Phase change detection method

Abstract

The invention relates to the technical field of electric power, and discloses a commutation detection method, which comprises the following steps: through collaborative analysis of double-valve current signals, realizing dynamic double-feature extraction and accurate quantitative characterization of an exit valve rebound rate and conduction valve rise time in a commutation process, and utilizing geometric constraint characteristics of a preset parabola criterion equation to obtain a commutation result; according to the method, commutation state judgment is converted into geometrical relationship judgment of a criterion value and a threshold boundary, so that real-time online detection of a commutation process and accurate identification of a fault state are realized, the delay defect of an extinction angle actual measurement method is overcome, the misjudgment risk of an alternating current voltage prediction method is avoided, meanwhile, the robustness of a single current characteristic method is improved, and the method is suitable for popularization and application. And finally, the real-time performance is remarkably improved on the basis of ensuring the detection accuracy, the detection reliability of the commutation process of the high-voltage direct-current power transmission system is effectively improved, and the stability and the safety of system operation are enhanced.

Description

Technical Field

[0001] This invention relates to the field of power technology, and in particular to a commutation detection method. Background Technology

[0002] In high-voltage direct current (HVDC) transmission systems based on grid-commutated converters, the commutation process of the six-pulse converter bridge is a crucial step in smoothly transferring DC current from the de-energizing valve (about to be turned off) to the energizing valve (about to be turned on). The driving force originates from the AC line voltage between the corresponding phases of the two valves. An ideal commutation process requires the de-energizing valve current to smoothly decay from its rated value to zero, while the energizing valve current synchronously and smoothly rises to its rated value, ultimately achieving a seamless transfer of DC current. This is essential for the stable operation of the system.

[0003] However, existing commutation detection technologies have significant limitations: While measurement methods based on the arc extinction angle can determine commutation results through valve voltage or zero-current time, data acquisition is required after commutation, resulting in inherent delays and a post-hoc diagnostic approach that fails to meet real-time requirements. Predictive methods based on AC voltage provide rapid early warning by monitoring precursors such as voltage dips, but the correlation between voltage disturbances and commutation failure is affected by fault type, location, and system operating status, making them prone to misjudgment or missed detection due to non-fault disturbances. Methods based on single current characteristics directly monitor current change rate or waveform distortion, but threshold setting is difficult and they are susceptible to noise and non-fault disturbances, resulting in insufficient robustness. These technical bottlenecks lead to a triple challenge for commutation detection: delay, misjudgment, and insufficient robustness. It is difficult to balance accuracy and real-time performance, becoming a key factor restricting the reliable operation of high-voltage direct current transmission systems. Summary of the Invention

[0004] Based on this, it is necessary to propose a commutation detection method to address the above problems. This method enables real-time online detection of the commutation process and accurate identification of fault states. It overcomes the delay defect of the arc extinction angle measurement method, avoids the misjudgment risk of the AC voltage prediction method, and improves the robustness of the single current characteristic method. Ultimately, it significantly improves the real-time performance while ensuring detection accuracy, effectively enhances the detection reliability of the commutation process in the high-voltage direct current transmission system, and strengthens the stability and safety of system operation.

[0005] To achieve the above objectives, the present invention provides a commutation detection method in a first aspect, the method comprising: Acquire the exit valve current signal and the on valve current signal of the current commutation event to be tested; Based on the exit valve current signal and the conduction valve current signal of the current commutation event under test, determine the exit valve bounce rate and conduction valve rise time of the current commutation event under test. Substituting the exit valve rebound rate and the conduction valve rise time of the current commutation event under test into the preset parabolic criterion equation, the parabolic criterion value is obtained. Based on the parabolic criterion value, the commutation detection result of the current commutation event to be tested is determined.

[0006] Optionally, determining the exit valve bounce rate and on-valve rise time of the current commutation event under test based on the exit valve current signal and on-valve current signal of the current commutation event under test includes: Determine the current values ​​of the exit valve current signal of the current commutation event under test from the minimum current value to the end of commutation, and determine the exit valve rebound rate of the current commutation event under test based on the current values ​​of the exit valve current signal from the minimum current value to the end of commutation. The first moment corresponding to the current value equal to the first current value after the start of commutation in the current commutation event to be tested, and the second moment corresponding to the current value equal to the second current value are determined. Based on the first moment and the second moment, the rise time of the commutation valve in the current commutation event to be tested is determined, wherein the first current value is a first preset percentage of the preset DC current value, the second preset current value is a second preset percentage of the preset DC current value, and the first preset percentage is less than the second preset percentage.

[0007] Optionally, the exit valve rebound rate of the current commutation event under test can be determined using the following formula: ; in, The exit valve rebound rate is the current commutation event to be tested. To find the maximum value function, It is the derivative of the current value at time t from the time of the minimum value to the time of the end of commutation with respect to time t. At the time of the minimum value, The time when the commutation ends is mentioned.

[0008] Optionally, determining the on-valve rise time of the current commutation event under test based on the first time and the second time includes: The difference between the second time point and the first time point is used as the on-valve rise time of the current commutation event to be tested.

[0009] Optionally, the method further includes: Acquire the exit valve current signal, conduction valve current signal and tag of various historical commutation events, including various normal operating conditions and various commutation failure scenarios; Based on the exit valve current signal and the on valve current signal of each historical commutation event, determine the exit valve bounce rate and the on valve rise time of each historical commutation event. Using the least squares method, multiple coefficients of a pre-set second-order polynomial regression equation are determined based on the exit valve rebound rate, conduction valve rise time, and labels of various historical commutation events. Substitute multiple coefficients of the preset second-order polynomial regression equation into the preset second-order polynomial regression equation to obtain the target second-order polynomial regression equation. Based on the target second-order polynomial regression equation, determine the parabola criterion parameters; Based on the parabola criterion parameters, the preset parabola criterion equation is constructed.

[0010] Optionally, the parabola criterion parameters include the x-coordinate of the vertex, the y-coordinate of the vertex, and the opening coefficient of the parabola. Determining the parabola criterion parameters based on the target second-order polynomial regression equation includes: By eliminating cross-product terms and completing the square on the target second-order polynomial regression equation, a parabolic equation is obtained. Based on the equation of the parabola, determine the x-coordinate of the vertex, the y-coordinate of the vertex, and the opening coefficient of the parabola.

[0011] Optionally, the method further includes: Acquire the exit valve current signal, conduction valve current signal and tag of various historical commutation events, including various normal operating conditions and various commutation failure scenarios; Based on the exit valve current signal and the on valve current signal of each historical commutation event, determine the exit valve bounce rate and the on valve rise time of each historical commutation event. The exit valve rebound rate, conduction valve rise time and labels of various historical commutation events are input into the initial classification prediction model for training, and the target classification prediction model is obtained. Based on the model parameters of the target classification prediction model, determine the parabolic criterion parameters; Based on the parabola criterion parameters, the preset parabola criterion equation is constructed.

