of the accompanying QKD device at the corresponding historical input voltage, the preset initial multi-output Gaussian process model is iteratively trained to obtain the preset multi-output Gaussian process model.

[0031] In an optional implementation, acquiring multiple sets of historical operating parameters of the QKD device under test and the accompanying QKD device includes:

[0032] Multiple sets of initial historical operating parameters are obtained for the QKD device under test and the accompanying QKD device. Each set of initial historical operating parameters includes: an initial historical input voltage, the corresponding initial performance detection parameters of the multiple dimensions, and the corresponding initial historical operating current.

[0033] The multiple sets of initial historical operating parameters are scanned according to the first preset step size to determine the abnormal voltage range;

[0034] The multiple sets of initial historical operating parameters within the abnormal voltage range are scanned according to the second preset step size to determine the multiple sets of historical operating parameters.

[0035] Secondly, embodiments of this application also provide an electronic device, including: a processor, a storage medium, and a bus. The storage medium stores program instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the program instructions to perform the steps of the QKD device power supply voltage tolerance detection method as described in any of the first aspects above.

[0036] The beneficial effects of this application are:

[0037] This application provides a power supply voltage tolerance detection method and electronic device for QKD devices, comprising: controlling a programmable power supply to provide power supply voltage to the QKD device under test, and collecting multiple input voltages of the QKD device under test; constructing multiple sets of input variables based on the multiple input voltages, wherein each set of input variables includes at least one input voltage; using a preset multi-output Gaussian process model to perform multi-output prediction on the multiple sets of input variables to obtain multiple sets of output variables, each set of output variables including: multiple dimensions of performance prediction parameters corresponding to one input voltage; and performing power supply voltage tolerance analysis on the QKD device under test based on the multiple input voltages and the corresponding multiple dimensions of performance prediction parameters to obtain the power supply voltage tolerance range of the QKD device under test.

[0038] The method in this application collects the input voltage of the QKD device, constructs multiple sets of input variables, and uses a multi-output Gaussian process model to predict multi-dimensional performance parameters. This method can accurately locate the safe operating range of the device voltage, avoid misjudgment of a single parameter, significantly improve the accuracy of anomaly detection, and proactively avoid risks such as increased bit error rate and decreased key generation rate caused by voltage anomalies. By comprehensively analyzing multi-dimensional performance to capture parameter correlations, it provides a quantitative basis for power supply design optimization, efficiently and comprehensively ensuring the stable operation and reliable performance of the QKD device under test under voltage fluctuations. Furthermore, it automates the entire process of power supply characteristic testing of the QKD device under test, improving testing efficiency and making it suitable for multi-device collaborative testing and batch device testing. Attached Figure Description

[0039] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 A schematic diagram of a power supply voltage tolerance detection system for QKD equipment provided in this application embodiment;

[0041] Figure 2 This is one of the flowcharts illustrating a power supply voltage tolerance detection method for a QKD device provided in an embodiment of this application;

[0042] Figure 3 A second schematic flowchart illustrating a power supply voltage tolerance detection method for a QKD device provided in this application embodiment;

[0043] Figure 4 A third schematic flowchart illustrating a power supply voltage tolerance detection method for QKD equipment provided in this application embodiment;

[0044] Figure 5 A fourth schematic flowchart of a power supply voltage tolerance detection method for a QKD device provided for the application embodiment;

[0045] Figure 6 Fifth flowchart illustrating a power supply voltage tolerance detection method for a QKD device provided in the application embodiment;

[0046] Figure 7 A schematic flowchart, number six, of a power supply voltage tolerance detection method for a QKD device provided in the application embodiment;

[0047] Figure 8 A functional module diagram of a power supply voltage tolerance detection device for QKD equipment provided in this application embodiment;

[0048] Figure 9 This is a schematic diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0050] Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of the application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0051] In the description of this application, it should be noted that if the terms "upper", "lower", etc. appear to indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship that the product of this application is usually placed in, it is only for the convenience of describing this application and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0052] Furthermore, the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Additionally, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0053] It should be noted that, where there is no conflict, the features in the embodiments of this application can be combined with each other.

[0054] Figure 1 A schematic diagram of a power supply voltage tolerance detection system for a QKD device provided in this application embodiment is shown below. Figure 1 As shown, the hardware layer system includes: an adaptive test executor, a switch, a programmable power supply, a data acquisition module, and a quantum key distribution (QKD) device cluster. The QKD device cluster includes: a quantum preparation device, a QKD device under test, a quantum detection device, and a QKD device under test.

[0055] The system establishes communication connections via a switch using the TCP / IP protocol for unified management of multiple devices. The adaptive test executor is configured with an IP address, the switch with a gateway address, and the programmable power supply and data acquisition module are also configured with IP addresses, forming a TCP / IP star topology network. The adaptive test executor can then remotely control the programmable power supply via SCPI commands, while the data acquisition module transmits acquired data to the adaptive test executor via TCP. The adaptive test executor then pushes the received data to the database in real time.

[0056] For example, a TCP / IP communication network is built through a switch, and the IP addresses of the following devices are configured: quantum preparation device (192.168.1.98), QKD device under test (192.168.1.99), QKD device under test (network segment 192.168.1.110~1.120), quantum detection device (192.168.1.100); programmable power supply (192.168.1.10), adaptive test executor (192.168.1.200); data acquisition module (192.168.1.11), realizing the physical connection of communication links between devices and network protocol initialization.

[0057] The programmable power supply can be configured with the following rated output parameters via the SCPI commands VOLT:RANGE 0-220 and CURR:LIMIT 5: voltage range 0~220V (accuracy 0.1V) and current limit ≤5A; the initial output voltage is set to the device's nominal value of 220V, and the programmable power supply output is started to power the QKD device under test.

[0058] Initialize the operating parameters of the quantum preparation equipment, the accompanying QKD equipment, the QKD equipment under test, and the quantum detection equipment. Specifically, this includes configuring parameters such as photon emission frequency and wavelength for the quantum preparation equipment; setting a synchronization clock and communication protocol for the accompanying QKD equipment and establishing a key distribution link with the QKD equipment under test; loading test firmware and enabling the performance monitoring module for the QKD equipment under test; calibrating the sensitivity of the single-photon detector for the quantum detection equipment, and initializing the data acquisition interface.

