Fast Prediction Method and System for HVDC Commutation Failure Based on Commutation Failure Risk Factor

Abstract

The present invention relates to a rapid prediction method and system for commutation failure of HVDC based on commutation failure risk factors, belonging to the technical field of commutation failure defense in high-voltage direct current transmission systems. This method can rapidly predict whether commutation failure will occur after a fault in the LCC-HVDC system, providing support for subsequent commutation failure defense. During the implementation process, this method defines the traditional commutation area and the leading commutation area, and calculates the commutation failure risk factor by integrating the traditional commutation area and the leading commutation area, and then predicts the commutation failure. The method comprehensively considers the current moment information and future moment information of the characteristic quantity during the prediction process, improving the prediction speed of commutation failure. At the same time, it can adaptively adjust the weight coefficients of the traditional commutation area and the leading commutation area according to the prediction error of the characteristic quantity, avoiding the reduction of the accuracy of commutation failure caused by a large prediction error of the characteristic quantity, and ensuring the reliability of the prediction result.

Description

Technical Field

[0001] The present invention relates to a method and system for quickly predicting commutation failure of a high-voltage direct current (HVDC) transmission system based on a commutation failure risk factor, and belongs to the technical field of commutation failure prevention of HVDC transmission systems. Background Art

[0002] The line-commutated converter high-voltage direct current (LCC-HVDC) technology has the advantages of large transmission capacity, low loss, and low cost, and can effectively address the problem of uneven distribution of energy centers and load centers in China, ensuring the reliable and stable supply of electric energy. Commutation failure (CF) is one of the most common faults in LCC-HVDC. It occurs during the commutation process, manifested as the thyristor being unable to turn off or on normally, threatening the stability and reliability of the power system supply. After commutation failure occurs, abnormal fluctuations will appear in the AC-DC system, bringing active and reactive power impacts to the LCC-HVDC system and threatening the stability and reliability of the entire power system. Continuous commutation failures may even trigger serious faults or large-scale power outages. Therefore, the commutation failure prevention technology has become a research hotspot, and the prediction of commutation failure is a key link among them. It is an important basis for the action of the commutation failure control and protection system. If commutation failure can be predicted in advance, the system operating state can be adjusted before the actual commutation failure occurs to increase the commutation margin, thereby avoiding the occurrence of commutation failure and improving the safety and stability of system operation.

[0003] Regarding the above problems, current domestic and foreign research mostly analyzes the commutation process of the converter based on the commutation voltage-time integral area theory, and predicts commutation failure by comparing the defined commutation demand area and commutation supply area.

[0004] However, the traditional commutation failure prediction method based on the commutation area theory predicts commutation failure at future moments based on the characteristic quantities at the measurement moment. It ignores the transient changes of characteristic quantities such as commutation voltage and DC current from the current measurement moment to future moments, resulting in a limit value for the commutation failure early prediction time of this method, that is, the size of half a power frequency cycle. Therefore, the traditional prediction method has deficiencies in prediction speed when predicting commutation failure and is prone to misjudgment phenomena. This problem has become the key to restricting the prediction accuracy of commutation failure in the field of commutation failure prevention of HVDC transmission. Summary of the Invention

[0005] In view of the deficiencies of the prior art and to solve the technical problems existing in the above-mentioned background art, the present invention provides a fast prediction method for commutation failure of HVDC based on commutation failure risk factors, which can more quickly predict the occurrence of commutation failure after a fault in the LCC-HVDC system.

[0006] The technical solution of the present invention is as follows:

[0007] A fast prediction method for commutation failure of HVDC based on commutation failure risk factors includes the following steps:

[0008] Step 1: Obtain the commutation voltage amplitude at the current moment, U com.t the DC current at the current moment, I d.t the commutation voltage amplitude at the future moment, U com.c the DC current at the future moment, I d.c and the trigger angle command value α , all of which are per-unit values;

