Short circuit simulation test method and device for high voltage direct current power supply system

Abstract

The present application relates to the technical field of data processing, in particular to a short-circuit simulation test method and device for a high-voltage direct-current power supply system, comprising: performing load removal test on the high-voltage direct-current power supply system to obtain test data; extracting an equivalent stable time period from an input voltage curve; extracting an allowable fluctuation feature and an excitation fluctuation feature; training a model to predict the excitation fluctuation feature during short-circuit; obtaining an excitation fault coefficient based on the difference between the predicted excitation fluctuation feature and the allowable fluctuation feature, and determining a delay fault coefficient based on the average length of the equivalent stable time period; and determining a test result based on the excitation fault coefficient and the delay fault coefficient. The present application can comprehensively evaluate the backfire influence of the electromechanical coupling effect on the stability of the controller power supply in the short-circuit fault, avoid the test deviation caused by the existing technology which only focuses on the delay of the controller itself and ignores the power supply abnormality, and improve the accuracy and reliability of the short-circuit simulation test.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and more specifically to a short-circuit simulation test method and equipment for high-voltage DC power supply systems. Background Technology

[0002] High-voltage direct current (HVDC) power supply systems are crucial auxiliary power sources for aircraft, typically consisting of a coaxial permanent magnet auxiliary exciter, a main exciter, and a controller. When a short circuit occurs at the system load, the controller must quickly cut off the excitation current to prevent the short-circuit current from persisting for too long and thus avoid equipment burnout. In engineering practice, to ensure reliability, short-circuit simulation tests are performed before the system is put into use. Current mainstream testing methods primarily focus on evaluating the controller's own response performance. This involves measuring and analyzing the controller's short-circuit current sensing delay, internal program processing delay, power drive delay, and protection logic triggering delay, using these inherent delays as the core basis for assessing the risk of short-circuit burnout. If the tests show excessively long delays and a high risk, it is determined that the controller needs to be replaced or redesigned.

[0003] However, existing technology suffers from a fundamental blind spot in understanding and a testing loophole. During the dynamic process of a real short-circuit fault, the massive short-circuit current (several times the rated current) instantaneously flowing through the stator windings of the main exciter causes drastic changes in the excitation energy. This change, through coaxial mechanical coupling, significantly interferes with the instantaneous speed of the permanent magnet auxiliary exciter. The permanent magnet auxiliary exciter, responsible for powering the controller, directly causes input voltage instability (severe fluctuations or drops) in the controller. As a precision electronic device, the controller's normal operation depends on a stable input voltage; voltage instability can trigger malfunctions (such as resets, misjudgments, or calculation errors), adding and constituting unpredictable response delays.

[0004] Therefore, existing technologies only test the controller's delay under ideal power supply conditions, completely ignoring the fact that short-circuit faults themselves can, through electromechanical coupling, negatively impact the controller's power supply, leading to test results that deviate significantly from reality. This testing method assesses short-circuit risk in an incomplete and unreliable manner, failing to accurately pinpoint the root cause of risk in a real short circuit. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a short-circuit simulation test method and equipment for high-voltage DC power supply systems.

[0006] The short-circuit simulation test method and equipment for high-voltage DC power supply systems of the present invention adopts the following technical solution: One embodiment of the present invention provides a short-circuit simulation test method for high-voltage DC power supply systems, the method comprising the following steps: A load removal test was performed on the high-voltage DC power supply system, and several test data points were obtained. Each test data point includes the rotational speed, the controller's input voltage curve, the controller's response time, and the load removed. Extract the equivalent steady-state time period of the voltage from the input voltage curve of each test data, so that the duration of the equivalent steady-state time period in all test data has the minimum difference. The voltage fluctuation characteristics of the input voltage curve within the equivalent stable time period of all test data are extracted and denoted as the allowable fluctuation characteristics; the voltage fluctuation characteristics of the input voltage curve outside the equivalent stable time period of each test data are denoted as the excitation fluctuation characteristics. The first model is trained using the rotational speed, removed load, and excitation fluctuation characteristics from all test data. The excitation fluctuation characteristics during a short circuit are predicted using the first model. The excitation fault coefficient caused by the excitation energy impact of the main exciter during a short circuit is obtained based on the difference between the predicted excitation fluctuation characteristics and the allowable fluctuation characteristics. The delay fault coefficient of the controller is determined based on the average duration of the equivalent stable time period in all test data. The test results are determined based on the excitation fault coefficient and the delay fault coefficient.

[0007] Preferably, the specific steps for extracting the equivalent stable time period of voltage from the input voltage curve of each test data point, so that the duration of the equivalent stable time period in all test data points has the minimum difference, are as follows: S1: For any acquisition time point in the input voltage curve of each test data, calculate the stability characteristics of that acquisition time point based on the voltage within the preset neighborhood time period of that acquisition time point; take the neighborhood time period of the acquisition time point with the largest stability characteristics as the temporary stable time period. S2: Determine the first indicator based on the length difference of the temporary stable time period corresponding to all test data, and the first indicator is positively correlated with the length difference; S3: Among all the temporary stable time periods corresponding to all test data, for the shortest temporary stable time period, on the input voltage curve where the temporary stable time period is located, the temporary stable time period is extended according to the stability characteristics of all sampling time points within the temporary stable time period. S4: Repeat S2~S3 until the average length of the temporary stable time period corresponding to all test data is greater than the preset length; among them, the temporary stable time period corresponding to all test data obtained when the first index is at its minimum is recorded as the equivalent stable time period.

[0008] Preferably, the specific steps for obtaining the permissible fluctuation characteristics and the excitation fluctuation characteristics are as follows: Perform a continuous wavelet transform on any input voltage curve, and use the transform coefficients of all frequency components under each shift factor as each spectrum vector; the spectrum vectors corresponding to all shift factors in the input voltage curve that fall within the equivalent steady time period are denoted as the allowable spectrum vectors; the spectrum vectors corresponding to all shift factors outside the equivalent steady time period are denoted as the excitation spectrum vectors. For all permissible spectrum vectors obtained from the input voltage curves in all test data, the closed interval formed by the maximum and minimum values ​​of all transformation coefficients of the same frequency component in all permissible spectrum vectors is denoted as the fluctuation range of the same frequency component, and the fluctuation range of all frequency components obtained from all permissible spectrum vectors constitutes the permissible fluctuation characteristic. For all excitation spectrum vectors obtained from the input voltage curve in each test data, the closed interval formed by the maximum and minimum values ​​of all transformation coefficients of the same frequency component in all excitation spectrum vectors is denoted as the fluctuation range of the same frequency component. The fluctuation range of all frequency components obtained from all excitation spectrum vectors constitutes the excitation fluctuation characteristic.