[0012] Optionally, the model parameters of the target classification prediction model include a support vector set, a label set corresponding to the support vector set, a dual coefficient set corresponding to the support vector set, and a bias term; the parabolic criterion parameters include the x-coordinate of the vertex, the y-coordinate of the vertex, and the opening coefficient of the parabola; determining the parabolic criterion parameters based on the model parameters of the target classification prediction model includes: Substituting the support vector set, the label set corresponding to the support vector set, the dual coefficient set corresponding to the support vector set, and the bias term into the decision function of the target classification prediction model, a second-order polynomial equation is obtained. Eliminating cross-product terms and completing the square of the second-order polynomial equation yields the parabolic equation; Based on the equation of the parabola, determine the x-coordinate of the vertex, the y-coordinate of the vertex, and the opening coefficient of the parabola.

[0013] Optionally, the preset parabola criterion equation is: ; in, The opening coefficient of the parabola in the parabola criterion parameters. The x-coordinate of the vertex of the parabola in the parabola criterion parameters is... The ordinate of the vertex of the parabola in the parabola criterion parameters is... The rise time of the valve is the opening time. For the exit valve rebound rate, This is the criterion value for a parabola.

[0014] Optionally, determining the commutation detection result of the current commutation event to be tested based on the parabolic criterion value includes: If the parabolic criterion value is less than the criterion threshold, the commutation detection result of the current commutation event to be tested is determined to be a commutation failure; If the parabolic criterion value is greater than or equal to the criterion threshold, the commutation detection result of the current commutation event to be tested is determined to be a successful commutation.

[0015] To achieve the above objectives, the present invention provides a commutation detection device in a second aspect, the device comprising: The acquisition module is used to acquire the exit valve current signal and the on valve current signal of the current commutation event to be tested. The quantization parameter determination module is used to determine the exit valve rebound rate and conduction valve rise time of the current commutation event under test based on the exit valve current signal and conduction valve current signal of the current commutation event under test. The calculation module is used to substitute the exit valve rebound rate and the conduction valve rise time of the current commutation event under test into the preset parabolic criterion equation to obtain the parabolic criterion value. The detection result determination module is used to determine the commutation detection result of the current commutation event to be tested based on the parabolic criterion value.

[0016] To achieve the above objectives, the present invention provides, in a third aspect, a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the commutation detection method as described in any one of the first aspects.

[0017] To achieve the above objectives, the present invention provides a computer device in a fourth aspect, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor causes the processor to perform the commutation detection method as described in any one of the first aspects.

[0018] The present invention provides the following advantages: The method acquires the exit valve current signal and the on-valve current signal of the current commutation event under test. Then, based on these signals, it determines the exit valve rebound rate and the on-valve rise time. These values ​​are then substituted into a preset parabolic criterion equation to obtain the parabolic criterion value. Finally, the commutation detection result of the current commutation event is determined based on the parabolic criterion value. In other words, through the collaborative analysis of the dual-valve current signals, the exit valve rebound during the commutation process is achieved. Dynamic dual-feature extraction and precise quantitative characterization of rate and conduction valve rise time, utilizing the geometric constraint characteristics of the preset parabolic criterion equation, transforms the commutation state judgment into a geometric relationship judgment between the criterion value and the threshold boundary. This achieves real-time online detection of the commutation process and accurate identification of fault states, overcoming the delay defect of the arc extinction angle measurement method and avoiding the misjudgment risk of the AC voltage prediction method. At the same time, it improves the robustness of the single current feature method. Ultimately, it significantly improves real-time performance while ensuring detection accuracy, effectively enhancing the detection reliability of the commutation process in the high-voltage direct current transmission system and strengthening the stability and safety of system operation. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] in: Figure 1 This is a schematic diagram of a commutation detection method according to an embodiment of this application; Figure 2 This is a schematic diagram of a commutation detection device according to an embodiment of this application; Figure 3 This is a diagram showing the internal structure of a computer device in some embodiments. Detailed Implementation

[0021] 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 only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] In high-voltage direct current (HVDC) transmission systems based on grid-commutated converters, the commutation process of the six-pulse converter bridge is a crucial step in smoothly transferring DC current from the de-energizing valve (about to be turned off) to the energizing valve (about to be turned on). The driving force originates from the AC line voltage between the corresponding phases of the two valves. An ideal commutation process requires the de-energizing valve current to smoothly decay from its rated value to zero, while the energizing valve current synchronously and smoothly rises to its rated value, ultimately achieving a seamless transfer of DC current. This is essential for the stable operation of the system.

[0023] However, existing commutation detection technologies have significant limitations: While measurement methods based on the arc extinction angle can determine commutation results through valve voltage or zero-current time, data acquisition is required after commutation, resulting in inherent delays and a post-hoc diagnostic approach that fails to meet real-time requirements. Predictive methods based on AC voltage provide rapid early warning by monitoring precursors such as voltage dips, but the correlation between voltage disturbances and commutation failure is affected by fault type, location, and system operating status, making them prone to misjudgment or missed detection due to non-fault disturbances. Methods based on single current characteristics directly monitor current change rate or waveform distortion, but threshold setting is difficult and they are susceptible to noise and non-fault disturbances, resulting in insufficient robustness. These technical bottlenecks lead to a triple challenge for commutation detection: delay, misjudgment, and insufficient robustness. It is difficult to balance accuracy and real-time performance, becoming a key factor restricting the reliable operation of high-voltage direct current transmission systems.

[0024] To address the aforementioned issues, this application proposes a commutation detection method that enables real-time online detection of the commutation process and accurate identification of fault states. This method overcomes the delay inherent in the arc extinction angle measurement method and avoids the misjudgment risk associated with the AC voltage prediction method. Furthermore, it enhances the robustness of the single current characteristic method. Ultimately, while ensuring detection accuracy, it significantly improves real-time performance, effectively enhancing the reliability of commutation detection in high-voltage direct current transmission systems and strengthening the stability and safety of system operation. The specific implementation principle will be detailed in the following embodiments.

[0025] This application provides a commutation detection method in its first aspect.

[0026] Please see Figure 1 The diagram below illustrates a commutation detection method according to an embodiment of this application. The method includes: Step 110: Obtain the exit valve current signal and the on valve current signal of the current commutation event to be tested.

[0027] Among them, the current commutation event to be tested refers to the commutation event between the exit valve and the on valve that are currently undergoing commutation, and this commutation event needs to be detected as either a commutation failure or a success.