[0009] Step 2: Based on the commutation voltage amplitude at the current moment U com.t , the DC current at the current moment I d.t and the trigger angle command value α , calculate the commutation demand area and the maximum commutation supply area, and define them as the traditional commutation demand area and the traditional maximum commutation supply area respectively; at the same time, in order to improve the prediction speed of commutation failure and overcome the problem of the limit of the advance prediction time existing in the traditional prediction method, based on the predicted commutation voltage amplitude at the future moment U com.c , the DC current at the future moment I d.c and the trigger angle command value α , calculate the commutation demand area and the maximum commutation supply area, and define them as the advanced commutation demand area and the advanced maximum commutation supply area respectively;

[0010] Preferably, in Step 2, based on the commutation voltage amplitude at the current moment U com.t , the DC current at the current moment I d.t and the trigger angle command value α , calculate the traditional commutation demand area S need.t and the traditional maximum commutation supply area S pro.t respectively according to Equations (2) and (3);

[0011] The commutation voltage amplitude at the future moment obtained based on prediction U com.c , the direct current at the future moment I d.c and the trigger angle command value α calculate the leading commutation demand area and the leading maximum commutation supply area respectively according to formulas (4) and (5) S need.c and the leading maximum commutation supply area S pro.c ;

[0012] S need.t = 2 X c I d.t (2)

[0013] (3)

[0014] S need.c = 2 X c I d.c (4)

[0015] (5)

[0016] In the formulas, X c is the equivalent commutation reactance, γ min is the minimum turn-off angle, ω 0 is the system angular frequency, with a magnitude of 2π / T , T is the fundamental wave period of the system; t represents the integration variable.

[0017] Step 3: Considering that there may be errors in the commutation voltage amplitude at the future moment U com.c and the direct current at the future moment I d.c calculated by prediction, calculate the weight coefficients of the traditional commutation demand area and the leading commutation demand area;

[0018] Preferably, in Step 3, based on the direct current at the current moment I d.t and the predicted value of the current direct current in the historical data I d.c0 calculate the weight coefficients C 1 and C 2 of the traditional commutation demand area and the leading commutation demand area according to formula (6);

[0019] (6)

[0020] Wherein, k i1 is the normalization coefficient of the commutation demand area weight, k i2 is the sensitivity coefficient of the commutation demand area weight.

[0021] Step 4: Calculate the weight coefficients of the traditional maximum commutation provided area and the leading maximum commutation provided area;

[0022] Preferably, in Step 4, based on the commutation voltage amplitude at the current moment U com.t and the predicted value of the current commutation voltage amplitude in the historical data U com.c0 Calculate the weight coefficients of the traditional maximum commutation provided area and the leading maximum commutation provided area according to Equation (7) C 3 and C 4;

[0023] (7)

[0024] Wherein, k u1 is the normalization coefficient of the maximum commutation provided area weight, k u2 is the sensitivity coefficient of the maximum commutation provided area weight.

[0025] Step 5: Weight and integrate the calculated traditional commutation demand area, traditional maximum commutation provided area, leading commutation demand area, and leading maximum commutation provided area through the weight coefficients to obtain the commutation failure risk factor, and then predict whether commutation failure will occur; and during the prediction process, adaptively adjust the weight coefficients of each commutation area according to the prediction error of the characteristic quantity at the previous moment. The prediction error of the characteristic quantity is the absolute value part in Equation (6) and Equation (7). The prediction error of the DC current is | I d.t - I d.c0 |, and the prediction error of the commutation voltage is | U com.t - U com.c0|, the adjustment of the weight coefficient is shown in Equations (6) and (7). As the system operating state changes, the prediction error of the characteristic quantity changes in real time, and the weight coefficient of each commutation area is adaptively adjusted according to the real-time prediction error of the characteristic quantity (the process of real-time calculation according to Equations (6) and (7) is the adaptive adjustment). If the prediction error of the commutation voltage is large, the weight coefficient of the leading largest commutation-providing area is small, while the weight coefficient of the traditional largest commutation-providing area is large. The same applies to the DC current. On the premise of ensuring the prediction accuracy, the prediction speed of commutation failure is improved.