[0059] The data acquisition module includes multiple sensors that, based on preset sample size and acquisition frequency, accurately acquire the input voltage of the QKD device under test (DUT) in real time. For example, a voltage sensor can acquire the input voltage of the DUT at 10ms intervals with an accuracy of 0.1V. It can also acquire the operating current of the DUT, for example, a Hall sensor can acquire the operating current of the DUT at 10ms intervals with an accuracy of 0.01A. A temperature sensor can also acquire the ambient temperature of the DUT. A sliding window filter is used to denoise the acquired input voltage and operating current, removing spike noise introduced by electromagnetic interference, generating smooth input voltage and operating current time series, etc., which are then uploaded to the adaptive test executor in real time via TCP protocol. Similarly, the data acquisition module can also acquire the input voltage and operating current of the accompanying QKD device.

[0060] In a QKD device cluster, the quantum preparation device, the companion QKD device, the quantum detection device, and the QKD device under test work together to generate system performance indicators such as the average key generation rate and quantum channel error rate of the companion QKD device and the QKD device under test. The quantum detection device then reads the performance parameters of the QKD device under test and the companion QKD device, namely the average key generation rate and quantum channel error rate, through a network protocol and uploads them to the adaptive test executor. Each of the performance parameters is accompanied by a timestamp and device number to ensure the differentiation of the performance data of the QKD device under test and the companion QKD device.

[0061] The received data is transmitted from the adaptive test executor to the real-time database in the data layer. The adaptive test executor receives data including hardware operating parameters such as input voltage and operating current, and QKD device performance data such as average key generation rate and quantum channel bit error rate. The real-time database, acting as a data hub, provides data support to the algorithm layer by extracting data for subsequent training and processing of the multi-output Gaussian process model. It also receives feedback from the algorithm layer regarding dynamically adjusted thresholds after processing by the multi-output Gaussian process model. Furthermore, it provides the basic data source for application layer functions such as data visualization and device control, enabling real-time data interaction and management.

[0062] The power supply voltage tolerance detection method for QKD devices provided in this application will be explained in detail below with reference to the accompanying drawings and specific examples. The power supply voltage tolerance detection method for QKD devices provided in this application can be implemented by an electronic device pre-installed with a preset multi-output Gaussian process model, through the execution of an algorithm. The electronic device can be, for example, a server or a terminal, and the terminal can be a user computer. Figure 2 This is one of the flowcharts illustrating a power supply voltage tolerance detection method for a QKD device provided in an embodiment of this application. Figure 2 As shown, the method includes:

[0063] S101. Control the programmable power supply to provide power voltage to the QKD device under test, and collect multiple input voltages of the QKD device under test.

[0064] In this embodiment, a QKD device power supply voltage tolerance detection task is created through the electronic device's task management system. The detection voltage range, step size, acquisition frequency, etc. are configured, and the SCPI command is input to start the detection task. After the detection task is started, the task management system displays the task progress in real time. Users can pause, resume, or export the task as needed.

[0065] After receiving the SCPI command, the programmable power supply in the QKD device power supply voltage tolerance detection system outputs the corresponding voltage value to provide power supply voltage to the QKD device under test. The voltage sensor in the data acquisition module collects multiple input voltages of the QKD device under test in real time according to the set sample size and acquisition frequency, and uploads them to the adaptive test executor. The adaptive test executor stores the received multiple input voltages in the real-time database, so that the electronic equipment in the algorithm layer can obtain multiple input voltages of the QKD device under test from the real-time database.

[0066] S102. Construct multiple sets of input variables based on multiple input voltages.

[0067] Each set of input variables includes at least one input voltage.

[0068] Specifically, noise reduction is performed on multiple input voltages by applying moving average filtering to further suppress random noise and improve data smoothness. Then, normalization is performed to map the multiple input voltages to the [0, 1] interval, as shown in the formula:

[0069]

[0070] in, , For the global extrema of multiple input voltages (e.g., voltage 0~220V), the dimensional differences are eliminated to adapt to the training of the multi-output Gaussian process model, and multiple sets of input variables are constructed based on the normalized multiple input voltages.

[0071] S103. Using a preset multi-output Gaussian process model, perform multi-output prediction on multiple sets of input variables to obtain multiple sets of output variables.

[0072] Each set of output variables includes: performance prediction parameters in multiple dimensions corresponding to an input voltage.

[0073] Specifically, multiple sets of input variables are input into a preset multi-output Gaussian process model, which then performs multi-output prediction on the multiple sets of input variables to obtain multiple sets of output variables.

[0074] Optionally, the performance prediction parameters across multiple dimensions include: multiple basic performance prediction parameters and extended performance prediction parameters;

[0075] Several basic performance prediction parameters include: average key generation rate prediction parameter and quantum channel error rate prediction parameter. Extended performance prediction parameters include: system line loss margin prediction parameter.

[0076] S104. Based on multiple input voltages and corresponding performance prediction parameters of multiple dimensions, perform power supply voltage tolerance analysis on the QKD device under test to obtain the power supply voltage tolerance range of the QKD device under test.

[0077] Specifically, based on multiple input voltages, corresponding performance prediction parameters in multiple dimensions, and performance thresholds in multiple dimensions, a power supply voltage tolerance analysis is performed on the QKD device under test to obtain the power supply voltage tolerance range of the QKD device under test.

[0078] In summary, this application provides a power supply voltage tolerance detection method for QKD devices, comprising: controlling a programmable power supply to provide power supply voltage to the QKD device under test, and collecting multiple input voltages of the QKD device under test; constructing multiple sets of input variables based on the multiple input voltages, wherein each set of input variables includes at least one input voltage; using a preset multi-output Gaussian process model to perform multi-output prediction on the multiple sets of input variables to obtain multiple sets of output variables, each set of output variables including: multiple dimensions of performance prediction parameters corresponding to one input voltage; and performing power supply voltage tolerance analysis on the QKD device under test based on the multiple input voltages and the corresponding multiple dimensions of performance prediction parameters to obtain the power supply voltage tolerance range of the QKD device under test.

[0079] The method in this application collects the input voltage of the QKD device, constructs multiple sets of input variables, and uses a multi-output Gaussian process model to predict multi-dimensional performance parameters. This method can accurately locate the safe operating range of the device voltage, avoid misjudgment of a single parameter, significantly improve the accuracy of anomaly detection, and proactively avoid risks such as increased bit error rate and decreased key generation rate caused by voltage anomalies. By comprehensively analyzing multi-dimensional performance to capture parameter correlations, it provides a quantitative basis for power supply design optimization, efficiently and comprehensively ensuring the stable operation and reliable performance of the QKD device under test under voltage fluctuations. Furthermore, it automates the entire process of power supply characteristic testing of the QKD device under test, improving testing efficiency. It is suitable for multi-device collaborative testing and batch device testing, reducing the workload of personnel.