[0028] Regarding the acquisition method of valve current signal, in some embodiments, when a commutation event occurs, the current signals of the exit valve and the conduction valve corresponding to the commutation event can be collected simultaneously to obtain the exit valve current signal and the conduction valve current signal of the current commutation event to be tested.

[0029] It should be noted that the occurrence of a commutation event is essentially an interaction between the exit valve and the on valve. Therefore, detecting whether the commutation has failed or succeeded is not an isolated behavior of the exit valve. When commutation fails, the current signal of both the exit valve and the on valve will leave a corresponding distortion mark. In this regard, this application performs commutation detection by synchronously acquiring the current signals of the exit valve and the on valve corresponding to the commutation event. Compared with existing commutation detection methods, this can improve the reliability and robustness of the detection.

[0030] It should be further noted that, since this application performs commutation detection by synchronously acquiring the current signals of the exit valve and the on valve corresponding to the commutation event, this joint analysis mechanism of dual valve current signals naturally has common-mode noise suppression capability and higher recognition. Therefore, even if one of the valve current signals is affected by noise or slight disturbance, the reliability and robustness of the detection can be ensured.

[0031] Step 120: Based on the exit valve current signal and the conduction valve current signal of the current commutation event to be tested, determine the exit valve rebound rate and conduction valve rise time of the current commutation event to be tested.

[0032] It should be noted that, for the exit valve, if commutation fails when the exit valve exits, the exit valve current signal will rebound at the moment of exit. The rebounded current is equivalent to applying a reverse electromotive force in the commutation circuit, which will hinder the rise of the conduction valve current signal. Therefore, for the conduction valve, if commutation fails when the conduction valve is turned on, the current of the conduction valve current signal at the moment of conduction will be hindered, and its rise will become abnormally slow or even stop.

[0033] Based on the above principle, in some embodiments, the exit valve rebound rate of the current test commutation event can be determined by the current value of the exit valve current signal of the current test commutation event when it exits, and the conduction valve rise time of the current test commutation event can be determined by the current value of the conduction valve current signal of the current test commutation event when it conducts.

[0034] Step 130: Substitute the exit valve rebound rate and the conduction valve rise time of the current commutation event under test into the preset parabolic criterion equation to obtain the parabolic criterion value.

[0035] The preset parabola criterion equation can be obtained and preset by the operator based on extensive experience, experiments, or statistics. Alternatively, it can be preset by the operator according to actual needs. In some embodiments, the determination of the preset parabolic criterion equation can be achieved by acquiring the exit valve current signal and the conduction valve current signal of various historical commutation events. Then, for each historical commutation event, the exit valve rebound rate and conduction valve rise time are determined. Parabolic fitting analysis is then performed on the exit valve rebound rate and conduction valve rise time of various historical commutation events to obtain the fixed parameters of the parabola. The fixed parameters of the parabola are then substituted into the parabolic equation to obtain the preset parabolic criterion equation. The various historical commutation events include various normal operating conditions (i.e., non-commutation failure conditions) and various commutation failure scenarios (i.e., commutation failure scenarios). The fixed parameters include the abscissa, ordinate, and opening coefficient of the parabola's vertex.

[0036] Step 140: Determine the commutation detection result of the current commutation event to be tested based on the parabolic criterion value.

[0037] It should be noted that since the preset parabolic criterion equation is determined in advance based on the exit valve current signal and the conduction valve current signal under various normal operating conditions and various commutation failure scenarios, the preset parabolic criterion equation is equivalent to using geometry to constrain the boundary between all commutation detection results being commutation failure and all commutation detection results being commutation success. That is, outside the parabola on one side of the boundary, all commutation detection results are commutation failure, while on the parabola on the other side of the boundary or inside, all commutation detection results are commutation success. When the commutation detection result is commutation failure, the point corresponding to the conduction valve rise time and the exit valve rebound rate should fall outside the parabola on one side of the boundary, while when the commutation detection result is commutation success, the point corresponding to the conduction valve rise time and the exit valve rebound rate should fall on the parabola on the other side of the boundary or inside.

[0038] Based on the above principle, in some embodiments, the commutation detection result of the current commutation event to be tested can be determined by comparing the parabolic criterion value with the preset threshold. The preset threshold can be obtained and set in advance by the operator based on a large amount of experience, experiments or statistics. Of course, it can also be set in advance by the operator according to actual needs.

[0039] In this embodiment, through the collaborative analysis of dual-valve current signals, dynamic dual-feature extraction and precise quantitative characterization of the commutation process exit valve rebound rate and conduction valve rise time are achieved. Utilizing the geometric constraint characteristics of the preset parabolic criterion equation, the commutation state judgment is transformed into a geometric relationship judgment between the criterion value and the threshold boundary. This enables real-time online detection of the commutation process and precise identification of fault states. It overcomes the delay defect of the arc extinction angle measurement method and avoids the misjudgment risk of the AC voltage prediction method. At the same time, it improves the robustness of the single current feature method. Ultimately, while ensuring detection accuracy, it significantly improves real-time performance, effectively enhances the detection reliability of the commutation process in the high-voltage direct current transmission system, and strengthens the stability and safety of system operation.

[0040] In addition to the advantages mentioned above, this commutation detection method also has the following beneficial effects: Optimized maintenance plans: Through long-term real-time monitoring and data analysis of the commutation process, the stability and reliability of the system's commutation under different operating conditions can be understood. Based on this information, more scientific and reasonable maintenance plans can be formulated, such as arranging in advance the repair or replacement of converter valves prone to commutation problems, avoiding more serious system failures caused by commutation failures, and improving the overall availability of the system; Reduced power fluctuations: Accurate commutation detection can promptly identify commutation failures. The system can quickly adjust its control strategy based on the detection results, reducing DC power fluctuations caused by commutation failures. Stable DC power output helps improve the power quality of the grid, reduces the impact on connected AC grids, and improves the overall operating efficiency of the power system; Optimized transmission capacity: Reliable commutation detection can... Ensuring the system operates within a safe range and fully utilizes its transmission capacity is crucial. When commutation stability is detected, the system can appropriately increase transmission power to fully leverage the advantages of high-voltage direct current (HVDC) transmission systems in terms of large capacity and long-distance transmission, thereby improving energy utilization efficiency. Reducing operating costs is also essential: improving the accuracy and real-time performance of commutation detection reduces commutation failures, thus lowering operating costs such as equipment wear and power outages. Optimized maintenance plans and operating strategies also reduce unnecessary equipment repair and replacement costs, improving the system's economic efficiency. Furthermore, a stable and reliable HVDC transmission system contributes to improved energy trading efficiency and effectiveness. Accurate commutation detection ensures the system completes energy transmission tasks according to plan, reducing the risk of energy trading defaults due to system failures and enhancing the stability and liquidity of the energy market.