[0026] Preferably, in step 5, based on the traditional commutation demand area calculated in steps 2 - 4 S need.t , the traditional largest commutation-providing area S pro.t , the leading commutation demand area S need.c and the leading largest commutation-providing area S pro.c and the weight coefficient C 1 - C 4 calculates the commutation failure risk factor in real time according to Equation (8) F ;

[0027] F = C 1 S need.c + C 2 S need.t + C 3 S pro.c + C 4 S pro.t (8)

[0028] If F ≥ 0, it is considered that the system state meets the normal commutation condition and commutation failure will not occur; if F < 0, it is considered that the system state does not meet the normal commutation condition and commutation failure will occur.

[0029] A computer-readable storage medium, on which a program is stored, and when the program is executed by a processor, it implements the steps in the HVDC commutation failure rapid prediction method based on the commutation failure risk factor.

[0030] An electronic device, including a memory, a processor, and a program stored on the memory and executable on the processor, and when the processor executes the program, it implements the steps in the HVDC commutation failure rapid prediction method based on the commutation failure risk factor.

[0031] The beneficial effects of the present invention are as follows:

[0032] This paper proposes a fast prediction method for commutation failure in HVDC based on the commutation failure risk factor, which essentially predicts the occurrence of the first commutation failure in advance for subsequent defense. This method can quickly and accurately predict whether commutation failure will occur after a fault in the LCC-HVDC system, providing support for subsequent defense against commutation failure. In the implementation process, this method defines the traditional commutation area and the leading commutation area, and comprehensively uses the traditional commutation area and the leading commutation area to predict commutation failure. The method comprehensively considers the current moment information and future moment information of the characteristic quantity during the prediction process, ensuring the rapidity and accuracy of the prediction and improving the prediction speed of commutation failure. At the same time, it can adaptively adjust the weight coefficients of the traditional commutation area and the leading commutation area according to the prediction error of the characteristic quantity, define the commutation failure risk factor for commutation failure prediction, avoid the reduction of commutation failure accuracy caused by large prediction errors of the characteristic quantity, and ensure the reliability of the prediction results. Description of the Drawings

[0033] Figure 1 It is a structure diagram of a 6-pulse converter;

[0034] Figure 2 It is a flowchart of the commutation failure prediction method based on the commutation failure risk factor of this application;

[0035] Figure 3 It is a main structure diagram of the LCC-HVDC system;

[0036] Figure 4 It is a waveform diagram of DC current, three-phase AC bus voltage, and valve current under the single-phase 0.17H grounding fault condition; (a) is the DC current waveform diagram near the fault moment, (b) is the three-phase AC bus voltage waveform diagram of the inverter station near the fault moment, and (c) is the valve current waveform diagram of the two valves where commutation failure first occurs;

[0037] Figure 5 It is a waveform diagram of DC current, three-phase AC bus voltage, and valve current under the three-phase 0.18H symmetrical grounding fault condition; (a) is the DC current waveform diagram near the fault moment, (b) is the three-phase AC bus voltage waveform diagram of the inverter station near the fault moment, and (c) is the valve current waveform diagram of the two valves where commutation failure first occurs. Detailed Embodiment

[0038] In order to make the technical solution of the present invention clearer, the following further details the fast prediction process of commutation failure in this invention patent in combination with the drawings and examples, but not limited thereto.

[0039] Embodiment 1:

[0040] Fast Prediction Method for Commutation Failure of HVDC Based on Commutation Failure Risk Factor, the process is as follows Figure 2 as shown. The steps are as follows:

[0041] Step 1: During the operation of LCC-HVDC, collect and calculate in real time the commutation voltage amplitude at the current moment of the converter station U com.t , the DC current at the current moment I d.t , the commutation voltage amplitude at the future moment U com.t , the DC current at the future moment I d.c , and the trigger angle command value α (all in per-unit values).

[0042] The DC current at the current moment I d.t and the trigger angle command value α are obtained by real-time acquisition. The commutation voltage amplitude at the current moment U com.t cannot be directly acquired, but is calculated through a conventional known algorithm from the instantaneous value of the commutation voltage collected in real time. The commutation voltage amplitude at the future moment U com.c , and the DC current at the future moment I d.c are respectively calculated through a conventional prediction algorithm based on the commutation voltage amplitude at the current moment U com.t and the DC current at the current moment I d.t . Any actual prediction method can be selected as long as it can calculate the predicted value containing future moment information

[0043] In this embodiment, the method for calculating the DC current at the future moment I d.t using the DC current at the current moment I d.c is based on the principle of Taylor expansion, as shown in the following formula. In the formula t 0 is the current moment, t c is the predicted future moment. In the formula is the approximate first-order differential of the DC current at the current moment, is the approximate second-order differential of the DC current at the current moment. In the present invention T 1 is taken as 1 ms, T 2 is taken as 8 ms

[0044] (1)

[0045] The values of the above characteristic quantities can all be obtained by known conventional methods and are regarded as known quantities in step 1.