[0080] This application also provides another possible implementation of the power supply voltage tolerance detection method for QKD devices. Figure 3 This is a second schematic flowchart illustrating a power supply voltage tolerance detection method for a QKD device, as provided in an embodiment of this application. Figure 3 As shown, the method also includes:

[0081] S201. Collect multiple ambient temperatures of the QKD device under test.

[0082] Based on the above, multiple sets of input variables are constructed according to multiple input voltages, including:

[0083] S202. Construct multiple sets of input variables based on multiple input voltages and multiple ambient temperatures.

[0084] Each set of input variables also includes: the ambient temperature corresponding to an input voltage.

[0085] In this embodiment, the temperature sensor in the data acquisition module collects multiple ambient temperatures of the QKD device under test in real time according to the set sample size and acquisition frequency, and uploads them to the adaptive test executor. The adaptive test executor also stores the received multiple ambient temperatures in the real-time database, so that the electronic device of the algorithm layer can obtain multiple ambient temperatures of the QKD device under test from the real-time database.

[0086] Based on multiple input voltages and corresponding ambient temperatures, multiple sets of input variables are constructed. This can be understood as inputting multiple input voltages and corresponding ambient temperatures into a preset multi-output Gaussian process model, which then performs multi-output predictions on the multiple input voltages and corresponding ambient temperatures to obtain multiple sets of output variables.

[0087] The method provided in this application extends the ambient temperature variable to the original input voltage, constructing multiple sets of input variables including the input voltage and the corresponding ambient temperature. By using a multi-output Gaussian process model to simultaneously predict multi-dimensional performance parameters, it can accurately analyze the performance change law of QKD device under the coupling effect of voltage and temperature, and improve the accuracy of predicting the power supply voltage tolerance range of the QKD device under test.

[0088] This application also provides another possible implementation of the power supply voltage tolerance detection method for QKD devices. Figure 4 This is the third flowchart illustrating a power supply voltage tolerance detection method for QKD equipment provided in this application embodiment. Figure 4 As shown, based on multiple input voltages and corresponding performance prediction parameters across multiple dimensions, a power supply voltage tolerance analysis is performed on the QKD device under test to obtain its power supply voltage tolerance range, including:

[0089] S301. Based on multiple input voltages and corresponding performance prediction parameters of multiple dimensions, determine the voltage range from the multiple input voltages where multiple dimensions meet the corresponding performance thresholds.

[0090] S302. Determine the power supply voltage tolerance range based on the voltage range that meets the corresponding performance threshold.

[0091] In this embodiment, if the performance prediction parameters of multiple dimensions include: average key generation rate prediction parameter and quantum channel error rate prediction parameter, the performance threshold corresponding to the average key generation rate prediction parameter is 90%, and the performance threshold corresponding to the quantum channel error rate prediction parameter is 3%.

[0092] Then, multiple input voltages are filtered, and only the voltage range formed by multiple target input voltages with an average key generation rate prediction parameter greater than or equal to 90% and a quantum channel bit error rate prediction parameter less than or equal to 3% is determined as the power supply voltage tolerance range, for example (195.0V-205.5V).

[0093] The method provided in this application selects voltage ranges that meet the threshold values ​​in each dimension from multiple input voltages and corresponding multi-dimensional performance prediction parameters, and then determines the power supply voltage tolerance range accordingly. This can accurately eliminate invalid voltage points caused by substandard performance in a single dimension, avoid the one-sidedness of judging the tolerance range based on only some indicators, and ensure that the finally determined voltage range can simultaneously meet the multi-dimensional performance requirements such as average key generation rate and quantum channel bit error rate. This effectively improves the reliability and practicality of the tolerance range, provides a more accurate safety boundary reference for QKD devices to resist voltage fluctuations in practical applications, and reduces the risk of system performance degradation caused by voltage exceeding the true tolerance range.

[0094] This application also provides another possible implementation of the power supply voltage tolerance detection method for QKD devices. Figure 5 This is the fourth flowchart illustrating a power supply voltage tolerance detection method for a QKD device provided in the application embodiment. Figure 5 As shown, the voltage range corresponding to the performance threshold is used to determine the power supply voltage tolerance range, including:

[0095] S401. Based on multiple input voltages and corresponding performance prediction parameters of multiple dimensions, determine the abnormal voltage range that does not meet the corresponding performance threshold from the multiple input voltages.

[0096] S402. Correct the voltage range that meets the corresponding performance threshold according to the abnormal voltage range, and determine the corrected voltage range as the power supply voltage tolerance range.

[0097] In this embodiment, the performance prediction parameters of multiple dimensions corresponding to multiple input voltages are evaluated. If the average key generation rate prediction parameter corresponding to an input voltage is less than 90%, or the quantum channel bit error rate prediction parameter is greater than 3%, then the input voltage is determined to be an abnormal voltage point, an alarm is triggered, and the abnormal voltage range is determined based on the abnormal voltage point.

[0098] Then, based on the abnormal voltage range, the voltage range that meets the corresponding performance threshold is corrected, and the corrected voltage range is the power supply voltage tolerance range.

[0099] The method provided in this application first identifies abnormal voltage ranges that do not meet the threshold among multiple input voltages and corresponding multi-dimensional performance parameters, and then corrects the voltage ranges that meet the performance requirements based on this. This can accurately eliminate hidden voltage anomaly areas caused by environmental interference, equipment characteristic fluctuations, and other factors, and avoid including critical failure voltages in the tolerance range. By correcting abnormal ranges, the final determined power supply voltage tolerance range can better match the actual working characteristics of the equipment, effectively improving the robustness of tolerance analysis. This provides a more reliable voltage safety boundary for the stable operation of QKD equipment under complex working conditions and reduces the potential risk of performance failure triggered by voltage fluctuations in practical applications.

[0100] This application also provides another possible implementation of the power supply voltage tolerance detection method for QKD devices. Based on multiple input voltages and corresponding performance prediction parameters of multiple dimensions, multiple performance curves of the QKD device under test are generated, and abnormal voltage points are marked in each performance curve. Each performance curve includes: an average key generation rate prediction curve and a quantum channel bit error rate prediction curve.