[0041] In one feasible implementation, step 120 in the above embodiment, determining the exit valve rebound rate and conduction valve rise time of the current commutation event under test based on the exit valve current signal and conduction valve current signal of the current commutation event under test, includes: determining the current value of all times in the exit valve current signal of the current commutation event under test from the minimum value time corresponding to the minimum current value to the end commutation time, and determining the exit valve rebound rate of the current commutation event under test based on the current value of all times in the period from the minimum value time to the end commutation time; determining the first time corresponding to the current value equal to the first current value and the second time corresponding to the current value equal to the second current value in the conduction valve current signal of the current commutation event under test after the commutation start time, and determining the conduction valve rise time of the current commutation event under test based on the first time and the second time, wherein the first current value is a first preset percentage of the preset DC current value, the second preset current value is a second preset percentage of the preset DC current value, and the first preset percentage is less than the second preset percentage.

[0042] The preset DC current value, the first preset percentage, and the second preset percentage can all be preset by the operator based on extensive experience, experiments, or statistics. Of course, they can also be preset by the operator according to actual needs.

[0043] Regarding the selection of the preset DC current value, in some embodiments, the steady-state value of the DC current in the six-pulse converter bridge of the high-voltage DC transmission system based on the grid commutation converter can be used as the preset DC current value.

[0044] Regarding the values ​​of the first preset percentage and the second preset percentage, in some embodiments, this application preferably sets the first preset percentage to 10% and the second preset percentage to 90%.

[0045] In this embodiment, by accurately extracting dynamic dual feature parameters (exit valve rebound rate and conduction valve rise time) and combining them with the geometric constraint characteristics of the preset parabolic criterion equation, real-time and accurate determination of commutation state is achieved, while improving the robustness and anti-interference capability of detection.

[0046] Understandably, the rebound rate quantification optimization involves analyzing the current value of the exit valve current signal from its minimum value to the end of commutation. This accurately captures the dynamic characteristics of current rebound during commutation failure. This method avoids the sensitivity to noise inherent in traditional single-threshold detection. By comprehensively analyzing current values ​​at multiple times, it improves the stability of rebound rate calculation and effectively suppresses the impact of common-mode noise on the detection results. The dynamic extraction of rise time involves setting 10% (first current value) and 90% (second current value) of the preset DC current value as key thresholds. This dynamically calculates the time required for the conduction valve current to rise from 10% to 90% of its rated value. This dual-percentage threshold design balances the rapid response in the initial stage of commutation with the steady-state characteristics in the later stages. Compared to the fixed threshold method, it more sensitively reflects the rise stagnation phenomenon caused by commutation failure, while reducing the impact of current... Risk of misjudgment caused by transient fluctuations; Dual-feature synergistic anti-interference: The joint analysis of the exit valve rebound rate and the conduction valve rise time forms a complementary feature set. When the current signal of a single valve is interfered with by noise, the feature parameters of the other valve can still provide a reliable basis for judgment. For example, the rebound of the exit valve current may be partially masked by noise, but the abnormal extension of the conduction valve rise time can still clearly indicate commutation failure. This dual-valve synergistic mechanism significantly improves the robustness of the detection system under complex working conditions; Geometric criteria simplify decision-making: Substituting the dual feature parameters into the preset parabolic criterion equation, the commutation state is directly determined by the geometric positional relationship between the criterion value and the threshold (inside / outside the parabola), avoiding complex logic judgment of multiple parameters. This method not only simplifies the decision-making process, but also naturally distinguishes between normal commutation and fault scenarios through the geometric constraint characteristics of the parabolic boundary, further reducing the misjudgment rate.

[0047] In one feasible implementation, the exit valve rebound rate of the current commutation event under test is determined using the following formula: ; in, The exit valve rebound rate is the current commutation event to be tested. To find the maximum value function, This is the derivative of the current value at time t with respect to time t during the period from the moment of minimum value to the moment of end of commutation. At the moment of minimum value, This is to mark the end of the commutation phase.

[0048] In this embodiment, the formula achieves accurate quantification of commutation failure characteristics by dynamically capturing the runaway recovery rate of the exit valve current, significantly improving the stability and anti-interference capability of the rebound rate calculation.

[0049] Understandably, the following features are employed: Dynamic feature extraction: By calculating the instantaneous rate of change (derivative) of current at each moment from the minimum point to the end of commutation, the dynamic process of current rebound is fully characterized, avoiding the sensitivity of traditional single-threshold detection to transient noise; Maximum value quantization mechanism: Taking the maximum value of the derivative at each moment as the rebound rate index, the most severe stage of current runaway recovery when commutation fails is accurately captured, effectively suppressing the influence of common-mode noise on the detection results; Enhanced anti-interference capability: Compared with single-point threshold detection, the comprehensive analysis of derivatives at multiple moments significantly reduces the risk of misjudgment caused by current transient fluctuations or local interference, and improves the stability of feature parameters; Fault feature enhancement: When commutation failure occurs, the instantaneous rate of change of current rebound will increase significantly. This formula quantifies this feature, forming a clear distinction from the normal commutation process, providing a reliable basis for subsequent judgment.

[0050] In one feasible implementation, determining the rise time of the conduction valve of the current commutation event to be tested based on the first time and the second time in the above embodiments includes: taking the difference between the second time and the first time as the rise time of the conduction valve of the current commutation event to be tested.

[0051] In some embodiments, the absolute value of the difference between the first moment and the second moment can be used as the rise time of the current commutation event to be measured for determining the valve rise time.

[0052] In this embodiment, the method for determining the rise time of the conduction valve is simple and efficient, and can accurately reflect the current rise characteristics of the conduction valve during the commutation process, effectively assisting in commutation detection.

[0053] Understandably, by setting a specific percentage (such as 10% and 90%) of the preset DC current value as a key threshold, the difference between the second moment (when the current reaches 90% of the preset DC current value) and the first moment (when the current reaches 10% of the preset DC current value) is used as the rise time of the conduction valve. This method, with its dynamic dual-percentage threshold design, takes into account both the rapid response in the early stage of commutation and the steady-state characteristics in the middle and later stages. Compared with the fixed threshold method, it can more sensitively reflect the rise stagnation phenomenon caused by commutation failure. At the same time, the calculation process is simple and direct, requiring only the acquisition of two key moments and the calculation of the difference, avoiding complex calculations, reducing the possibility of calculation errors, and quickly and accurately obtaining the rise time of the conduction valve, providing a reliable basis for subsequent commutation detection, and improving detection efficiency and accuracy.