[0046] Step 2: Based on the commutation voltage amplitude at the current moment U com.t , the direct current at the current moment I d.t and the trigger angle command value α , calculate the traditional commutation demand area S need.t and the traditional maximum commutation supply area S pro.t respectively according to formulas (2) and (3). Based on the predicted commutation voltage amplitude at the future moment U com.c , the direct current at the future moment I d.c and the trigger angle command value α , calculate the leading commutation demand area S need.c and the leading maximum commutation supply area S pro.c .

[0047] S need.t = 2 X c I d.t (2)

[0048] (3)

[0049] S need.c = 2 X c I d.c (4)

[0050] (5)

[0051] In the formula, X c is the equivalent commutation reactance, γ min is the minimum turn-off angle, ω 0 is the system angular frequency, with a magnitude of 2π / T , T is the fundamental period of the system; t represents the integration variable.

[0052] Step 3: Based on the direct current at the current moment I d.t and the predicted value of the current direct current in the historical dataI d.c0 Calculate the weight coefficients of the traditional commutation demand area and the leading commutation demand area according to Equation (6). C 1 and C 2.

[0053] (6)

[0054] In the formula, k i1 is the normalization coefficient of the commutation demand area weight, k i2 is the sensitivity coefficient of the commutation demand area weight.

[0055] Step 4: Based on the commutation voltage amplitude at the current moment U com.t and the predicted value of the current commutation voltage amplitude in historical data U com.c0 Calculate the weight coefficients of the traditional maximum commutation supply area and the leading maximum commutation supply area according to Equation (7). C 3 and C 4.

[0056] (7)

[0057] In the formula, k u1 is the normalization coefficient of the maximum commutation supply area weight, k u2 is the sensitivity coefficient of the maximum commutation supply area weight.

[0058] Step 5: Based on the traditional commutation demand area calculated in Steps 2 - 4 S need.t , the traditional maximum commutation supply area S pro.t , the leading commutation demand area S need.c and the leading maximum commutation supply area S pro.c and the weight coefficients C 1 - C 4, calculate the commutation failure risk factor F .

[0059] F = C 1 S need.c + C 2 S need.t + C 3 S pro.c + C 4S pro.t (8)

[0060] If F ≥ 0, it is considered that the system state meets the normal commutation condition and commutation failure will not occur; if F < 0, it is considered that the system state does not meet the normal commutation condition and commutation failure will occur.

[0061] Experimental example

[0062] To verify the feasibility of the proposed commutation failure rapid prediction method of the present invention, the following is verified by combining the method of Embodiment 1. The parameters of the LCC-HVDC are shown in Table 1.

[0063] Table 1 LCC-HVDC parameters

[0064]

[0065] In Figure 3 The verification is carried out in the shown monopolar LCC-HVDC standard model, where the converter stations on the rectifier side and the inverter side of the LCC-HVDC both adopt 12-pulse converters (formed by two 6-pulse converters connected in series). The converter transformer adopts a three-phase three-winding transformer, with the high-voltage side connected in Yg type, and the two low-voltage side windings are connected in Y type and D1 type respectively, and provide commutation voltages for the two 6-pulse converters respectively. The LCC-HVDC parameters are shown in Table 1. In this example, a grounding fault is set at the AC bus of the inverter side at 0.75 s during the normal operation of the LCC-HVDC, and the commutation failure of the system is predicted based on the traditional commutation failure prediction method and the method proposed in the present invention respectively. The prediction effects of the two methods are compared to verify the method proposed in the present invention.