[0101] Optionally, each set of output variables also includes: operating current prediction parameters; and generating a relationship curve between input voltage and operating current based on multiple input voltage and operating current prediction parameters.

[0102] In this embodiment, the electronic device's display interface combines real-time database data to present a multi-dimensional dashboard that displays the input voltage waveform, the relationship curve between input voltage and operating current, the average key generation rate prediction curve, and the quantum channel bit error rate prediction curve of the QKD device under test in real time. It also marks the normal operating range and outliers and displays the power supply voltage tolerance range.

[0103] This application also provides another possible implementation of the power supply voltage tolerance detection method for QKD devices. Figure 6 This is the fifth flowchart illustrating a power supply voltage tolerance detection method for a QKD device provided in the application embodiment. Figure 6 As shown, before obtaining multiple sets of output variables by using a pre-defined multi-output Gaussian process model to perform multi-output prediction on multiple sets of input variables, the method further includes:

[0104] S501. Obtain multiple sets of historical operating parameters for the QKD device under test and the accompanying QKD device.

[0105] Each set of historical operating parameters includes: a historical input voltage, corresponding performance detection parameters in multiple dimensions, and corresponding historical operating current.

[0106] In this embodiment, the historical input voltage and historical operating current of the QKD device under test and the accompanying QKD device are collected by the data acquisition module. The performance detection parameters of the QKD device under test and the accompanying QKD device are read by the quantum detection device and uploaded to the adaptive test executor. The adaptive test executor then transmits the data to the real-time database. The performance detection parameters include the average key generation rate and the quantum channel bit error rate.

[0107] S502. Using a preset initial multi-output Gaussian process model, multi-output prediction is performed on the historical input voltage of the QKD device under test to obtain the performance prediction parameters and current prediction parameters of the QKD device under test under the corresponding historical input voltage.

[0108] S503. Using a preset initial multi-output Gaussian process model, multi-output prediction is performed on the historical input voltage of the QKD device under test to obtain the performance prediction parameters and current prediction parameters of the QKD device under test under the corresponding historical input voltage.

[0109] S504. Based on the performance prediction parameters, current prediction parameters, performance detection parameters, and historical operating current of the QKD device under test at the corresponding historical input voltage, and the performance prediction parameters, current prediction parameters, performance detection parameters, and historical operating current of the accompanying QKD device at the corresponding historical input voltage, iteratively train the preset initial multi-output Gaussian process model to obtain the preset multi-output Gaussian process model.

[0110] Specifically, firstly, the initial input data... and initial output data Implementing standardization:

[0111]

[0112]

[0113] in, The input dataset is the initial multi-output Gaussian process model, where N is the number of samples and 1 is the input dimension (containing only input voltage). Each element represents the input voltage measurement of one sample. The indicator represents the output dataset of the initial multi-output Gaussian process model, where N is the number of samples and 3 is the output dimension, corresponding to the performance prediction parameters and current prediction parameters, respectively. The performance prediction parameters include the average key generation rate prediction parameter and the quantum channel bit error rate prediction parameter. Each element represents the corresponding output measurement value of a sample. , These are the mean and standard deviation of the input data, respectively. , These are the mean and standard deviation of the output data, respectively. The input data consists of historical input voltages, and the output data includes performance prediction parameters and current prediction parameters.

[0114] The core of a multi-output Gaussian process is to construct a joint Gaussian process among multiple outputs, assuming an initial 3D output. Their joint distribution follows a multivariate Gaussian distribution:

[0115]

[0116] in, The indicator represents the joint probability distribution of the outputs of the initial multi-output Gaussian process model, which falls under the category of conditional probability. Its core meaning is "given the input dataset of the initial multi-output Gaussian process model..." Model parameter set Under the premise of observing the output dataset The probability. This indicates the set of parameters for the initial model, including the latent variable weight matrix, kernel function parameters, noise covariance matrix parameters, etc. The indicator is the mean. The covariance matrix is The multivariate normal distribution is the output of the initial multi-output Gaussian process model. The type of distribution it follows. The covariance matrix of the initial multi-output Gaussian process model describes the correlation between output variables and is composed of the kernel function of the latent variables, the weight matrix, and the noise covariance matrix.

[0117] Wherein, the covariance matrix Defined by the kernel function:

[0118]

[0119] in, This indicates the number of latent variables, i.e., the number of latent variables shared by the model's output variables. The indicator is the weight vector of the r-th latent variable to the three outputs. One latent variable corresponds to one weight value for each of the three outputs, which describes the degree of influence of the latent variable on different outputs. The weight vector of the r-th latent variable The transpose of , with a dimension of 1×3, is used in conjunction with the kernel function matrix to construct related terms of the output covariance matrix.

[0120] For the squared exponent kernel: , indicating the r-th latent variable in the input dataset The covariance matrix is ​​N×N, and the element (i, j) represents the covariance between the i-th sample and the j-th sample on the r-th latent variable.

[0121] The noise covariance matrix of the initial multi-output Gaussian process model characterizes the correlation of the three output observation noises. If it is a diagonal matrix, it means that the output noises are independent of each other.

[0122] Based on the initial multi-output Gaussian process model, multi-output prediction is performed on the historical input voltage of the QKD device under test to obtain the performance prediction parameters and current prediction parameters of the QKD device under test under the corresponding historical input voltage. These parameters are then compared with the performance detection parameters in multiple dimensions and the corresponding historical operating current to determine whether parameter expansion is required. If the error between the performance prediction parameters and the performance detection parameters and the error between the current prediction parameters and the historical operating current are greater than the error range, then parameter expansion is required.

[0123] At this point, for extended input data Extended output data Implement standardization:

[0124]

[0125]

[0126] in, Indicates the input dataset for the extended model. 2 represents the number of samples, and 2 represents the input dimension (including input voltage and ambient temperature). Each element represents the voltage or temperature measurement value of one sample. Indicates the output dataset for the extended model. 4 represents the number of samples, and 4 represents the output dimension (corresponding to the current prediction parameter and the performance prediction parameter, respectively. The performance prediction parameter includes: the average key generation rate prediction parameter, the quantum channel bit error rate prediction parameter, and the system line loss margin). Each element represents the corresponding output measurement value of a sample.