[0009] Preferably, the specific steps for obtaining the excitation fault coefficient are as follows: For any frequency component, obtain the crossover ratio of the predicted excitation fluctuation characteristic fluctuation range in that frequency component and the allowable fluctuation characteristic fluctuation range in that frequency component. When the predicted excitation fluctuation characteristic fluctuation range in that frequency component is completely within the allowable fluctuation characteristic fluctuation range in that frequency component, set the crossover ratio to 1. The first mean of the cross-parallel ratio calculated for all frequency components is obtained. The difference between the predicted excitation fluctuation characteristics and the allowable fluctuation characteristics is negatively correlated with the first mean. The difference between the predicted excitation fluctuation characteristics and the allowable fluctuation characteristics is used as the excitation fault coefficient.

[0010] Preferably, the specific steps for determining the test results based on the excitation fault coefficient and the delay fault coefficient are as follows: When the excitation fault coefficient is greater than the first preset threshold, the system will display "Excitation impact / power supply design test failed"; when the delay fault coefficient is greater than the second preset threshold, the system will display "Controller code / hardware test failed"; wherein, the delay fault coefficient is positively correlated with the average duration of the equivalent stable time period in all test data.

[0011] Preferably, the specific steps for determining the first index based on the length difference of the temporary stable time period corresponding to all test data are as follows: The average length of the horizontal axis range of all input voltage curves is recorded as the reference length; the length difference of the temporary stable time period corresponding to all test data is equal to the ratio of the standard deviation of the length of all temporary stable time periods to the reference length; the mean of the length of the temporary stable time period corresponding to all test data is obtained, and the ratio of this mean to the reference length is recorded as the average length index; the mean of the stability characteristics of all sampling time points within the temporary stable time period corresponding to all test data is obtained and recorded as the average stability characteristic index. The first indicator is positively correlated with the length difference, and negatively correlated with the average length indicator and the average stability characteristic indicator, respectively.

[0012] Preferably, the specific steps for extending the temporary stable period based on the stability characteristics of all sampling time points within the temporary stable period are as follows: For all sampling time points within the temporary stable period, the sampling time point with the largest stability feature is obtained, and the union of the neighborhood time point of the sampling time point and the temporary stable period is taken as the extension result of the temporary stable period. If the temporary stable time periods before and after the extension are the same, then the sampling time point with the second largest stable feature is obtained within the temporary stable time period. The extension result is obtained by taking the union of the neighborhood time period of the sampling time point and the temporary stable time period. If the temporary stable time period before and after the extension is still the same, then obtain the third largest sampling time point of the stable feature within that temporary stable time period, and so on, until the temporary stable time periods before and after the extension are different.

[0013] Preferably, the stability characteristics are negatively correlated with the standard deviation of the voltage in the neighborhood time period and the difference between the average voltage in the neighborhood time period and the rated input voltage of the controller.

[0014] Preferably, when using the first model to predict the excitation fluctuation characteristics during a short circuit, the inputs of the first model also include the minimum speed and rated load of the high-voltage DC power supply system.

[0015] Another embodiment of the present invention provides a short-circuit simulation test device for a high-voltage direct current power supply system. The device includes a load cabinet and a voltage sensor. The load cabinet contains several resistors, each representing a different load, including a minimum safe load. Different loads are switched to the minimum safe load within the load cabinet via controlled thyristor switches. The load reduction when switching to the minimum safe load is considered as the removed load. The voltage sensor is used to acquire the input voltage of the controller. The device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor reads the removed load from the load cabinet and the input voltage acquired by the voltage sensor, and executes the computer program to implement all steps of the aforementioned short-circuit simulation test method for a high-voltage direct current power supply system.

[0016] The beneficial effects of the technical solution of the present invention are: This invention simulates short-circuit impacts through load removal testing, acquiring test data including rotational speed, controller input voltage curves, controller response time, and load removal. An equivalent stable time period is extracted from the input voltage curve, distinguishing between permissible fluctuation characteristics and excitation fluctuation characteristics. A mapping model is established between short-circuit impact conditions and abnormal power supply fluctuation modes. The excitation fluctuation characteristics under actual short-circuit conditions are predicted and compared with the permissible voltage fluctuation characteristics for stable controller operation, quantifying the impact of excitation energy impacts on controller power supply stability. Combined with the controller's own delay benchmark, a root cause decomposition assessment of short-circuit risks is achieved. Overall, this invention comprehensively evaluates the backlash effect of electromechanical coupling effects on controller power supply stability during short-circuit faults, avoiding the shortcomings of existing technologies that only focus on controller delays while ignoring test deviations caused by power supply anomalies, thus improving the accuracy and reliability of short-circuit simulation testing. Specifically, by extracting equivalent stable time periods and minimizing their duration differences, a comparable controller stable operating time benchmark is ensured under different test conditions; voltage fluctuation characteristics are extracted through continuous wavelet transform, enabling a comprehensive characterization of voltage fluctuation patterns from a frequency domain perspective; excitation fluctuation characteristics under real short-circuit conditions are predicted by training the first model, enabling quantitative assessment of the most severe impact consequences without the need for destructive real short-circuit testing; and the root causes of short-circuit risks can be accurately located by calculating the excitation fault coefficient and the delay fault coefficient, providing a direct and clear decision-making basis for system improvement. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the steps of a short-circuit simulation test method for a high-voltage DC power supply system provided in one embodiment of the present invention. Figure 2 This is the input voltage curve. Detailed Implementation

[0018] The following description, in conjunction with the accompanying drawings, details the specific scheme of the short-circuit simulation test method and equipment for high-voltage DC power supply systems provided by this invention.

[0019] Please see Figure 1 The diagram illustrates a flowchart of a short-circuit simulation test method for a high-voltage DC power supply system according to an embodiment of the present invention. The method includes the following steps: Step S101: Perform a load removal test on the high-voltage DC power supply system to obtain several test data points; each test data point includes the rotational speed, the controller's input voltage curve, the controller's response time, and the removed load.

[0020] The high-voltage direct current power supply system is a three-stage brushless structure, comprising a coaxially rotating permanent magnet auxiliary exciter and a main exciter. Its working principle is as follows: a power source (e.g., a gasoline engine) drives the permanent magnet auxiliary exciter to generate electricity, which, after rectification, powers the controller; the controller outputs controllable DC current as the stator excitation current of the main exciter; the main exciter rotates to generate electricity, and its output, after rectification, becomes the final high-voltage direct current of the high-voltage direct current power supply system.