[0066] Next, the rapidity of the proposed method is verified by comparing the prediction results of the commutation failure rapid prediction method proposed in the present invention and the existing commutation area-based prediction method. The relevant parameter settings are as follows: the normalization coefficient of the commutation demand area weight k i1 = 1.7, the sensitivity coefficient of the commutation demand area weight k i2 = -20, the normalization coefficient of the maximum commutation supply area weight k u1 = 1.8, the sensitivity coefficient of the maximum commutation supply area weight k u2 = -60. It can be seen from Equation (6) that k i1 The role of is to limit the calculated weight coefficients C 1 and C 2 within [-0.9, 0] to prevent C1 and C 2 are too large or too small, k i2 then the weight coefficients C 1 and C 2's sensitivity to the change of DC current prediction error k i2 The larger the absolute value of C 1 and C 2 is, the more sensitive the weight coefficients k u1 The function of C 3 and C 4 is to limit the calculated weight coefficients k u2 to [0, 0.9], C 3 and C 4's sensitivity to the change of commutation voltage amplitude prediction error k u2 The larger the absolute value of C 3 and C 4 is, the more sensitive the weight coefficients k i1 , k i2 , k u1 , k u2 and other parameters directly affect the rapidity and accuracy of the prediction method. Therefore, these parameters need to be reasonably set before prediction.

[0067] Since commutation failure in the LCC-HVDC operation process is usually caused by single-phase grounding fault, the method is verified by taking single-phase grounding fault as an example first. Set a phase A grounding fault at the AC bus on the inverter side of the LCC-HVDC system, and the grounding impedance is 0.17H. The relevant waveforms under this condition are as Figure 4 shown. Figure 4 Among them, (a) is the DC current waveform near the fault moment, (b) is the three-phase voltage waveform of the inverter station AC bus near the fault moment, and (c) is the valve current waveforms of the two valves when commutation failure occurs for the first time. From Figure 4It can be seen that after a single-phase ground fault occurs, the DC current fluctuates, the AC voltage of the fault phase drops significantly, and the first commutation failure occurs during the commutation process of converter valves 4 - 6 corresponding to the Y-connected transformer at 0.8279 s. After the commutation failure occurs, the DC current rises rapidly, the distortion of the AC voltage becomes more serious, and the system becomes unstable. The commutation failure under this condition is predicted based on the traditional commutation area method and the method proposed in the present invention. The prediction results are shown in case 4 of Table 2. The "×" indicates that the traditional method has a missed judgment, while the method proposed in the present invention successfully predicts at 0.8232 s. Continue to set single-phase ground faults with different impedance values on the inverter AC side for verification. The impedance values cover 7 different fault impedances ranging from 0.14 H to 0.2 H, with a step size of 0.1 H, and the fault duration is 100 ms for all cases. The prediction effects of the two methods are compared, and the prediction results are shown in Table 2. The first column of data in Table 2 is the fault test condition, the second column is the magnitude of the grounding impedance, the third column is the actual commutation failure occurrence time after the fault, and "not occurred" means that no commutation failure occurs during the fault and system recovery period; the fourth column and the fifth column of data are the moments when the traditional commutation area method and the method proposed in the present invention judge commutation failure (the moment when the criterion is less than 0), respectively. The "×" indicates a missed judgment of commutation failure, that is, the failure to successfully predict before the actual commutation failure occurs, and the "√" indicates a false judgment where no commutation failure occurs, that is, the criterion is always greater than 0 when the system does not have a commutation failure. It can be seen from the data in Table 2 that in single-phase fault cases 1, 2, and 3, both the traditional method and the proposed method can successfully predict before the actual commutation failure occurs, and the method of the present invention can judge commutation failure earlier than the traditional method, with a faster prediction speed; in single-phase fault case 4, the traditional method has a missed judgment of commutation failure, while the method of the present invention successfully predicts before the actual commutation failure; in single-phase fault cases 5, 6, and 7, no commutation failure occurs, and neither the traditional method nor the proposed method has a false judgment.