[0127] After adding a temperature variable, the kernel function is updated as follows:

[0128]

[0129] in, The pointer indicates the input kernel function of the extended model, used to characterize the two input samples. and The correlation is formed by the superposition of the voltage kernel function and the temperature kernel function. For the i-th input sample of the expanded model, with a dimension of 1×2, it contains the voltage and temperature values ​​of that sample. The j-th input sample of the extended model has a dimension of 1×2 and contains the voltage and temperature values ​​of that sample.

[0130] The kernel function is indicated by the voltage variable, and the input is the input voltage value of the i-th sample. The input voltage value of the j-th sample It is inherited from the initial model and describes the spatial correlation of voltage variables. The kernel function is indicated by the temperature variable, and the input is the ambient temperature value of the i-th sample. Compared with the ambient temperature value of the j-th sample To characterize the spatial correlation of temperature variables, its parameters need to be initialized.

[0131] The weight matrix is ​​expanded as follows:

[0132]

[0133] in, The indicator is the weight matrix of the latent variables in the extended model, 4 is the output dimension (including the newly added system line loss margin), and R is the number of latent variables, which describes the degree of influence of each latent variable on the four outputs. The indicator is the latent variable weight matrix of the initial model (inherited from the initial model), 3 is the initial output dimension, and R is the number of latent variables, which describes the degree of influence of each latent variable on the initial 3 outputs. The indicator is the weight vector of the system line loss margin (new output) on each latent variable. R is the number of latent variables, and 1 is the dimension of the weight vector. It describes the degree of influence of each latent variable on the system line loss margin. Its parameters need to be initialized. The system line loss margin weight vector The transpose of , with dimension 1×R, is used to... The weight matrix of the extended model is formed by concatenating the components. .

[0134] The noise covariance matrix is ​​updated as follows:

[0135]

[0136] in, The indicator is the noise covariance matrix (diagonal matrix) of the extended model, which characterizes the variance of the observed noise of the four outputs (including the newly added system line loss margin), with the diagonal elements corresponding to the noise variance of each output. The noise variance of the operating current output (inherited from the initial model) characterizes the degree of deviation between the measured operating current value and the true value. The noise variance of the bitrate output (inherited from the initial model) characterizes the degree of deviation between the bitrate measurement and the true value. The noise variance of the bit error rate output (inherited from the initial model) characterizes the degree of deviation between the measured bit error rate value and the true value. The noise variance (new addition) for the system line loss margin output characterizes the degree of deviation between the measured value and the true value of the system line loss margin. Its parameters need to be initialized.

[0137] Hyperparameter θ optimization for multi-output Gaussian processes is often performed using the EM algorithm, which consists of an E-step (calculating the posterior distribution of latent variables) and an M-step (updating parameters). The E-step calculates the posterior distribution of latent variables given the observed data. Latent variables The posterior distribution is:

[0138]

[0139] Specifically:

[0140]

[0141]

[0142]

[0143] in, The indicator is the set of latent variables in the E-step (latent variable inference phase) of incremental learning for the extended model. The posterior probability distribution, belonging to the category of conditional probability, has the core meaning of "given the observed output dataset of the extended model". Current model parameter set Under the premise of, the set of latent variables The probability of taking a specific value. The indicator is the mean. The covariance matrix is The multivariate normal distribution is a type of posterior distribution of the latent variable f. Let f be the posterior mean of the latent variable f, characterizing the observed output. Under the given conditions, the expected value of the latent variable. Let f be the posterior covariance matrix of the latent variable f, characterizing the observed output. Under the given conditions, the correlation between the values ​​of the latent variables.

[0144] Let f be the prior covariance matrix (block diagonal matrix) of the latent variable f, with dimension 1. Each block represents the r-th latent variable in covariance matrix on It characterizes the prior correlation between latent variables (the prior independence between different latent variables). The structure is a block-diagonal matrix, where the diagonal blocks are the covariance matrices of the 1st to Rth latent variables, and the off-diagonal blocks are zero matrices, indicating prior independence between different latent variables.

[0145] To extend the model, output the covariance matrix. The inverse matrix is ​​used to calculate the weighting coefficients for the posterior mean and posterior covariance. Indicator is the posterior mean of a latent variable. The calculation formula is output through observation. The result is obtained by correcting the prior mean (0) of the latent variable. Indicator is the posterior covariance matrix of latent variables The calculation formula is output through observation. Prior covariance of latent variables The result was obtained after correction.

[0146] M-step: Update parameters by maximizing the lower bound of the log-marginal likelihood:

[0147]

[0148]

[0149] in, Represented as the logarithmic marginal likelihood function, it indicates that under the model parameters Under the conditions, the observed output The logarithmic probability of the model can be maximized to achieve the optimal estimate of the model parameters. The core objective function for the M-step incremental learning (parameter update phase) of the extended model is the log-marginal likelihood function. This falls under the category of conditional logarithmic probability, and its core meaning is that given the input dataset of the extended model... ), current model parameter set Under the premise of observation, the extended output dataset was observed. The logarithmic probability. Let M be the set of model parameters to be updated in M ​​steps, including the latent variable weight matrix. The noise variance of the four outputs and the parameters of the first to Rth kernel functions. Represented as the latent variable weight matrix of the extended model (same as above) The dimension is 4×R, which characterizes the degree of influence of each latent variable on the four outputs. The parameters of the r-th kernel function (such as the length scale of the RBF kernel and the signal variance) determine how and how strongly the kernel function characterizes the correlation of the input variables.

[0150] Weight matrix update:

[0151]

[0152] in, Represented as the log-marginal likelihood function with respect to the weight vector of the r-th latent variable. The partial derivatives are used to update the gradient ascent method. Maximize it . Represented as the r-th latent variable in the i-th input sample The value at the location With the output of the i-th sample The covariance characterizes the linear correlation between latent variables and the output. Represented as the r-th latent variable in the i-th input sample The value at that location, It is represented as the output vector (dimension 1×4) of the i-th sample, which contains the measured values ​​of operating current, code generation rate, bit error rate, and system line loss margin. This is represented as the r-th kernel function on the i-th input sample. The autocovariance at a given point, i.e., the covariance of the sample itself on that latent variable, is a scalar.