[0021] A core function of the controller is to prevent the short-circuit current from lasting too long and thus prevent the high-voltage DC power supply system from burning out when a short circuit occurs at the load end. This is achieved through the controller's response (such as cutting off the excitation current or triggering the protective actions of safety devices like relays). However, due to inherent delays in the controller's response (e.g., delays in short-circuit current sensing, internal program processing, power drive, and protection logic triggering), the short circuit cannot be eliminated in a timely manner, resulting in a certain duration of short-circuit current. When these delays cause the controller's response time to be too long, the short-circuit current duration becomes excessively prolonged, posing a high risk of burnout.

[0022] This embodiment safely simulates a short circuit and collects data in the laboratory. Specifically: For the high-voltage DC power supply system under test, multiple speeds were set up, and different known loads (denoted as target loads) were applied sequentially at each speed. After each target load was applied, the system immediately switched to the minimum safe load (i.e., load removal operation), simulating the electrical impact process of a sudden short circuit at the load end. The minimum safe load was designed to protect the system from burnout during repeated tests. This operating environment ensures that tests can be safely and repeatedly performed under laboratory conditions.

[0023] During the above process, the following are recorded: the rotational speed corresponding to each load removal operation; the size of the load removed (the difference between the target load and the minimum safe load); the controller response time, i.e., the time from the moment the load is switched to the time it takes for the controller to complete the regulation (such as the output current returning to zero); and the controller's input voltage curve during the entire response time period.

[0024] Each record corresponding to a load removal operation constitutes one test data point; the input voltage curves contained in two test data points are as follows: Figure 2 Specifically, in some embodiments, if the controller fails to complete regulation (e.g., bring the output current to zero) for an extended period (e.g., within 3 seconds), the test data is considered invalid and deleted.

[0025] After completing all load removal operations, multiple test data points are obtained. These test data points represent multiple tests conducted under different test conditions (such as different rotational speeds and different load impact intensities), ensuring the diversity and coverage of the data samples. This allows the subsequently extracted features and trained models to have better generalization ability and representativeness.

[0026] The rotational speed in each test data point directly reflects the initial instantaneous state of the permanent magnet exciter during the test, directly demonstrating the electromechanical coupling effect. The controller's input voltage curve represents the change in the controller's input voltage over time, while the controller's response time quantifies the time required for the controller to complete the control action from sensing a load change. The removed load quantifies the intensity of the short-circuit impact simulated in each test. These four parameters, as a whole data package, collectively and completely characterize a load removal event.

[0027] Step S102: Extract the equivalent stable time period of voltage from the input voltage curve of each test data, so that the duration of the equivalent stable time period in all test data has the smallest difference.

[0028] This embodiment further considers that when the load of the high-voltage DC power supply system is short-circuited, the huge short-circuit current (abrupt change in excitation energy) in the main exciter will instantaneously affect the speed of the permanent magnet auxiliary exciter through mechanical coupling of the shaft. The speed of the permanent magnet auxiliary exciter directly determines the voltage it supplies to the controller, thus causing instability in the controller's input voltage. Therefore, a short-circuit fault creates a feedback loop: short-circuit fault → affects the controller's input voltage → interferes with the controller's short-circuit response process → may increase the duration of the short-circuit circuit.

[0029] This embodiment uses the equivalent settling time period to describe the continuous time in the input voltage curve of each test that supports the controller in stably completing the short-circuit response operation without interfering with the controller's short-circuit response process. Furthermore, regardless of changes in test conditions (speed, load), the shortest time required for the controller to operate stably is relatively fixed. Therefore, the time period with the smallest difference in duration extracted from the input voltage curves of each test is the equivalent settling time period for each test.

[0030] Step S103: Extract the voltage fluctuation characteristics of the input voltage curve within the equivalent stable time period of all test data, and record them as allowable fluctuation characteristics; extract the voltage fluctuation characteristics of the input voltage curve outside the equivalent stable time period of each test data, and record them as excitation fluctuation characteristics.

[0031] Since the equivalent stable time period is constrained to the time interval during which the controller can operate stably, the voltage fluctuation patterns extracted from these intervals of all test data collectively constitute a set of voltage fluctuation ranges characterizing the controller's ability to tolerate and operate normally. This embodiment uses the permissible fluctuation feature to quantify this set of voltage fluctuation ranges. This permissible fluctuation feature is the power supply resilience baseline for the controller's stable operation. The smaller and more stable the quantified fluctuation range, the higher the controller's requirements for power supply quality, or the better the power supply quality provided by the system during stable periods.

[0032] The time period outside the equivalent stable period corresponds to the stage where the excitation energy of the main exciter changes abruptly due to the removal of a specific load (simulated short circuit) in each test, which in turn causes fluctuations in the speed of the permanent magnet auxiliary exciter through electromechanical coupling, ultimately resulting in abnormal fluctuations in the controller input voltage. In this embodiment, the voltage fluctuation characteristics of the input voltage curve outside the equivalent stable period are denoted as excitation fluctuation characteristics, which represent the fluctuation pattern extracted from the input voltage curve segment outside the equivalent stable period in a single test. This represents the abnormal voltage fluctuations caused by the specific short-circuit excitation energy impact of this test, making it impossible for the controller to operate stably.

[0033] Thus, each test data point has a corresponding excitation fluctuation characteristic, and together they establish a mapping relationship between different external impact conditions and abnormal power supply response.

[0034] Step S104: Train the first model using the rotational speed, removed load, and excitation fluctuation characteristics from all test data. Use the first model to predict the excitation fluctuation characteristics during a short circuit. Based on the difference between the predicted excitation fluctuation characteristics and the allowable fluctuation characteristics, obtain the excitation fault coefficient caused by the excitation energy impact of the main exciter during a short circuit. Determine the delay fault coefficient of the controller based on the average duration of the equivalent stable time period in all test data.

[0035] Considering that the voltage fluctuations caused by the excitation energy impact of the short-circuit current, which prevent the controller from operating stably (i.e., excitation fluctuation characteristics), depend on the impact of the short-circuit current on the permanent magnet auxiliary exciter or the main exciter, the aforementioned excitation fluctuation characteristics are related to data such as the rotational speed of the permanent magnet auxiliary exciter or the main exciter and the removed load. Based on this, this embodiment uses the rotational speed, removed load, and excitation fluctuation characteristics from all test data to train the first model. Through training, the model learns and establishes a complex nonlinear mapping relationship between the short-circuit impact condition and the resulting abnormal power supply fluctuation mode. The purpose of introducing this model is to extrapolate and predict the power supply fluctuation under the worst-case scenario (i.e., a real short circuit with 100% load) from limited, non-full-load test data.