[0054] In one feasible implementation, the method in the above embodiments further includes: acquiring the exit valve current signal, conduction valve current signal, and tags for various historical commutation events, wherein the various historical commutation events include various normal operating conditions and various commutation failure scenarios; determining the exit valve rebound rate and conduction valve rise time for each historical commutation event based on the exit valve current signal and conduction valve current signal for each historical commutation event; using the least squares method, determining multiple coefficients of a preset second-order polynomial regression equation based on the exit valve rebound rate, conduction valve rise time, and tags for various historical commutation events; substituting the multiple coefficients of the preset second-order polynomial regression equation into the preset second-order polynomial regression equation to obtain a target second-order polynomial regression equation; determining parabolic criterion parameters based on the target second-order polynomial regression equation; and constructing a preset parabolic criterion equation based on the parabolic criterion parameters.

[0055] Among them, various historical commutation events refer to various normal operating conditions (i.e., non-commutation failure conditions) and various commutation failure scenarios (i.e., commutation failure scenarios); the labels here are used to distinguish between normal operating conditions and commutation failure scenarios; the parabola criterion parameters refer to the various fixed parameters of the parabola, such as the x-coordinate of the vertex, the y-coordinate of the vertex, and the opening coefficient of the parabola.

[0056] In some embodiments, the least squares method can be used to determine the preset second-order polynomial regression equation. The exit valve rebound rate, conduction valve rise time and label of various historical commutation events are substituted into the preset objective function for coefficient optimization to obtain multiple coefficients of the preset second-order polynomial regression equation.

[0057] Furthermore, in some embodiments, the expression for the preset objective function is: ; in, The function value of the preset objective function, To find the minimum value function, The total number of historical phase-change events. These are the labels for the m-th historical commutation event. to This refers to the first to sixth coefficients of the pre-defined second-order polynomial regression equation. Let be the rise time of the conduction valve for the m-th historical commutation event. Let be the exit valve bounce rate for the m-th historical commutation event.

[0058] In some embodiments, the expression for the predefined second-order polynomial regression equation is: ; in, to This refers to the first to sixth coefficients of the pre-defined second-order polynomial regression equation. and All are variables (in the following, The rise time of the valve is the opening time. (Exit valve rebound rate).

[0059] In this embodiment, by constructing a second-order multinomial regression model based on historical data, the parabolic criterion equation is accurately parameterized, which significantly improves the accuracy and adaptability of boundary determination in commutation detection.

[0060] Understandably, the data-driven criterion optimization involves: collecting historical data covering both normal operation and commutation failure scenarios, combining label classification, and using the least squares method to fit a second-order polynomial regression equation. This allows the criterion equation to dynamically adapt to different operating conditions, avoiding the limitations of empirically set parameters and improving the accuracy of boundary judgment. Enhanced robustness of geometric constraints: based on the parabolic parameters (vertex coordinates, opening coefficient) extracted from the target second-order polynomial regression equation, the constructed criterion equation clearly distinguishes between successful and failed commutation regions through geometric boundaries. This nonlinear constraint mechanism, compared to linear thresholds, better reflects the actual current distribution characteristics, effectively reducing the risk of misjudgment. Balance between computational efficiency and practicality: the second-order polynomial regression equation ensures sufficient fitting accuracy while avoiding the overfitting risk of higher-order models. Furthermore, its low computational complexity facilitates real-time implementation, balancing detection accuracy with engineering practicality.

[0061] In one feasible implementation, the parabola criterion parameters in the above embodiments include the x-coordinate of the vertex, the y-coordinate of the vertex, and the opening coefficient of the parabola.

[0062] In the above embodiments, determining the parabola criterion parameters based on the target second-order polynomial regression equation includes: eliminating cross-product terms and completing the square of the target second-order polynomial regression equation to obtain the parabola equation; and determining the x-coordinate, y-coordinate, and opening coefficient of the parabola based on the parabola equation.

[0063] It should be noted that, in the process of eliminating cross-product terms and completing the formula for the target second-order polynomial regression equation, this application preferably rearranges the parabolic equation to a value with an opening coefficient greater than 0.

[0064] In this embodiment, the parabolic geometric parameters are accurately extracted through mathematical transformation, providing a geometric boundary with clear physical meaning for commutation state determination, which significantly improves the adaptability and reliability of the criterion equation.

[0065] Understandably, the physical meaning of the parameters is clarified: through cross-product elimination and the completing the square process, the second-order polynomial regression equation is transformed into a standard parabolic form, so that the vertex coordinates (horizontal and vertical coordinates) and the opening coefficient directly correspond to the current characteristic distribution law of the commutation process. Among them, the vertex coordinates reflect the critical point between normal commutation and fault state, and the opening coefficient characterizes the degree of separation between the two states. The robustness of boundary judgment is enhanced: the opening coefficient of the parabolic equation after the completing the square process is forcibly set to a positive value to ensure that the geometric boundary is an upward-opening parabola. This constraint makes the division of the successful commutation region (on the parabola and inside the parabola) and the failed region (outside the parabola) more consistent with the actual distribution of current characteristics. The system effectively avoids boundary ambiguity caused by parameter fluctuations; the calculation process is standardized: through a unified mathematical transformation process (elimination of cross-product terms, formula processing, and parameter extraction), the interference of cross terms in the original regression equation on the geometric shape is eliminated, making the calculation of vertex coordinates and opening coefficients repeatable and engineering practical, providing a stable basis for real-time criterion calculation; the adaptability to operating conditions is improved: the parabolic parameters obtained by fitting historical data can automatically adapt to the changes in current characteristics under different operating conditions after formula processing, so that the geometric boundary always maintains a dynamic match with the actual commutation process, which has stronger environmental adaptability compared with the fixed threshold method.

[0066] In one feasible implementation, the method in the above embodiments further includes: acquiring the exit valve current signal, conduction valve current signal, and tags for various historical commutation events, wherein the various historical commutation events include various normal operating conditions and various commutation failure scenarios; determining the exit valve rebound rate and conduction valve rise time for each historical commutation event based on the exit valve current signal and conduction valve current signal for each historical commutation event; inputting the exit valve rebound rate, conduction valve rise time, and tags for various historical commutation events into an initial classification prediction model for training to obtain a target classification prediction model; determining parabolic criterion parameters based on the model parameters of the target classification prediction model; and constructing a preset parabolic criterion equation based on the parabolic criterion parameters.

[0067] Here, the target classification prediction model refers to a pre-trained model that can output commutation detection results based on the input exit valve rebound rate and conduction valve rise time.

[0068] Regarding the selection of the initial classification prediction model, in some embodiments, a Support Vector Machine (SVM) with a second-order multinomial kernel function can be used as the initial classification prediction model.