[0153] Nuclear parameter update:

[0154]

[0155] in, Represented as the log-marginal likelihood function with respect to the r-th kernel function parameter The partial derivatives are used to update the gradient ascent method. Maximize it . Represented as the trace operation of a matrix, it is the summation of the elements on the main diagonal of the matrix, used to convert the result of a matrix operation into a scalar (the logarithmic marginal likelihood is a scalar). This represents the output covariance matrix of the extended model. For the r-th kernel function parameter The partial derivatives of the kernel function are used to characterize the effect of changes in kernel function parameters on the output covariance matrix. This represents the output dataset of the extended model. The product of its transpose and its transpose, with a dimension of 4N×4N, characterizes the correlation between output samples.

[0156] Noise variance update:

[0157]

[0158] in, Represented as system line loss margin and noise variance The updated formula, calculated using sample observations, posterior mean of latent variables, and covariance, ensures that the noise variance estimate is consistent with the observed data and the distribution of latent variables. It is represented as the observed value of the system line loss margin for the i-th sample. This is represented as the weight of the system line loss margin with respect to the r-th latent variable (i.e., (the r-th element). Represented as the r-th latent variable in the i-th input sample The posterior mean at (i.e.) (The element corresponding to the r-th latent variable and the i-th sample). This is represented as the r-th latent variable and the s-th latent variable in the i-th input sample. The posterior covariance of the values ​​taken at each point (i.e. The element at the corresponding position in the middle.

[0159] Calculate the parameter change between two iterations , The iteration stops when the threshold is set, or when the ELBO increment is less than the threshold.

[0160] After completing the EM iteration, for the new input Predict multiple outputs Distribution:

[0161]

[0162] in;

[0163]

[0164]

[0165] This is represented as a new input point, with a dimension of 2×1, containing the AC voltage value to be predicted. and ambient temperature value . This represents the core output of the extended model's prediction phase, the joint prediction probability distribution of multiple outputs from new input points, falling under the category of conditional probability. Its core meaning is that given the training dataset of the extended model (input... Output ) and new input points to be predicted Given the new input point, the output vector The probability of taking a specific value. Indicates a new input point The corresponding prediction output vector has a dimension of 1×4 and includes predicted values ​​for operating current, code generation rate, bit error rate, and system line loss margin. Expressed as mean The covariance matrix is The multivariate normal distribution is used to predict output. The type of distribution it follows. Represented as predicted output The mean of the kernel function, i.e. the optimal predicted value of each output, is calculated using the training data and the kernel function value of the new input. Represented as predicted output The covariance matrix is ​​used to characterize the correlation and prediction uncertainty among the predicted outputs.

[0166] Represented as the r-th kernel function on new input With training input The covariance vector between them has a dimension of 1×N, and the element (1,i) represents... The covariance of the i-th training sample with respect to the r-th latent variable. Represented as the r-th kernel function on new input The autocovariance at the r-th latent variable is a scalar that characterizes the prior variance of the new input itself. This is represented as the r-th kernel function in the training input. With new input The covariance vector between them, with dimension N×1, is The transpose of .

[0167] Based on the predicted distribution, calculate the confidence interval of the system line loss margin L:

[0168]

[0169] in, The indicator is the predicted distribution of the system line loss margin L, which follows a mean of 1 / 2. variance is The normal distribution of L is used to characterize the predicted value and uncertainty of L. This is expressed as the predicted mean of the system line loss margin L (i.e. (The element corresponding to the line loss margin). This is expressed as the prediction variance of the system line loss margin L (i.e. (The diagonal element corresponding to the line loss margin).

[0170] Similarly, the confidence interval of the test QKD device is calculated using a preset initial multi-output Gaussian process model. By comparing the confidence interval of the test QKD device with that of the QKD device to be tested, it is determined whether the model training has stopped. If the two confidence intervals are consistent, then the model trained at this time is determined to be the preset multi-output Gaussian process model.

[0171] It should be noted that the power supply voltage tolerance range of the QKD device under test can be calculated using a preset multi-output Gaussian process model, expressed as:

[0172]

[0173] in, This is the voltage tolerance set, which is the range of input voltage values ​​when the ambient temperature is fixed. It requires that the predicted values ​​of all output variables (operating current, code generation rate, bit error rate, system line loss margin) within this range meet the performance requirements within a 95% confidence interval.

[0174] For current prediction parameters, This is the minimum allowable value for the operating current. This is the maximum allowable value for the operating current. The average key generation rate prediction parameter. This is the minimum allowed value for average key generation rate. For the prediction parameters of the quantum channel bit error rate, This represents the maximum allowable value for the quantum channel bit error rate prediction parameter. This is the system's line loss margin. This represents the minimum allowable bit rate value within the system's line loss margin. To establish a confidence level, meaning the probability that the predicted output meets performance requirements is no less than 95%, the confidence interval needs to be calculated based on the variance of the predicted distribution. This involves iterating through the voltage input. Filter those that meet the threshold The range refers to the power supply voltage tolerance range.

[0175] This application also provides another possible implementation of the power supply voltage tolerance detection method for QKD devices. Figure 7 This is a sixth flowchart illustrating a power supply voltage tolerance detection method for a QKD device, provided as an example of the application. Figure 7 As shown, multiple sets of historical operating parameters of the QKD device under test and the accompanying QKD device are obtained, including:

[0176] S601. Obtain multiple sets of initial historical operating parameters for the QKD device under test and the accompanying QKD device.

[0177] Each set of initial historical operating parameters includes: an initial historical input voltage, corresponding initial performance detection parameters in multiple dimensions, and corresponding initial historical operating current.

[0178] S602. Scan multiple sets of initial historical working parameters according to the first preset step size to determine the abnormal voltage range.

[0179] S603. Scan multiple sets of initial historical operating parameters within the abnormal voltage range according to the second preset step size, and determine multiple sets of historical operating parameters.

[0180] In this embodiment, the first preset step size can be set to 5V. The programmable power supply is configured to scan within the nominal voltage range (e.g., 180~240V) in 5V steps. Data is collected after each voltage point stabilizes for 5 seconds. Based on the preprocessed data, the code rate drop point (e.g., code rate <80% below 200V) and the bit error rate jump point (e.g., bit error rate >5% above 230V) are initially fitted and located through linear regression. The abnormal voltage range (e.g., 190~210V) is determined as the initial range for fine scanning.

[0181] The second preset step size can be set to 0.1V. Within the range determined by coarse scanning (such as 190~210V), the step size can be dynamically adjusted for encrypted sampling through the SCPI command VOLT:STEP 0.1. Multiple sets of data are collected for each voltage point, the mean and variance are calculated, and the critical values ​​of the voltage-performance relationship are refined (such as the upper and lower voltage limits corresponding to a code generation rate of 90%), thereby obtaining multiple sets of historical working parameters.