[0036] The first model uses parameters based on real short-circuit conditions (such as removing the load to 100% rated load and corresponding speed) to predict the excitation fluctuation characteristics during a short circuit. This predicted excitation fluctuation characteristic represents the abnormal fluctuation pattern of the controller input voltage that a real short-circuit fault would cause (specifically, the input voltage fluctuation caused by the short-circuit excitation energy impact, which would prevent the controller from operating stably). This makes it possible to quantitatively assess the most severe impact consequences without performing destructive real short-circuit tests.

[0037] The predicted short-circuit excitation fluctuation characteristics are compared with the baselined allowable fluctuation characteristics, and the difference is quantified as the excitation fault coefficient. The larger this coefficient is, the more likely the impact of the short circuit on the power supply stability of the controller is to cause controller malfunction, thereby significantly prolonging the actual response time.

[0038] Furthermore, the delay fault coefficient of the controller is determined based on the average duration of the equivalent stable time period in all test data. This coefficient represents the inherent average response time benchmark of the controller's hardware and software in completing protection actions under relatively stable power supply conditions. The larger this coefficient, the greater the performance delay of the controller itself, and the higher the risk of burnout due to its slow response.

[0039] Step S105: Determine the test results based on the excitation fault coefficient and the delay fault coefficient.

[0040] The final test results simultaneously output the excitation fault coefficient and the delay fault coefficient. This achieves a root cause decomposition of short-circuit risk: whether the risk stems from external short-circuit energy impact causing power instability (high excitation fault coefficient), or from insufficient controller performance (high delay fault coefficient), or both. This precise identification provides a direct and clear basis for decision-making regarding improvements to the high-voltage DC power supply system (whether to strengthen mechanical decoupling / power supply design, or optimize controller code / hardware).

[0041] As an example, the output DC voltage of the high-voltage DC power supply system under test used in the load removal test is 270V when the speed is 21600r / min~26400r / min, and the rated output power is 50kW, while the rated input voltage of the controller is 28V.

[0042] As an example, the speeds set during the load removal test were 21,500 r / min, 22,000 r / min, 22,500 r / min, ..., 26,500 r / min.

[0043] As an example, during the load removal test, the target load is selected using alloy resistors with different resistance values, namely 200%, 133%, 100%, 80%, 67%, and 50% of the rated load. The rated load is the ratio of the square of the rated output voltage to the rated output power, for example, it equals (270V). 2 / 50kW. In this example, the minimum safe load resistance is set to 0.3Ω. The removed load is equal to the difference between the target load and the minimum safe load.

[0044] In other examples, different values ​​can be set for the above parameters depending on the specific circumstances of the high-voltage DC power supply system. This embodiment does not impose specific limitations.

[0045] As an example, a voltage sensor is used to acquire the input voltage of the controller at a frequency of 1 kHz.

[0046] As a preferred example, the equivalent steady-state time period of the voltage is extracted from the input voltage curve of each test data point, minimizing the difference in the duration of the equivalent steady-state time periods across all test data points. This can be achieved through methods such as: (1) The vertical axis of the input voltage curve is voltage, and the horizontal axis is the voltage acquisition time point. For any acquisition time point in the input voltage curve of each test data, the multiple acquisition time points with the smallest absolute value of the difference from the acquisition time point (e.g., 11 acquisition time points including the acquisition time point) are recorded as the neighborhood time points of the acquisition time point; the time period formed by all neighborhood time points is recorded as the neighborhood time period of the acquisition time point; the stability characteristics of the acquisition time point are calculated based on the voltage within the neighborhood time period, and the stability characteristics are negatively correlated with the voltage fluctuation characteristics within the neighborhood time period and the difference between the voltage within the neighborhood time period and the rated input voltage of the controller. The more severe the voltage fluctuations within the neighborhood time period (the larger the fluctuation characteristics) or the greater the difference between the voltage within the neighborhood time period and the controller's rated input voltage, the smaller the stability characteristics, indicating that the voltage near that sampling time point is more unstable and more likely to cause abnormal controller operation. Conversely, the less severe the voltage fluctuations within the neighborhood time period (the smaller the fluctuation characteristics) or the smaller the difference between the voltage within the neighborhood time period and the controller's rated input voltage, the larger the stability characteristics, indicating that the voltage near that sampling time point is more stable and more likely to ensure stable controller operation.

[0047] In the input voltage curve, the sampling time point with the largest stability characteristic is obtained, and its neighborhood time period is taken as the temporary stable time period. Thus, the input voltage curve of each test data corresponds to a temporary stable time period, and each temporary stable time period corresponds to a sampling time point.

[0048] In particular, when there are multiple sampling time points with the greatest stability characteristics in the input voltage curve, the neighborhood time period of the sampling time point with the latest timing sequence is taken as the temporary stable time period. This is because the later the timing sequence, the more likely the controller's input voltage is to stabilize.

[0049] (2) The first index is determined based on the length difference of the temporary stable time period corresponding to all test data. The first index is positively correlated with the length difference. The smaller the length difference (that is, the more similar the length of the temporary stable time period of all test data), the smaller the first index.

[0050] (3) Among all the temporary stable time periods corresponding to all test data, for the shortest temporary stable time period, the temporary stable time period is extended on the input voltage curve where the temporary stable time period is located, based on the stability characteristics of all sampling time points within the temporary stable time period.

[0051] (4) Repeat steps (2) to (3) until the average length of the temporary stable time period corresponding to all test data is greater than the preset length (e.g., 60% of the average length of the horizontal axis range of all input voltage curves). During this process, the temporary stable time periods corresponding to all test data obtained when the first index is at its minimum are all recorded as equivalent stable time periods; such as... Figure 2 The equivalent settling time periods of the two input voltage curves are marked in the figure.

[0052] As an example, the stability characteristics of a sampling time point are calculated based on the voltage within a neighboring time period, including the following methods: Obtain the absolute value of the difference between the average voltage over the neighborhood time period and the controller's rated input voltage (which is equal to 28V in this example). The ratio of this absolute value of the difference to the controller's rated input voltage is denoted as the voltage offset value.

[0053] Obtain the standard deviation of all voltages within the neighborhood time period. The ratio of this standard deviation to the rated input voltage of the controller is recorded as the voltage fluctuation value.

[0054] The purpose of comparing the value with the controller's rated input voltage is to remove the dimensions.

[0055] The stability characteristics are negatively correlated with both voltage offset and voltage fluctuation values.