[0068] Table 2 Prediction Results of Commutation Failure under Single-Phase Fault Conditions

[0069]

[0070] Note: The data in the fourth column are the moments when the criterion of the traditional method is less than 0; the data in the fifth column are the moments when the criterion of the proposed method is less than 0; the "×" in Table 2 indicates the failure to successfully predict before the actual commutation failure occurs, that is, a missed judgment; the "√" in Table 2 indicates that the criterion is always greater than 0 when the system does not have a commutation failure, that is, no false judgment.

[0071] Taking the three-phase symmetrical ground fault as an example for verification. A three-phase symmetrical ground fault is set at the AC bus of the inverter side of the system, with a grounding impedance of 0.18 H. The relevant waveforms under this condition are as Figure 5 shown. Figure 5Among them, (a) is the DC current waveform near the fault moment, (b) is the three-phase voltage waveform of the inverter station AC bus near the fault moment, and (c) is the valve current waveforms of the two valves where commutation failure first occurs. It can be seen from Figure 5 that after the three-phase grounding fault occurs, the DC current fluctuates, the three-phase AC voltages of A, B, and C all drop significantly, and the first commutation failure occurs during the commutation process of valves 2 - 4 of the converter corresponding to the D1-type connection transformer at 0.7628 s. The commutation failure under this condition is predicted based on the traditional commutation area method and the method proposed in the present invention. The prediction results are shown in Table 3, Condition 2. The traditional method successfully predicts at 0.7560 s, while the method proposed in the present invention successfully predicts at 0.7532 s. A three-phase symmetrical grounding fault is continuously set at the AC bus on the inverter side of the system. The impedance values cover 4 different fault impedances from 0.13 H - 0.28 H, with a step size of 0.5 H, and the fault duration is 100 ms for all cases. The prediction effects of the two methods are shown in Table 3. It can be seen from the data in Table 3 that in the three-phase fault Condition 1, the traditional method misses the judgment of commutation failure, while the method proposed in the present invention successfully predicts in advance; in the three-phase fault Condition 2, both the traditional method and the method proposed in the present invention successfully predict before the actual commutation failure occurs, and the method proposed in the present invention can judge commutation failure earlier than the traditional method; in the three-phase fault Conditions 3 and 4, commutation failure does not occur, and neither the traditional method nor the method proposed in the present invention makes a misjudgment. Thus, it can be seen that the method proposed in the present invention can predict the occurrence of commutation failure faster under the premise of ensuring accuracy after a single-phase grounding fault occurs in the AC system on the inverter side of the LCC-HVDC, which is more conducive to the subsequent operation of the commutation failure control and protection system.

[0072] Table 3 Prediction results of commutation failure under three-phase symmetrical fault conditions

[0073]

[0074] Note: The data in the 4th column is the moment when the criterion of the traditional method is less than 0; the data in the 5th column is the moment when the criterion of the proposed method is less than 0; the × in Table 3 indicates that the prediction fails to succeed before the actual commutation failure occurs, that is, a missed judgment occurs; the √ in Table 3 indicates that the criterion is always greater than 0 when the system does not have commutation failure, that is, no misjudgment occurs.