[0182] The method provided in this application first acquires multiple sets of initial historical operating parameters of the QKD device under test and the accompanying device, including initial historical input voltage, multi-dimensional initial performance detection parameters, and initial historical operating current. Then, it scans the device with a first preset step size to initially locate abnormal voltage ranges. Finally, it uses a smaller second preset step size to finely scan the parameters within the abnormal range to determine multiple sets of historical operating parameters. This method can cover more comprehensive operating conditions by utilizing the complete historical operating data of the two devices, quickly narrow the scope of abnormal voltage investigation and improve analysis efficiency by using a larger step size. It can also capture local performance fluctuations that are easily missed by coarse scanning (such as performance degradation accompanied by small current changes within a voltage range) by using fine scanning with a small step size within the abnormal range. This allows the determined historical operating parameters to more accurately reflect the true characteristics of the device in the abnormal voltage region, providing more reliable data support for subsequent power supply voltage tolerance analysis and effectively reducing the risk of misjudging abnormal ranges due to insufficient parameter scanning accuracy, which in turn affects the accuracy of the tolerance range.

[0183] The following will continue to explain the power supply voltage tolerance detection device and electronic device for QKD equipment provided in any of the above embodiments of this application. The specific implementation process and the resulting technical effects are the same as those in the corresponding method embodiments. For the sake of brevity, the parts not mentioned in this embodiment can be referred to the corresponding content in the method embodiment.

[0184] Figure 8 This is a functional module diagram of a QKD equipment power supply voltage tolerance detection device provided in an embodiment of this application. Figure 8 As shown, the power supply voltage tolerance detection device 100 for the QKD equipment includes:

[0185] The acquisition module 110 is used to control the programmable power supply to provide power voltage to the QKD device under test and to acquire multiple input voltages of the QKD device under test.

[0186] The construction module 120 is used to construct multiple sets of input variables based on multiple input voltages, wherein each set of input variables includes at least one input voltage;

[0187] Prediction module 130 is used to perform multi-output prediction on multiple sets of input variables using a preset multi-output Gaussian process model to obtain multiple sets of output variables. Each set of output variables includes: multiple dimensions of performance prediction parameters corresponding to an input voltage.

[0188] Analysis module 140 is used to perform power supply voltage tolerance analysis on the QKD device under test based on multiple input voltages and corresponding performance prediction parameters of multiple dimensions, so as to obtain the power supply voltage tolerance range of the QKD device under test.

[0189] Optionally, the acquisition module 110 is also used to acquire multiple ambient temperatures of the QKD device under test;

[0190] The construction module 120 is also used to construct multiple sets of input variables based on multiple input voltages and multiple ambient temperatures. Each set of input variables also includes: the ambient temperature corresponding to an input voltage.

[0191] Optionally, the performance prediction parameters in multiple dimensions include: multiple basic performance prediction parameters and extended performance prediction parameters; the multiple basic performance prediction parameters include: average key generation rate prediction parameters and quantum channel bit error rate prediction parameters, and the extended performance prediction parameters include: system line loss margin prediction parameters.

[0192] Optionally, the analysis module 140 is further configured to determine, based on multiple input voltages and corresponding performance prediction parameters of multiple dimensions, a voltage range in which multiple dimensions meet the corresponding performance thresholds; and to determine the power supply voltage tolerance range based on the voltage range that meets the corresponding performance thresholds.

[0193] Optionally, the analysis module 140 is further configured to determine, based on multiple input voltages and corresponding performance prediction parameters of multiple dimensions, abnormal voltage ranges that do not meet the corresponding performance thresholds from multiple input voltages; and to correct the voltage ranges that meet the corresponding performance thresholds based on the abnormal voltage ranges, thereby determining the corrected voltage range as the power supply voltage tolerance range.

[0194] Optionally, the device further includes:

[0195] The generation module is used to generate multiple performance curves for the QKD device under test based on multiple input voltages and corresponding performance prediction parameters of multiple dimensions, and to mark abnormal voltage points in each performance curve. Each performance curve includes: an average key generation rate prediction curve and a quantum channel bit error rate prediction curve.

[0196] Optionally, each set of output variables also includes: operating current prediction parameters. The generation module is further configured to generate a relationship curve between input voltage and operating current based on multiple input voltages and the operating current prediction parameters.

[0197] Optionally, the device further includes:

[0198] The acquisition module is used to acquire multiple sets of historical operating parameters of the QKD device under test and the accompanying QKD device. Each set of historical operating parameters includes: a historical input voltage, corresponding performance detection parameters in multiple dimensions, and corresponding historical operating current.

[0199] The prediction module 130 is also used to perform multi-output prediction of the historical input voltage of the QKD device under test using a preset initial multi-output Gaussian process model, so as to obtain the performance prediction parameters and current prediction parameters of the QKD device under test under the corresponding historical input voltage.

[0200] The prediction module 130 is also used to perform multi-output prediction on the historical input voltage of the QKD device under test using a preset initial multi-output Gaussian process model, so as to obtain the performance prediction parameters and current prediction parameters of the QKD device under test under the corresponding historical input voltage.

[0201] The training module is used to iteratively train a preset initial multi-output Gaussian process model based on the performance prediction parameters, current prediction parameters, performance detection parameters, and historical operating current of the QKD device under test at the corresponding historical input voltage, as well as the performance prediction parameters, current prediction parameters, performance detection parameters, and historical operating current of the QKD device under companion test at the corresponding historical input voltage, to obtain the preset multi-output Gaussian process model.

[0202] Optionally, the acquisition module is also used to acquire multiple sets of initial historical operating parameters of the QKD device under test and the accompanying QKD device. Each set of initial historical operating parameters includes: an initial historical input voltage, corresponding initial performance detection parameters of multiple dimensions, and corresponding initial historical operating current; scanning multiple sets of initial historical operating parameters according to a first preset step size to determine abnormal voltage ranges; and scanning multiple sets of initial historical operating parameters within the abnormal voltage range according to a second preset step size to determine multiple sets of historical operating parameters.

[0203] The above-described device is used to execute the method provided in the foregoing embodiments, and its implementation principle and technical effect are similar, so they will not be described again here.