[0056] As an example, the formula for calculating stable features includes: F = exp(-x), where F represents the stability characteristic, x represents the average of the voltage offset and voltage fluctuation, and exp() represents an exponential function with the natural constant as the base.

[0057] As an example, the first metric is determined based on the difference in the length of the temporary stable time period corresponding to all test data, including the following methods: The average length of the horizontal axis range of all input voltage curves is recorded as the reference length.

[0058] The length difference of the temporary stable time periods corresponding to all test data is equal to the ratio of the standard deviation of the length of all temporary stable time periods to the reference length. The mean length of the temporary stable time periods corresponding to all test data is obtained, and the ratio of this mean to the reference length is denoted as the average length index. The mean of the stability characteristics at all sampling time points within the temporary stable time periods corresponding to all test data is obtained and denoted as the average stability characteristic index.

[0059] The purpose of comparing the length with the reference length is to remove the dimension.

[0060] The first indicator is positively correlated with the length difference and negatively correlated with the average length indicator and the average stability characteristic indicator, respectively. A smaller length difference indicates a smaller difference in the length of the temporary stable time periods corresponding to all test data. Furthermore, a larger average length indicator or average stability characteristic indicator (in which case the first indicator is smaller) indicates that the temporary stable time periods corresponding to all test data have a large length or that the temporary stable time periods can ensure stable operation of the controller.

[0061] When the first indicator is minimized, on the one hand, the difference in the length of the equivalent stable time period corresponding to all test data is minimized, and on the other hand, the temporary stable time period corresponding to all test data has the maximum length and the maximum working stability. This allows the equivalent stable time period to comprehensively include and cover all input voltage fluctuation modes allowed by the controller as much as possible.

[0062] As an example, the formula for calculating the first indicator includes: P = w1 × P1 - w2 × P2 - w3 × P3, where P represents the first index, P1 represents the length difference, P2 and P3 represent the average length index and the average stability characteristic index, respectively, and w1, w2, and w3 represent preset attention coefficients. In this example, w1, w2, and w3 are set to 5, 1, and 1, respectively. In other embodiments, w1, w2, and w3 can be set to other values, and this embodiment does not limit this.

[0063] As an example, extending the temporary stable period based on the stability characteristics of all sampling time points within that temporary stable period includes the following methods: For all sampling time points within the temporary stable period, the sampling time point with the largest stability feature is obtained, and the union of the neighborhood time period of this sampling time point and the temporary stable period is calculated. The time period corresponding to this union is used as the extension result of the temporary stable period.

[0064] If the temporary stable time periods before and after the extension are the same, then the sampling time point with the second largest stable feature is obtained within the temporary stable time period. The union of the neighborhood time period of the sampling time point and the temporary stable time period is taken to obtain the extension result.

[0065] If the temporary stable time period before and after the extension is still the same, then obtain the third largest sampling time point of the stable feature within that temporary stable time period, and so on, until the temporary stable time periods before and after the extension are different.

[0066] In special cases, if there are multiple sampling time points with the largest or second largest stable features within the temporary stable time period, the neighboring time periods of all the largest or second largest sampling time points are first joined together and then joined together with the temporary stable time period.

[0067] As a preferred example, the voltage fluctuation characteristics of the input voltage curve within the equivalent stable time period of all test data are extracted and denoted as the allowable fluctuation characteristics; the voltage fluctuation characteristics of the input voltage curve outside the equivalent stable time period of each test data are denoted as the excitation fluctuation characteristics, including the following methods: Perform a continuous wavelet transform on any input voltage curve to obtain the transform coefficients of each scaling factor (scale) and translation factor (time). The transform coefficients of all scaling factors under the same translation factor constitute a spectrum vector (each dimension represents a scaling factor, and each scaling factor is regarded as a frequency component). The spectrum vector is used to describe the voltage response at different sampling time points under each frequency component.

[0068] For all translation factors in the input voltage curve that fall within the equivalent steady-state time period, the corresponding spectral vectors are denoted as the allowable spectral vectors. The spectral vectors corresponding to all translation factors outside the equivalent steady-state time period are denoted as the excitation spectral vectors.

[0069] For all permissible spectral vectors obtained from the input voltage curves in all test data, the closed interval formed by the maximum and minimum values ​​of all transform coefficients in the same dimension (or the same scaling factor or the same frequency component) of all permissible spectral vectors is denoted as the fluctuation range of the same dimension (or the same scaling factor or the same frequency component). The fluctuation range of all dimensions or all frequency components constitutes the permissible fluctuation characteristic.

[0070] For all excitation spectrum vectors obtained from the input voltage curve in each test data, the closed interval formed by the maximum and minimum values ​​of all transformation coefficients in the same dimension (or the same scaling factor or the same frequency component) in all excitation spectrum vectors is denoted as the fluctuation range of the same dimension (or the same scaling factor or the same frequency component). The fluctuation range of all dimensions or all frequency components constitutes the excitation fluctuation characteristic.

[0071] In other examples, the fluctuation range of the same dimension (or the same scaling factor or the same frequency component) can be obtained using the following method to avoid the noise influence of the transformation coefficients: Obtain the standard deviation g and mean u of all transformation coefficients for the same dimension (or the same scaling factor or the same frequency component), and use [u-3g, u+3g] as the fluctuation range.

[0072] As an example, the wavelet function used in the wavelet transform is the db4 wavelet.

[0073] As an optional example, the voltage fluctuation characteristics of the input voltage curve within the equivalent stable time period of all test data are extracted and denoted as the allowable fluctuation characteristics; the voltage fluctuation characteristics of the input voltage curve outside the equivalent stable time period of each test data are denoted as the excitation fluctuation characteristics, including the following methods: The above uses continuous wavelet transform, which has the advantages of high computational cost and data redundancy. This example performs discrete wavelet transform on any input voltage curve, decomposing it into J layers (e.g., J=8).

[0074] Considering that discrete wavelet transform cannot directly obtain the transform coefficients of different scaling factors under different translation factors (or different sampling time points), this example achieves a similar effect to the preferred example above through a mapping method. Specifically, for any sampling time point on the input voltage curve (equivalent to the aforementioned translation factor), assuming the index of this sampling time point is n; for the detail coefficient cDj of the j-th (j=1, 2, ..., J) layer, the information of this sampling time point is mainly contained in the k(j)-th coefficient of the j-th layer, where k(j) equals n / 2. j Rounding down, the vector [cD1[k(1)], cD2[k(2)], ..., cDj[k(j)], ..., cDJ[k(J)]] is now considered as the spectral vector at that sampling time point, where cDj[k(j)] represents the k(j)th coefficient (equivalent to the scaling factor mentioned above) in the detail coefficients of the j-th layer.

