Method for rapidly extracting amplitude and phase angle of HVDC (high voltage direct current) commutation voltage based on trend prediction

Abstract

The invention relates to an HVDC (high voltage direct current) commutation voltage amplitude phase angle rapid extraction method based on trend prediction, and belongs to the technical field of commutation failure defense of a high voltage direct current transmission system.The method comprises the steps that firstly, fitting data of a commutation voltage amplitude phase angle is preliminarily obtained by conducting trigonometric function least square fitting on data collected in real time of commutation voltage; under the condition that other collected data is not added, the trend of the amplitude and the phase angle is predicted based on the current fitting data, and the amplitude and the phase angle after the commutation voltage is changed are obtained in a shorter time by defining a memory component, a trend component and an oscillation component in the prediction process; therefore, rapid extraction of the commutation voltage amplitude and the phase angle after the HVDC fault is realized, and support is provided for rapid prediction and suppression of subsequent commutation failure.

Description

Technical Field

[0001] The present invention relates to a method for quickly extracting a commutation voltage amplitude and phase angle of a high voltage direct current (HVDC) power transmission system based on trend prediction, and belongs to the technical field of commutation failure defense of a high voltage direct current power transmission system. Background Art

[0002] Line-commutated Converter HVDC (LCC-HVDC) based on grid commutation has the characteristics of large transmission capacity, low loss and low cost, which effectively solves the problem of uneven distribution of energy centers and load centers and ensures the reliable supply of electric energy. Commutation failure (CF) is one of the most common faults in LCC-HVDC. It will cause the DC current of the system to rise rapidly, bringing active and reactive power impact to the DC system. In severe cases, it will cause DC transmission to be locked and trigger active power flow transfer. Commutation failure defense technology is a current research hotspot. In commutation failure defense, the commutation voltage is one of the most concerned feature quantities. The amplitude of the commutation voltage directly affects the time required for the commutation process, and the phase angle of the commutation voltage directly affects the zero-crossing time of the voltage to be shut off. Therefore, extracting the amplitude and phase angle information of the commutation voltage can provide support for the prediction of commutation failure. At the same time, if the amplitude and phase angle of the commutation voltage can be extracted in a shorter time after the system fails, the rapidity of commutation failure prediction can be improved, so that the system operating parameters can be adjusted earlier and the risk of commutation failure can be reduced. In summary, it is very important to quickly extract the amplitude and phase angle of HVDC commutation voltage.

[0003] For the above problems, current domestic and foreign research mostly uses the least squares fitting method of trigonometric functions to extract the commutation voltage amplitude and phase angle, as recorded in the following literature: Wang Zengping, Liu Xiyang, Zheng Bowen, et al. Rapid prediction and suppression measures of commutation failure based on voltage waveform fitting [J]. Transactions of the Chinese Society of Electrotechnical Engineering, 2020, 35(7): 1454-1463.

[0004] In the existing trigonometric least squares fitting method, during the operation of the LCC-HVDC system, no matter how the commutation voltage amplitude and phase angle change after the system fault, the expression can be written in the following form: (1); Formula (1) is a nonlinear function. u is the instantaneous value of the fitted commutation voltage, t For time, oh 0 is the system angular frequency, Um0 represents the amplitude of the commutation voltage fitting function, i 0 Represents the phase angle of the commutation voltage fitting function. Using the sum and difference angle formula of trigonometric functions, the above formula (1) can be rewritten into the following linear form: (2); in, U m1 represents the amplitude of the sinusoidal part, U m2 represents the magnitude of the cosine part; and U m0 , U m1 , U m2 , i 0 Satisfies the following relationship: (3); According to the least squares linear fitting formula, U m1 and U m2 The fitting results are as follows: (4); In the formula, , u 1 , u 2 , …, u m for m The instantaneous value of a voltage, t 1 , t 2 , …, t m for m The time signal of the instantaneous value of the voltage. Then U m1 and U m2 Substituting into equation (3), we can get the amplitude of the commutation voltage fitting function: U m0 and the phase angle of the commutation voltage fitting function i 0 , and get the fitting result.

[0005] However, the existing trigonometric function fitting method has shortcomings in the extraction speed of the commutation voltage amplitude and phase after the fault, resulting in a lag in the prediction and suppression of commutation failure, making the occurrence of commutation failure inevitable. This problem has become a bottleneck limiting the rapidity of commutation failure prediction in the field of HVDC transmission commutation failure prevention. Summary of the invention

[0006] In order to solve the above problems, the present invention provides a method for quickly extracting the amplitude and phase angle of HVDC commutation voltage based on trend prediction, which can quickly extract the amplitude and phase angle of the commutation voltage after a system fault.

[0007] The technical solution of the present invention is as follows: A method for quickly extracting HVDC commutation voltage amplitude and phase angle based on trend prediction comprises the following steps: S1: Set the sampling interval T and the number of fitted data m The instantaneous value of the commutation voltage collected is u The change over time is input into the voltage fitting module in real time, and the counter i The initial value of i ≥1); S2: Input the collected instantaneous value of the commutation voltage into the voltage fitting module in real time. Each time the instantaneous value of the commutation voltage is input, the counter i +1 for real-time input m The instantaneous value of the commutation voltage is fitted by trigonometric least squares function to calculate the amplitude and phase angle of the fitting curve; The fitting process of this step is carried out in real time. m Amplitude and phase fitting extraction is performed on each piece of data.