[0204] These modules can be one or more integrated circuits configured to implement the above methods, such as one or more Application Specific Integrated Circuits (ASICs), one or more microprocessors, or one or more Field Programmable Gate Arrays (FPGAs). Alternatively, when a module is implemented using processing element scheduler code, the processing element can be a general-purpose processor, such as a Central Processing Unit (CPU) or other processor capable of calling program code. Furthermore, these modules can be integrated together as a system-on-a-chip (SOC).

[0205] Figure 9 This is a schematic diagram of an electronic device provided in an embodiment of this application. This electronic device can be used for power supply voltage tolerance detection in QKD devices. Figure 9 As shown, the electronic device includes: a processor 210, a storage medium 220, and a bus 230.

[0206] Storage medium 220 stores machine-readable instructions executable by processor 210. When the electronic device is running, processor 210 communicates with storage medium 220 via bus 230, and processor 210 executes the machine-readable instructions to perform the steps of the above method embodiment. The specific implementation and technical effects are similar and will not be described again here.

[0207] Optionally, this application also provides a storage medium 220, on which a computer program is stored. When the computer program is run by a processor, it executes the steps of the above-described method embodiments. The specific implementation and technical effects are similar, and will not be repeated here.

[0208] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0209] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0210] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0211] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute some steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0212] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for detecting the power supply voltage tolerance of a QKD (Quick Knockout) device, characterized in that, The method includes: The programmable power supply is controlled to provide power voltage to the QKD device under test, and multiple input voltages of the QKD device under test are acquired. Based on the multiple input voltages, multiple sets of input variables are constructed, wherein each set of input variables includes at least one input voltage; A preset multi-output Gaussian process model is used to perform multi-output prediction on the multiple sets of input variables to obtain multiple sets of output variables. Each set of output variables includes: multiple dimensions of performance prediction parameters corresponding to the input voltage. Based on the multiple input voltages and the corresponding performance prediction parameters of the multiple dimensions, a power supply voltage tolerance analysis is performed on the QKD device under test to obtain the power supply voltage tolerance range of the QKD device under test. Before using a preset multi-output Gaussian process model to perform multi-output prediction on the multiple sets of input variables to obtain multiple sets of output variables, the method further includes: Multiple sets of historical operating parameters are obtained for the QKD device under test and the accompanying QKD device. Each set of historical operating parameters includes: a historical input voltage, the corresponding performance detection parameters of the multiple dimensions, and the corresponding historical operating current. Using a preset initial multi-output Gaussian process model, multi-output prediction is performed on the historical input voltage of the QKD device under test to obtain the performance prediction parameters and current prediction parameters of the QKD device under test under the corresponding historical input voltage. Using the preset initial multi-output Gaussian process model, multi-output prediction is performed on the historical input voltage of the QKD device under test to obtain the performance prediction parameters and current prediction parameters of the QKD device under the corresponding historical input voltage. Based on the performance prediction parameters, current prediction parameters, performance detection parameters, and historical operating current of the QKD device under test at the corresponding historical input voltage, and the performance prediction parameters, current prediction parameters, performance detection parameters, and historical operating current of the accompanying QKD device at the corresponding historical input voltage, the preset initial multi-output Gaussian process model is iteratively trained to obtain the preset multi-output Gaussian process model.

2. The method according to claim 1, characterized in that, The method further includes: Collect multiple ambient temperatures of the QKD device under test; The step of constructing multiple sets of input variables based on the multiple input voltages includes: Based on the multiple input voltages and the multiple ambient temperatures, the multiple sets of input variables are constructed, and each set of input variables further includes: the ambient temperature corresponding to one of the input voltages.

3. The method according to claim 1, characterized in that, The performance prediction parameters across multiple dimensions include: multiple basic performance prediction parameters and extended performance prediction parameters; The multiple basic performance prediction parameters include: average key generation rate prediction parameter and quantum channel error rate prediction parameter; the extended performance prediction parameters include: system line loss margin prediction parameter.

4. The method according to claim 1, characterized in that, The step of performing power supply voltage tolerance analysis on the QKD device under test based on the multiple input voltages and the corresponding performance prediction parameters of the multiple dimensions to obtain the power supply voltage tolerance range of the QKD device under test includes: Based on the multiple input voltages and the corresponding performance prediction parameters of the multiple dimensions, determine the voltage range in which all multiple dimensions satisfy the corresponding performance thresholds from the multiple input voltages; The power supply voltage tolerance range is determined based on the voltage range that meets the corresponding performance threshold.

5. The method according to claim 4, characterized in that, The step of determining the power supply voltage tolerance range based on the voltage range that meets the corresponding performance threshold includes: Based on the multiple input voltages and the corresponding performance prediction parameters of the multiple dimensions, determine the abnormal voltage ranges that do not meet the corresponding performance thresholds from the multiple input voltages; The voltage range that meets the corresponding performance threshold is corrected based on the abnormal voltage range, and the corrected voltage range is determined as the power supply voltage tolerance range.

6. The method according to claim 5, characterized in that, The method further includes: Based on the multiple input voltages and the corresponding performance prediction parameters of the multiple dimensions, multiple performance curves of the QKD device under test are generated, and abnormal voltage points are marked in each performance curve. Each performance curve includes: an average key generation rate prediction curve and a quantum channel bit error rate prediction curve.

7. The method according to claim 1, characterized in that, Each set of output variables further includes: operating current prediction parameters; the method further includes: Based on the multiple input voltages and the predicted operating current parameters, a curve showing the relationship between input voltage and operating current is generated.

8. The method according to claim 1, characterized in that, The acquisition of multiple sets of historical operating parameters of the QKD device under test and the accompanying QKD device includes: Multiple sets of initial historical operating parameters are obtained for the QKD device under test and the accompanying QKD device. Each set of initial historical operating parameters includes: an initial historical input voltage, the corresponding initial performance detection parameters of the multiple dimensions, and the corresponding initial historical operating current. The multiple sets of initial historical operating parameters are scanned according to the first preset step size to determine the abnormal voltage range; The multiple sets of initial historical operating parameters within the abnormal voltage range are scanned according to the second preset step size to determine the multiple sets of historical operating parameters.

9. An electronic device, characterized in that, include: The device includes a processor, a storage medium, and a bus. The storage medium stores program instructions executable by the processor. When the electronic device is running, the processor communicates with the storage medium via the bus, and the processor executes the program instructions to perform the steps of the power supply voltage tolerance detection method for the QKD device as described in any one of claims 1-8.

Images