[0075] For all sampling time points within the equivalent steady-state period of the input voltage curve, the corresponding spectral vectors are denoted as the allowable spectral vectors. The spectral vectors corresponding to all sampling time points outside the equivalent steady-state period are denoted as the excitation spectral vectors.

[0076] Then, the permissible fluctuation characteristics and excitation fluctuation characteristics are obtained according to the preferred example above.

[0077] This optional example requires less computation, but its accuracy is relatively low due to potential errors in the mapping relationship described above.

[0078] As a preferred example, the first model is trained using the rotational speed, removed load, and excitation fluctuation characteristics from all test data, including the following methods: The rotational speed, removed load, and minimum safe load in each test data point are taken as a sample, and the excitation fluctuation feature is taken as the label of the sample. All the test data points and the labels constitute a dataset. The first model is trained using this dataset. The input of the first model is the rotational speed, removed load, and minimum safe load, and the output is the excitation fluctuation feature.

[0079] In some examples, the minimum safe load is a constant, so it may not be used as an element in the sample or as input to the first model to reduce the number of parameters in the first model.

[0080] It should be noted that this example uses a vectorized representation of the excitation fluctuation characteristics (or allowable fluctuation characteristics), specifically including: Each excitation fluctuation characteristic (or permissible fluctuation characteristic) contains the fluctuation range of the input voltage under different frequency components. The lower and upper limits of the fluctuation range of each frequency component are regarded as a unit vector, and the vector obtained by concatenating the unit vectors of the fluctuation ranges of all frequency components is the vectorization result.

[0081] In other examples, considering that the excitation fluctuation features and permissible fluctuation features have too many dimensions and require a large amount of computation, which is not conducive to the training of the first model, the obtained magnetic fluctuation features and permissible fluctuation features are dimensionality reduced. Specifically, for any frequency component, the fluctuation ranges of all magnetic fluctuation features and all permissible fluctuation features under that frequency component are obtained, and the similarity of all fluctuation ranges is calculated and recorded as the similarity index of that frequency component. The larger the similarity index, the more likely that all magnetic fluctuation features and all permissible fluctuation features have the same or similar fluctuation situation under that frequency component. The less necessary that frequency component is to participate in the calculation and analysis of this embodiment, that is, the fluctuation range under that frequency component does not need to be concerned and needs to be deleted.

[0082] Specifically, the TOPK frequency components with the largest similarity index are obtained, and the fluctuation range of the TOPK frequency components is removed from all magnetic fluctuation features and all permissible fluctuation features, thereby obtaining the dimension-reduced magnetic fluctuation features and permissible fluctuation features.

[0083] As an example, the preferred value range for TOPK is 30% to 70% of the total number of frequency components. This example uses TOPK equal to 40% (rounded up) of the total number of frequency components.

[0084] As an example, the methods for calculating the similarity of all fluctuation ranges include: The crossover ratio (CRO) is calculated pairwise for each of all fluctuation ranges, and the mean of the CROs for all fluctuation ranges is used as the similarity.

[0085] As an example, the first model uses a fully connected neural network with three intermediate layers, each containing 5, 8, and 15 neurons respectively. During training, the mean squared error loss function is used, the optimizer is Adam, the learning rate is set to 0.01, the batch size is 6, and the maximum number of training iterations is set to 10. 4 Since fully connected neural networks and their training methods are well known, this embodiment will not elaborate on them.

[0086] In other embodiments, preprocessing can be performed when collecting or recording each test data to improve training efficiency. The preprocessing includes removing dimensions and reducing the order of magnitude, specifically including: dividing the rotational speed of each test data by 26500 r / min, dividing the removed load (and minimum safe load) by the rated resistance, dividing the collected input voltage by 28V, and dividing the controller response time by 50ms. This process achieves the removal of dimensions and reduction of the order of magnitude of each test data by dividing each test data by a preset value. Other preset values ​​can also be used in other embodiments, and this embodiment does not impose specific limitations (this embodiment does not need to guarantee that the preprocessed test data are all normalized).

[0087] As an example, the method for predicting excitation fluctuation characteristics during a short circuit using the first model includes: In this example, the speed of the high-voltage DC power supply system when it is operating stably is 21600 r / min to 26400 r / min. Therefore, this example uses 21600 r / min and the rated load as the inputs for the first model prediction (including the minimum safe load). The first model outputs the predicted excitation fluctuation characteristics, which include the fluctuation range under different frequency components. This means that when the high-voltage DC power supply system is operating stably with the rated load at the minimum speed (corresponding to 21600 r / min and the rated load), the input voltage fluctuation under different frequency components caused by the short-circuit excitation energy impact after a short circuit occurs, which makes the controller unable to operate stably.

[0088] It should be noted that if some embodiments use preprocessing methods to remove dimensions and orders of magnitude, then the input data or other data with dimensions in this example also need to have their dimensions and orders of magnitude removed in the same way.

[0089] As a preferred example, the excitation fault coefficient caused by the excitation energy impact of the main exciter during a short circuit is obtained based on the difference between the predicted excitation fluctuation characteristics and the allowable fluctuation characteristics. The method includes: The predicted excitation fluctuation characteristics also represent the range of fluctuations in the input voltage at different frequency components caused by the short-circuit excitation energy impact after a short circuit, which prevents the controller from operating stably; the permissible fluctuation characteristics represent the range of fluctuations in the input voltage at different frequency components that the controller is allowed to operate stably.

[0090] For any frequency component, obtain the crossover ratio of the predicted excitation fluctuation characteristic fluctuation range under that frequency component and the allowable fluctuation characteristic fluctuation range under that frequency component. When the predicted excitation fluctuation characteristic fluctuation range under that frequency component is completely within the allowable fluctuation characteristic fluctuation range under that frequency component (i.e., the latter includes the former), directly set the crossover ratio to 1.

[0091] For all frequency components, the crossover and parallelization ratio is calculated according to the above method, and the first mean of the crossover and parallelization ratios calculated for all frequency components is obtained. The difference between the predicted excitation fluctuation characteristics and the allowable fluctuation characteristics is negatively correlated with the first mean.

[0092] This example uses the difference between the predicted excitation fluctuation characteristics and the permissible fluctuation characteristics as the excitation fault coefficient.