[0069] In some embodiments, the pre-defined parabola criterion equation can be constructed by substituting each parameter in the parabola criterion parameters into the parabola equation.

[0070] In this embodiment, the dynamic adaptive adjustment of the commutation detection boundary is achieved through criterion optimization driven by a machine learning model, which significantly improves the adaptability of the criterion equation to complex working conditions and the detection accuracy.

[0071] Understandably, the data-driven criterion optimization involves using dual-valve current characteristics (rebound rate, rise time) and labeled data from historical commutation events. A support vector machine model learns the classification boundary between normal and fault states, enabling the criterion parameters to dynamically adapt to different operating conditions (such as load changes and system disturbances), avoiding the sensitivity of traditional empirical thresholds to changes in operating conditions. The geometricization of nonlinear criteria utilizes a second-order polynomial kernel support vector machine to capture the complex nonlinear relationship between current characteristics and commutation state. Parabolic criterion parameters (vertex coordinates, opening coefficient) are obtained through model parameter transformation, explicitly converting the classification boundary into geometric constraints, thus enabling the determination of commutation success / failure. It better conforms to the physical distribution law of current characteristics; enhanced anti-interference ability: during the model training stage, it automatically learns the difference between noise and fault characteristics, and suppresses the influence of common-mode noise on feature parameters through kernel function mapping. For example, when the current of a single valve is disturbed by noise, the model can still maintain the discrimination stability through joint analysis of dual-valve features, which significantly improves robustness compared with the fixed threshold method; balance between computational efficiency and practicality: the support vector machine model implicitly processes high-dimensional features through kernel functions, avoiding the high computational complexity of explicit feature engineering. After training, only parabolic parameters need to be stored. During real-time detection, only dual-valve features need to be calculated and substituted into the parabolic equation, which balances detection accuracy and engineering feasibility.

[0072] In one feasible implementation, the model parameters of the target classification prediction model in the above embodiments include a support vector set, a label set corresponding to the support vector set, a dual coefficient set corresponding to the support vector set, and a bias term. The parabola criterion parameters include the x-coordinate of the vertex of the parabola, the y-coordinate of the vertex, and the opening coefficient.

[0073] In the above embodiments, determining the parabola criterion parameters based on the model parameters of the target classification prediction model includes: substituting the support vector set, the label set corresponding to the support vector set, the dual coefficient set corresponding to the support vector set, and the bias term into the decision function of the target classification prediction model to obtain a second-order polynomial equation; performing cross-product term elimination and completing the square on the second-order polynomial equation to obtain the parabola equation; and determining the x-coordinate of the vertex, y-coordinate of the vertex, and opening coefficient of the parabola based on the parabola equation.

[0074] In some embodiments, the following decision function can be used to determine the second-order polynomial equation: ; in, Let be the decision function. For the nth dual coefficient in the set of dual coefficients, For the nth tag in the tag set, It is a second-order polynomial kernel function. For bias terms, The dot product symbol is a vector symbol. The first support vector in the support vector set. This is the second support vector in the support vector set.

[0075] It should be noted that, since the model in this application is a model with a second-order multinomial kernel function, the total number of its support vector set, label set, and dual coefficient set is 2, and the total number of elements in the support vector is also 2.

[0076] It should be further noted that, in the process of simplifying the above-mentioned determination of the second-order polynomial equation, we can first... Substitution In the middle, we get Then Substitution In the middle, and ordered , that is , and then and They are respectively and ,get ,Will And substitute elements from dual coefficients, labels, bias terms, and support vectors. And combine like terms to obtain and For variables (in the following, The rise time of the valve is the opening time. The second-order polynomial equation is given by the exit valve rebound rate; where, For the first element of the nth support vector in the support vector set, It is the second element of the nth support vector in the support vector set.

[0077] In this embodiment, the dynamic adaptive optimization of the commutation detection criterion is achieved through the deep integration of machine learning models and geometric constraints, which significantly improves the detection accuracy and robustness under complex working conditions.

[0078] Understandably, the data-driven dynamic boundary generation utilizes a support vector machine model to learn the mapping relationship between the dual-valve current characteristics (rebound rate, rise time) and commutation state in historical commutation events. Second-order polynomial equations are directly generated through model parameters (support vectors, dual coefficients, etc.), avoiding the subjectivity of manually setting thresholds and allowing the criterion boundary to automatically adjust with the system's operating state. Geometric explicitness of nonlinear criteria involves transforming the second-order polynomial equations output by the model into parabolic equations through cross-product elimination and completing the square, making the classification boundary explicit as geometric constraints (vertex coordinates, opening coefficients). This transformation not only preserves the model's adaptability to complex operating conditions but also simplifies the physical interpretation of the criterion through geometric intuition. Enhanced anti-interference capability is achieved through implicitly processing the interaction between features using a second-order polynomial kernel function during the model training phase. High-order interaction automatically suppresses the impact of common-mode noise on single features. For example, when the exit valve current is disturbed by noise, the model can still maintain discrimination stability through joint analysis of the valve rise time and bounce rate, which significantly improves robustness compared to the fixed threshold method. Balance between computational efficiency and accuracy: The support vector machine model maps high-dimensional features to low-dimensional space through kernel tricks, avoiding the high computational complexity of explicit feature engineering. After training, only parabolic parameters need to be stored, balancing detection speed and accuracy. Clarify the physical meaning of parameters: The opening coefficient of the parabolic equation after the formula is forcibly set to a positive value, ensuring that the geometric boundary is an upward-opening parabola. This makes the division between the commutation success region (on and inside the parabola) and the failure region (outside) more consistent with the actual distribution law of current characteristics, effectively avoiding the boundary ambiguity problem caused by parameter fluctuations.

[0079] In one feasible implementation, the preset parabolic criterion equation in the above embodiments is: ; in, The opening coefficient of the parabola in the parabola criterion parameters. The x-coordinate of the vertex of the parabola in the parabola criterion parameters. The ordinate of the vertex of the parabola in the parabola criterion parameters. The rise time of the valve is the opening time. For the exit valve rebound rate, This is the criterion value for a parabola.

[0080] In this embodiment, by explicitly constructing the parabolic equation, the geometric intuitiveness and computational efficiency of the commutation state determination are achieved, which significantly improves the real-time response capability and boundary determination accuracy of the detection system.