Claims

1. A fast prediction method for HVDC commutation failure based on commutation failure risk factors, characterized in that, Based on the commutation voltage amplitude at the current moment U com.t , the DC current at the current moment I d.t , the commutation voltage amplitude at the future moment U com.c , and the DC current at the future moment I d.c , the traditional commutation demand area S need.t , the traditional maximum commutation supply area S pro.t and the leading commutation demand area S need.c , the leading maximum commutation supply area S pro.c are calculated respectively. Then, through the adaptive weight coefficient C 1 - C 4, the traditional commutation area and the leading commutation area are integrated to obtain the commutation failure risk factor F , and further judge the occurrence of commutation failure, so as to realize the rapid prediction of commutation failure after the LCC - HVDC fault; It includes the following steps: Step 1: Obtain the commutation voltage amplitude at the current moment, the DC current at the current moment, the commutation voltage amplitude at the future moment, the DC current at the future moment, and the trigger angle command value in real time during the operation of the LCC-HVDC system. All are per-unit values; U com.t and the DC current at the current moment I d.t as well as the commutation voltage amplitude at the future moment U com.c and the DC current at the future moment I d.c and the trigger angle command value α are all per-unit values; Step 2: Based on the commutation voltage amplitude at the current moment U com.t , the DC current at the current moment I d.t and the trigger angle command value α calculate the commutation demand area and the maximum commutation supply area, and define them as the traditional commutation demand area and the traditional maximum commutation supply area respectively; meanwhile, based on the predicted commutation voltage amplitude at the future moment U com.c , the DC current at the future moment I d.c and the trigger angle command value α calculate the commutation demand area and the maximum commutation supply area, and define them as the leading commutation demand area and the leading maximum commutation supply area respectively; Step 3: Calculate the weight coefficients of the traditional commutation demand area and the leading commutation demand area; Step 4: Calculate the weight coefficients of the traditional maximum commutation supply area and the leading maximum commutation supply area; Step 5: Weightedly integrate the calculated traditional commutation demand area, traditional maximum commutation supply area, leading commutation demand area, and leading maximum commutation supply area through the weight coefficients to obtain the commutation failure risk factor, and then predict whether commutation failure will occur; and during the prediction process, adaptively adjust the weight coefficients of each commutation area according to the prediction error of the feature quantity at the previous moment; In step 5, based on the traditional commutation demand area calculated in steps 2 - 4 S need.t , the traditional maximum commutation supply area S pro.t , the leading commutation demand area S need.c and the leading maximum commutation supply area S pro.c and the weight coefficient C 1 - C 4, calculate the commutation failure risk factor in real time according to formula (8) F ; F = C 1 S need.c + C 2 S need.t + C 3 S pro.c + C 4 S pro.t (8) If F ≥ 0, it is considered that the system state meets the normal commutation condition and commutation failure will not occur; if F < 0, it is considered that the system state does not meet the normal commutation condition and commutation failure will occur.

2. The HVDC commutation failure rapid prediction method based on commutation failure risk factors according to claim 1, wherein In step 2, based on the commutation voltage amplitude at the current moment U com.t , the DC current at the current moment I d.t and the trigger angle command value α are respectively used to calculate the traditional commutation demand area S need.t and the traditional maximum commutation supply area S pro.t ; The amplitude of the commutation voltage at the future moment obtained based on prediction U com.c , the DC current at the future moment I d.c and the trigger angle command value α respectively calculate the leading commutation demand area and the leading maximum commutation supply area according to Equation (4) and Equation (5) S need.c ; S pro.c ; S need.t =2 X c I d.t (2) (3) S need.c =2 X c I d.c (4) (5) In the formula, X c is the equivalent commutation reactance, γ min is the minimum turn-off angle, ω 0 is the system angular frequency, with a magnitude of 2π / T , T is the fundamental period of the system; t represents the integration variable.

3. The fast prediction method for commutation failure of HVDC based on commutation failure risk factors according to claim 1, wherein In step 3, based on the DC current at the current moment I d.t and the predicted value of the current DC current in the historical data I d.c0 Calculate the weight coefficients of the traditional commutation demand area and the leading commutation demand area according to Equation (6) C 1 and C 2; （6） In the formula, k i1 is the normalization coefficient of the commutation demand area weight, k i2 is the sensitivity coefficient of the commutation demand area weight.

4. The fast prediction method for HVDC commutation failure based on commutation failure risk factors according to claim 1, characterized in that In step 4, based on the commutation voltage amplitude at the current moment U com.t and the predicted value of the current commutation voltage amplitude in historical data U com.c0 Calculate the weight coefficients of the traditional maximum commutation supply area and the leading maximum commutation supply area according to formula (7) C 3 and C 4; (7) In the formula, k u1 is the normalization coefficient providing the area weight for the maximum commutation, k u2 is the sensitivity coefficient providing the area weight for the maximum commutation.

5. A computer-readable storage medium, characterized in that, It stores a program, and when the program is executed by a processor, it implements the steps in the HVDC commutation failure rapid prediction method based on the commutation failure risk factor described in any one of claims 1-4.

6. An electronic device, characterized in that, It includes a memory, a processor, and a program stored on the memory and executable on the processor. When the processor executes the program, it implements the steps in the HVDC commutation failure rapid prediction method based on the commutation failure risk factor described in any one of claims 1-4.

Images