[0093] The larger the first mean, the more likely the predicted excitation fluctuation characteristics are similar to the allowable fluctuation characteristics in terms of voltage fluctuation across all frequency components. In this case, the smaller the difference between the predicted excitation fluctuation characteristics and the allowable fluctuation characteristics, the more likely the fluctuation range of the input voltage caused by the short-circuit excitation energy impact, which prevents the controller from operating stably, is the same as or similar to the fluctuation range of the input voltage allowed when the controller is operating stably (or even the former falls within the latter). This means that during a short circuit, the fluctuation of the input voltage caused by the short-circuit excitation energy impact falls within or is similar to the allowable fluctuation range of the controller, and will not significantly increase the controller's response delay. In this case, the excitation fault coefficient is smaller.

[0094] The smaller the first mean, the more significantly the predicted excitation fluctuation characteristics differ from the overall allowable fluctuation characteristics across all frequency components. The greater the difference between the predicted and allowable fluctuation characteristics, the more significant the difference between the predicted and allowable fluctuation characteristics. This indicates that the fluctuation range of the input voltage caused by the short-circuit excitation energy impact, which prevents the controller from operating stably, is significantly different from the allowable fluctuation range of the input voltage when the controller is operating stably. This means that during a short circuit, the input voltage fluctuation caused by the short-circuit excitation energy impact cannot guarantee that the controller will operate within its allowable fluctuation range, significantly increasing the controller's response delay. In this case, the excitation fault coefficient is larger.

[0095] As an example, the formula for calculating the difference between the predicted excitation fluctuation characteristics and the allowable fluctuation characteristics is: D = exp(-y), where D represents the difference between the predicted excitation fluctuation characteristics and the allowable fluctuation characteristics (i.e., the excitation fault coefficient), y represents the first mean, and exp() represents an exponential function with the natural constant as the base.

[0096] As an example, the delay failure coefficient of the controller is determined based on the average duration of the equivalent stable time period across all test data, using the following formula: The delay fault coefficient is obtained by comparing the average duration of the equivalent stable time period across all test data with the preset reference duration. A larger delay fault coefficient (i.e., a larger average duration of the equivalent stable time period) indicates that the controller experienced a longer response time before initiating short-circuit protection when the power supply was relatively stable, and its own delay is more severe. The purpose of comparing it with the preset reference duration is to remove dimensions.

[0097] It should be noted that the average duration of the equivalent settling time period is equivalent to the response time of the controller under the extreme condition of operating within its allowable fluctuation range. This is helpful in analyzing the test results under the special condition of unstable input voltage.

[0098] One method for obtaining the preset reference duration is as follows: Perform a withstand voltage test on the stator coils of the main exciter: input a current several times (e.g., 3 times) of the rated current into the stator coils of the main exciter, and obtain the time it takes for the stator coils to blow. Use this time as the reference duration. The rated current represents the output current when outputting rated power at rated voltage (e.g., 50kW / 270V).

[0099] As a preferred example, the method for determining test results based on the excitation fault coefficient and the delay fault coefficient includes: When the excitation fault coefficient is greater than the first preset threshold, it indicates that the short-circuit excitation energy impact during a short-circuit fault significantly increases the risk of short-circuit burnout. At this time, the message "Excitation impact / power supply design test unqualified" is displayed, and mechanical decoupling / power supply design needs to be strengthened.

[0100] When the delay fault coefficient is greater than the second preset threshold, it indicates that the controller's own delay has significantly increased the risk of short circuit and burnout. At this time, the message "Controller code / hardware test failed" will be displayed, and the controller code / hardware needs to be optimized.

[0101] This example uses a first preset threshold of 0.47 as an example and a second preset threshold of 0.58 as an example. In other embodiments, the first and second preset thresholds can be set to other values, and this embodiment does not impose specific limitations.

[0102] As an optional example, determining test results based on excitation fault coefficient and delay fault coefficient includes the following methods: The ratio of the excitation fault coefficient to the delay fault coefficient is calculated. When this ratio is greater than the third preset threshold, it indicates that the risk of short-circuit burnout caused by the short-circuit excitation energy impact is greater than the risk of short-circuit burnout caused by the controller's own delay. At this time, the message "Excitation impact / power supply design test failed, please strengthen the mechanical decoupling / power supply design and then retest" is displayed. In this example, the main problem of the excitation impact / power supply design test failure is solved first, and then this embodiment is rerun. When the ratio is less than the third preset threshold, the above preferred example is run.

[0103] This example uses a third preset threshold of 1.6 as an example. In other embodiments, the third preset threshold can be set to other values, and this embodiment does not impose specific limitations.

[0104] This optional example avoids the problem of unreliable assessment of the controller's own delay due to excessive short-circuit excitation energy surge.

[0105] Another embodiment of the present invention provides a short-circuit simulation test device for a high-voltage DC power supply system. The device includes a load cabinet and a voltage sensor. The load cabinet contains multiple alloy resistors, each representing a different target load and a minimum safe load. A controlled thyristor (SCR) switch within the load cabinet rapidly switches the different target loads to the minimum safe load to simulate a short-circuit impact. The voltage sensor is used to acquire the input voltage of the controller. The device also includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements all the steps of all the embodiments described above.

[0106] The device also includes an interactive module, in which a touchscreen is connected to the processor. The touchscreen is used to display test results. Additionally, several hyperparameters can be set via the interactive interface on the touchscreen. These hyperparameters include a first preset threshold, a second preset threshold, etc. The default value for each hyperparameter should be consistent with the above embodiment. Other embodiments may add more hyperparameters; this embodiment does not impose specific limitations.