[0081] Understandably, the physical explicitness of geometric constraints involves transforming the criterion equation into a parabolic form, allowing the opening coefficient and vertex coordinates to directly correspond to the current characteristic distribution law of the commutation process. The vertex coordinates clearly define the critical point between normal commutation and fault states, while the opening coefficient quantifies the degree of separation between the two states. This geometric explicitness avoids complex logical judgments, giving the criterion a clear physical meaning. Optimized computational efficiency is achieved because the parabolic equation only requires basic operations to calculate the criterion value. Compared to higher-order polynomials or neural network models, the computational complexity is greatly reduced, ensuring the feasibility of real-time detection and meeting the millisecond-level response requirements of high-voltage direct current transmission systems. Enhanced robustness of boundary judgment is achieved through the forced setting of the opening coefficient. This design ensures that the geometric boundary is an upward-opening parabola, making the division between the successful commutation region (on and inside the parabola) and the failed region (outside the parabola) more consistent with the actual distribution of current characteristics. Improved anti-interference capability: The quadratic characteristics of the parabolic equation give it a natural ability to suppress noise. When there are transient fluctuations in the current signal, the change in the criterion value is very small, while the fluctuation amplitude of the linear threshold method is relatively high, significantly improving detection stability. Ease of engineering implementation: The parabolic equation only needs to store 3 parameters, greatly reducing memory usage compared to the support vector machine model (which needs to store support vectors, dual coefficients, etc.), and it does not require complex matrix operations, making it particularly suitable for hardware implementation in embedded systems.

[0082] In one feasible implementation, step 140 in the above embodiment, which determines the commutation detection result of the current commutation event to be tested based on the parabolic criterion value, includes: determining the commutation detection result of the current commutation event to be tested as commutation failure when the parabolic criterion value is less than the criterion threshold; and determining the commutation detection result of the current commutation event to be tested as commutation success when the parabolic criterion value is greater than or equal to the criterion threshold.

[0083] The criterion threshold can be obtained and preset by the operator based on a large amount of experience, experiments or statistics. Of course, it can also be preset by the operator according to actual needs.

[0084] Regarding the value of the criterion threshold, in some embodiments, this application preferably sets the criterion threshold to 0.

[0085] In this embodiment, by directly comparing the parabolic criterion value with the threshold, the commutation state is quickly and accurately determined, which significantly improves the real-time performance and reliability of the detection system.

[0086] Understandably, the judgment logic is simple and efficient: after substituting the complex dual-feature parameters (exit valve rebound rate, conduction valve rise time) into the preset parabolic equation, the commutation status can be determined simply by comparing it with a single threshold (preferably 0). This simplification avoids complex multi-parameter logic judgments, reducing the judgment process from dozens of steps in the traditional method to two steps (calculating the criterion value and comparing the threshold), and shortening the response time to the microsecond level, meeting the millisecond-level real-time requirements of the high-voltage direct current transmission system. The boundary division is physically clear: through the geometric characteristics of the parabolic equation, the boundary between successful and failed commutation is explicitly represented as a critical point (criterion threshold) on the parabola. When the criterion value is less than 0, the corresponding point is located outside the parabola, and the commutation is judged as failed. When the criterion value is greater than or equal to 0, the corresponding point is located on or inside the parabola, and the commutation is judged as successful. This geometric division is highly consistent with the actual distribution law of current characteristics, and its judgment accuracy is greatly improved compared with the traditional linear threshold method.

[0087] In a second aspect, this application provides a commutation detection device.

[0088] Please see Figure 2 This is a schematic diagram of a commutation detection device according to an embodiment of this application. The device 210 includes: The acquisition module 211 is used to acquire the exit valve current signal and the conduction valve current signal of the current commutation event to be tested. The quantization parameter determination module 212 is used to determine the exit valve rebound rate and the conduction valve rise time of the current commutation event under test based on the exit valve current signal and the conduction valve current signal of the current commutation event under test. Substitute the values ​​into the calculation module 213 to input the exit valve rebound rate and the conduction valve rise time of the current commutation event to be tested into the preset parabolic criterion equation to obtain the parabolic criterion value. The detection result determination module 214 is used to determine the commutation detection result of the current commutation event to be tested based on the parabolic criterion value.

[0089] In this embodiment of the application, the relevant contents of the above-mentioned acquisition module 211, quantization parameter determination module 212, substitution calculation module 213 and detection result determination module 214 can be found in the following references. Figure 1 The contents of the illustrated embodiments will not be repeated here.

[0090] It should be noted that the device 210 of this application also includes other modules. It can be understood that the method of this application and the device 210 have a one-to-one correspondence. Therefore, the other modules of the device 210 of this application are the contents corresponding to the method of this application in the above embodiments.

[0091] In this embodiment, through the collaborative analysis of dual-valve current signals, dynamic dual-feature extraction and precise quantitative characterization of the commutation process exit valve rebound rate and conduction valve rise time are achieved. Utilizing the geometric constraint characteristics of the preset parabolic criterion equation, the commutation state judgment is transformed into a geometric relationship judgment between the criterion value and the threshold boundary, realizing real-time online detection of the commutation process and precise identification of fault states. This overcomes the delay defects of the arc extinction angle measurement device and avoids the misjudgment risk of the AC voltage prediction device, while also improving the robustness of the single current feature device. Ultimately, while ensuring detection accuracy, real-time performance is significantly improved, effectively enhancing the detection reliability of the commutation process in the high-voltage direct current transmission system and strengthening the stability and safety of system operation.

[0092] In addition to the advantages mentioned above, this commutation detection device also has the following beneficial effects: Optimized maintenance plans: Through long-term real-time monitoring and data analysis of the commutation process, the stability and reliability of the commutation under different operating conditions can be understood. Based on this information, more scientific and reasonable maintenance plans can be formulated, such as arranging in advance the repair or replacement of converter valves prone to commutation problems, avoiding more serious system failures caused by commutation failure, and improving the overall availability of the system; Reduced power fluctuations: Accurate commutation detection can promptly identify commutation failures. The system can quickly adjust its control strategy based on the detection results, reducing DC power fluctuations caused by commutation failures. Stable DC power output helps improve the power quality of the grid, reduces the impact on the connected AC grid, and improves the overall operating efficiency of the power system; Optimized transmission capacity: Reliable commutation detection can... Ensuring the system operates within a safe range and fully utilizes its transmission capacity is crucial. When commutation stability is detected, the system can appropriately increase transmission power to fully leverage the advantages of high-voltage direct current (HVDC) transmission systems in terms of large capacity and long-distance transmission, thereby improving energy utilization efficiency. Reducing operating costs is also essential: improving the accuracy and real-time performance of commutation detection reduces commutation failures, thus lowering operating costs such as equipment wear and power outages. Optimized maintenance plans and operating strategies also reduce unnecessary equipment repair and replacement costs, improving the system's economic efficiency. Furthermore, a stable and reliable HVDC transmission system contributes to improved energy trading efficiency and effectiveness. Accurate commutation detection ensures the system completes energy transmission tasks according to plan, reducing the risk of energy trading defaults due to system failures and enhancing the stability and liquidity of the energy market.