[0107] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A short-circuit simulation test method for high-voltage DC power supply systems, characterized in that, The method includes the following steps: A load removal test was performed on the high-voltage DC power supply system, and several test data points were obtained. Each test data point includes the rotational speed, the controller's input voltage curve, the controller's response time, and the load removed. Extract the equivalent steady-state time period of the voltage from the input voltage curve of each test data, so that the duration of the equivalent steady-state time period in all test data has the minimum difference. The voltage fluctuation characteristics of the input voltage curve within the equivalent stable time period of all test data are extracted and denoted as the allowable fluctuation characteristics; the voltage fluctuation characteristics of the input voltage curve outside the equivalent stable time period of each test data are denoted as the excitation fluctuation characteristics. The first model is trained using the rotational speed, removed load, and excitation fluctuation characteristics from all test data. The excitation fluctuation characteristics during a short circuit are predicted using the first model. The excitation fault coefficient caused by the excitation energy impact of the main exciter during a short circuit is obtained based on the difference between the predicted excitation fluctuation characteristics and the allowable fluctuation characteristics. The delay fault coefficient of the controller is determined based on the average duration of the equivalent stable time period in all test data. The test results are determined based on the excitation fault coefficient and the delay fault coefficient. The specific steps for obtaining the permissible fluctuation characteristics and the excitation fluctuation characteristics are as follows: Perform a continuous wavelet transform on any input voltage curve, and use the transform coefficients of all frequency components under each shift factor as each spectrum vector; the spectrum vectors corresponding to all shift factors in the input voltage curve that fall within the equivalent steady time period are denoted as the allowable spectrum vectors; the spectrum vectors corresponding to all shift factors outside the equivalent steady time period are denoted as the excitation spectrum vectors. For all permissible spectrum vectors obtained from the input voltage curves in all test data, the closed interval formed by the maximum and minimum values ​​of all transformation coefficients of the same frequency component in all permissible spectrum vectors is denoted as the fluctuation range of the same frequency component, and the fluctuation range of all frequency components obtained from all permissible spectrum vectors constitutes the permissible fluctuation characteristic. For all excitation spectrum vectors obtained from the input voltage curve in each test data, the closed interval formed by the maximum and minimum values ​​of all transformation coefficients of the same frequency component in all excitation spectrum vectors is denoted as the fluctuation range of the same frequency component. The fluctuation range of all frequency components obtained from all excitation spectrum vectors constitutes the excitation fluctuation characteristic.

2. The short-circuit simulation test method for high-voltage DC power supply systems according to claim 1, characterized in that, The specific steps involved in extracting the equivalent stable time period of voltage from the input voltage curve of each test data point, so that the duration of the equivalent stable time period across all test data points has the least difference, are as follows: S1: For any acquisition time point in the input voltage curve of each test data, calculate the stability characteristics of that acquisition time point based on the voltage within the preset neighborhood time period of that acquisition time point; take the neighborhood time period of the acquisition time point with the largest stability characteristics as the temporary stable time period. S2: Determine the first indicator based on the length difference of the temporary stable time period corresponding to all test data, and the first indicator is positively correlated with the length difference; S3: Among all the temporary stable time periods corresponding to all test data, for the shortest temporary stable time period, on the input voltage curve where the temporary stable time period is located, the temporary stable time period is extended according to the stability characteristics of all sampling time points within the temporary stable time period. S4: Repeat S2~S3 until the average length of the temporary stable time period corresponding to all test data is greater than the preset length; among them, the temporary stable time period corresponding to all test data obtained when the first index is at its minimum is recorded as the equivalent stable time period.

3. The short-circuit simulation test method for high-voltage DC power supply systems according to claim 1, characterized in that, The specific steps for obtaining the excitation fault coefficient are as follows: For any frequency component, obtain the crossover ratio of the predicted excitation fluctuation characteristic fluctuation range in that frequency component and the allowable fluctuation characteristic fluctuation range in that frequency component. When the predicted excitation fluctuation characteristic fluctuation range in that frequency component is completely within the allowable fluctuation characteristic fluctuation range in that frequency component, set the crossover ratio to 1. The first mean of the cross-parallel ratio calculated for all frequency components is obtained. The difference between the predicted excitation fluctuation characteristics and the allowable fluctuation characteristics is negatively correlated with the first mean. The difference between the predicted excitation fluctuation characteristics and the allowable fluctuation characteristics is used as the excitation fault coefficient.

4. The short-circuit simulation test method for high-voltage DC power supply systems according to claim 1, characterized in that, The specific steps for determining the test results based on the excitation fault coefficient and the delay fault coefficient are as follows: When the excitation fault coefficient is greater than the first preset threshold, the message "Excitation impact / power supply design test failed" is displayed; when the delay fault coefficient is greater than the second preset threshold, the message "Controller code / hardware test failed" is displayed; wherein, the delay fault coefficient is positively correlated with the average duration of the equivalent stable time period in all test data.

5. The short-circuit simulation test method for high-voltage DC power supply systems according to claim 2, characterized in that, The specific steps for determining the first indicator based on the difference in the length of the temporary stable time period corresponding to all test data are as follows: The average length of the horizontal axis range of all input voltage curves is recorded as the reference length; the difference in the length of the temporary stable time period corresponding to all test data is equal to the ratio of the standard deviation of the length of all temporary stable time periods to the reference length; the mean of the length of the temporary stable time period corresponding to all test data is obtained, and the ratio of this mean to the reference length is recorded as the average length index; the mean of the stability characteristics of all sampling time points within the temporary stable time period corresponding to all test data is obtained and recorded as the average stability characteristic index. The first indicator is positively correlated with the length difference, and negatively correlated with the average length indicator and the average stability characteristic indicator, respectively.

6. The short-circuit simulation test method for high-voltage DC power supply systems according to claim 2, characterized in that, The specific steps involved in extending the temporary stable period based on the stability characteristics of all sampling time points within the temporary stable period are as follows: For all sampling time points within the temporary stable period, the sampling time point with the largest stability feature is obtained, and the union of the neighborhood time point of the sampling time point and the temporary stable period is taken as the extension result of the temporary stable period. If the temporary stable time periods before and after the extension are the same, then the sampling time point with the second largest stable feature is obtained within the temporary stable time period. The extension result is obtained by taking the union of the neighborhood time period of the sampling time point and the temporary stable time period. If the temporary stable time period before and after the extension is still the same, then obtain the third largest sampling time point of the stable feature within that temporary stable time period, and so on, until the temporary stable time periods before and after the extension are different.

7. The short-circuit simulation test method for high-voltage DC power supply systems according to claim 2, characterized in that, The stability characteristics are negatively correlated with the standard deviation of the voltage in the neighborhood time period and the difference between the average voltage in the neighborhood time period and the rated input voltage of the controller.

8. The short-circuit simulation test method for high-voltage DC power supply systems according to claim 1, characterized in that, When using the first model to predict the excitation fluctuation characteristics during a short circuit, the inputs of the first model also include the minimum speed and rated load of the high-voltage DC power supply system.

9. A short-circuit simulation test device for a high-voltage DC power supply system, comprising a load cabinet and a voltage sensor; wherein the load cabinet contains several resistors, each representing a different load, including a minimum safe load; the load cabinet switches different loads to the minimum safe load via controlled thyristor switches, and the load reduction when switching to the minimum safe load is considered as the removed load; the voltage sensor is used to acquire the input voltage of the controller; the device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that... When the processor reads the removed load in the load cabinet and the input voltage collected by the voltage sensor, and runs the computer program, it implements all the steps of the short-circuit simulation test method for high-voltage DC power supply systems as described in any one of claims 1 to 8.

Images