[0093] In a third aspect, this application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform a commutation detection method as described in any of the first aspects.

[0094] This application provides a computer device in a fourth aspect, including a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform a commutation detection method as described in any of the first aspects.

[0095] Figure 3 The diagram illustrates the internal structure of a computer device in some embodiments. This computer device may specifically be a terminal, a server, or a gateway. Figure 3 As shown, the computer device includes a processor, memory, and network interface connected via a system bus.

[0096] The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium of the computer device stores an operating system and may also store a computer program. When executed by a processor, this computer program causes the processor to perform the steps in the above method embodiments. The internal memory may also store a computer program, which, when executed by a processor, causes the processor to perform the steps in the above method embodiments. Those skilled in the art will understand that... Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0097] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above methods.

[0098] Any references to memory, storage, database, or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0099] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0100] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A commutation detection method, characterized in that, The method includes: Acquire the exit valve current signal and the on valve current signal of the current commutation event to be tested; Based on the exit valve current signal and the conduction valve current signal of the current commutation event under test, determine the exit valve bounce rate and conduction valve rise time of the current commutation event under test. Substituting the exit valve rebound rate and the conduction valve rise time of the current commutation event under test into the preset parabolic criterion equation, the parabolic criterion value is obtained. Based on the parabolic criterion value, the commutation detection result of the current commutation event to be tested is determined.

2. The commutation detection method according to claim 1, characterized in that, The step of determining the exit valve bounce rate and conduction valve rise time of the current commutation event under test based on the exit valve current signal and conduction valve current signal of the current commutation event under test includes: Determine the current values ​​of the exit valve current signal of the current commutation event under test from the minimum current value to the end of commutation, and determine the exit valve rebound rate of the current commutation event under test based on the current values ​​of the exit valve current signal from the minimum current value to the end of commutation. The first moment corresponding to the current value equal to the first current value after the start of commutation in the current commutation event to be tested, and the second moment corresponding to the current value equal to the second current value are determined. Based on the first moment and the second moment, the rise time of the commutation valve in the current commutation event to be tested is determined, wherein the first current value is a first preset percentage of the preset DC current value, the second preset current value is a second preset percentage of the preset DC current value, and the first preset percentage is less than the second preset percentage.

3. The commutation detection method according to claim 2, characterized in that, The exit valve rebound rate of the current commutation event under test is determined using the following formula: ； in, The exit valve rebound rate is the current commutation event to be tested. To find the maximum value function, It is the derivative of the current value at time t from the time of the minimum value to the time of the end of commutation with respect to time t. At the time of the minimum value, The time when the commutation ends is mentioned.

4. The commutation detection method according to claim 2, characterized in that, The step of determining the on-valve rise time of the current commutation event under test based on the first time moment and the second time moment includes: The difference between the second time point and the first time point is used as the on-valve rise time of the current commutation event to be tested.

5. The commutation detection method according to claim 1, characterized in that, The method further includes: Acquire the exit valve current signal, conduction valve current signal and tag of various historical commutation events, including various normal operating conditions and various commutation failure scenarios; Based on the exit valve current signal and the on valve current signal of each historical commutation event, determine the exit valve bounce rate and the on valve rise time of each historical commutation event. Using the least squares method, multiple coefficients of a pre-set second-order polynomial regression equation are determined based on the exit valve rebound rate, conduction valve rise time, and labels of various historical commutation events. Substitute multiple coefficients of the preset second-order polynomial regression equation into the preset second-order polynomial regression equation to obtain the target second-order polynomial regression equation. Based on the target second-order polynomial regression equation, determine the parabola criterion parameters; Based on the parabola criterion parameters, the preset parabola criterion equation is constructed.

6. The commutation detection method according to claim 5, characterized in that, The parabola criterion parameters include the x-coordinate of the vertex, the y-coordinate of the vertex, and the opening coefficient of the parabola. Determining the parabola criterion parameters based on the target second-order polynomial regression equation includes: By eliminating cross-product terms and completing the square on the target second-order polynomial regression equation, a parabolic equation is obtained. Based on the equation of the parabola, determine the x-coordinate of the vertex, the y-coordinate of the vertex, and the opening coefficient of the parabola.

7. The commutation detection method according to claim 1, characterized in that, The method further includes: Acquire the exit valve current signal, conduction valve current signal and tag of various historical commutation events, including various normal operating conditions and various commutation failure scenarios; Based on the exit valve current signal and the on valve current signal of each historical commutation event, determine the exit valve bounce rate and the on valve rise time of each historical commutation event. The exit valve rebound rate, conduction valve rise time and labels of various historical commutation events are input into the initial classification prediction model for training, and the target classification prediction model is obtained. Based on the model parameters of the target classification prediction model, determine the parabolic criterion parameters; Based on the parabola criterion parameters, the preset parabola criterion equation is constructed.

8. The commutation detection method according to claim 7, characterized in that, The model parameters of the target classification prediction model include a support vector set, a label set corresponding to the support vector set, a dual coefficient set corresponding to the support vector set, and a bias term. The parabolic criterion parameters include the x-coordinate of the vertex, the y-coordinate of the vertex, and the opening coefficient of the parabola. Determining the parabolic criterion parameters based on the model parameters of the target classification prediction model includes: Substituting the support vector set, the label set corresponding to the support vector set, the dual coefficient set corresponding to the support vector set, and the bias term into the decision function of the target classification prediction model, a second-order polynomial equation is obtained. Eliminating cross-product terms and completing the square of the second-order polynomial equation yields the parabolic equation; Based on the equation of the parabola, determine the x-coordinate of the vertex, the y-coordinate of the vertex, and the opening coefficient of the parabola.

9. The commutation detection method according to any one of claims 1, 5 to 8, characterized in that, The preset parabola criterion equation is: ; in, The opening coefficient of the parabola in the parabola criterion parameters. The x-coordinate of the vertex of the parabola in the parabola criterion parameters is... The ordinate of the vertex of the parabola in the parabola criterion parameters is... The rise time of the valve is the opening time. For the exit valve rebound rate, This is the criterion value for a parabola.

10. The commutation detection method according to claim 1, characterized in that, The step of determining the commutation detection result of the current commutation event to be tested based on the parabolic criterion value includes: If the parabolic criterion value is less than the criterion threshold, the commutation detection result of the current commutation event to be tested is determined to be a commutation failure; If the parabolic criterion value is greater than or equal to the criterion threshold, the commutation detection result of the current commutation event to be tested is determined to be a successful commutation.

Images