[0008] S3: input the amplitude and phase angle obtained in step S2 into the trend prediction model, and set the parameters of the trend prediction model; S4: When the counter i ≤ p +1, p is the oscillation period length. p + The amplitude and phase angle of the first fitting curve are used to initialize the memory component, trend component and oscillation component of the trend prediction model, and the amplitude and phase angle data of the original fitting curve are used as the prediction results as the amplitude and phase angle after trend prediction respectively; S5: When the counter i > p +1, the memory component, trend component and oscillation component of the trend prediction model are updated, and the amplitude and phase angle are predicted based on the updated memory component, trend component and oscillation component; S6: Real-time output of amplitude and phase angle after trend prediction.

[0009] In step S1, the sampling interval T and the number of fitted data m Affects the fitting results, preferably the sampling interval T =0.2ms, number of fitting data m= 5.

[0010] Preferably, in step S2, the input m The instantaneous value of the commutation voltage is fitted by the least squares trigonometric function to obtain the i The commutation voltage amplitude and phase angle fitting results U m0 ( i )and i 0 ( i ); (5); (6); in, , oh 0 is the system angular frequency; u m ( i ) indicates the i Input m The instantaneous value of the commutation voltage; t m ( i ) indicates the i Input m The time signal corresponding to the instantaneous value of the commutation voltage; U m1 ( i )and U m2 ( i ) is an intermediate variable in the calculation process and has no actual physical meaning.

[0011] Preferably, in step S3, the parameters of the trend prediction model include the amplitude memory factor α u , Phase Angle Memory Factor α θ , Amplitude trend factor β u , phase angle trend factor β θ , Amplitude Oscillation Factor c u , phase angle oscillation factor c θ and the oscillation period lengthp .

[0012] The choice of parameters affects the prediction results. Preferably, the amplitude memory factor α u =0.5, phase angle memory factor α θ =0.2, amplitude trend factor β u =0.5, phase angle trend factor β θ =0.7, amplitude oscillation factor c u =0.3, phase oscillation factor c θ =0.5, oscillation period length p =6.

[0013] Preferably, in step S4, the memory component includes an amplitude memory component L u and phase angle memory component L θ , the trend component includes the amplitude trend component B u and phase angle trend component B θ , the oscillation component includes the amplitude oscillation component S u and phase angle oscillation component S θ ,in L u , B u , S u , α u , β u , c u For amplitude trend prediction, L θ , B θ , S θ , α θ , β θ , c θ Used for trend prediction of phase angle.

[0014] First, according to the previous p The amplitude and phase angle of the +1 fitting curve are used to initialize the memory component, trend component and oscillation component of the trend prediction model, as shown in equations (7) and (8): (7); (8); in, U m0 (1) represents the first commutation voltage amplitude fitting result; U m0 ( p +1) indicates the p +1 commutation voltage amplitude fitting result; U m0 (1: p ) indicates the first to the p An array consisting of the commutation voltage amplitude fitting results; i 0 (1) represents the first commutation voltage phase angle fitting result; i 0 ( p +1) indicates the p +1 commutation voltage phase angle fitting result; i 0 (1: p ) indicates the first to the p An array consisting of the commutation voltage phase angle fitting results; Then, the amplitude and phase angle are trended and the counter i ≤ p +1, the amplitude and phase angle data of the original fitting curve are stored as prediction results in U m ( i )and i ( i ), as shown in formula (9) and formula (10): (9); (10).

[0015] Preferably, in step S5, as the counter i Increase to greater than p +1, the program will U m0 ( i ), phase angle data i 0 ( i ), L u and L θ , B u and B θ, S u and S θ Input to the trend prediction module for update, specifically: L u and L θ Assign values ​​to L u _prev and L θ _prev ,Right now L u _prev = L u , L θ _prev = L θ , and according to formula (11) and formula (12) respectively L u , B u , S u , L θ , B θ , S θ To update: (11); (12); Among them, mod() is the remainder function; S u Update according to the third line of equation (11), because S u is a length of p Array, each update is to update the array p A value in a quantity; S θ Update according to the third line of equation (12), S θ is a length of p Array, each update is to update the array p A value in a quantity.

[0016] Then, based on the updated L u and L θ , B u and B θ, S u and S θ , the amplitude and phase angle are predicted by equations (13) and (14) respectively: (13); (14).

[0017] In the present invention, α u , β u , c u Separately control the amplitude memory component L u , amplitude trend component B u and amplitude oscillation component S u The sensitivity of the update process to recent input data; α θ , β θ , c θ Separately control the phase angle memory component L θ , phase angle trend component B θ and phase angle oscillation component S θ The sensitivity of the update process to recent input data.

[0018] when i > p +1, for one input ( m Each time the prediction component is updated, the prediction component (memory component, trend component and oscillation component) is updated and the trend prediction is calculated. After each prediction component update is completed, their current values ​​are returned to participate in the next component update. As long as the system keeps inputting the collected instantaneous value of the commutation voltage into the extraction system of the present invention, the prediction component is always updated and returned. In other words, the extraction method proposed by the present invention extracts the commutation voltage amplitude and phase angle in real time. Finally, the amplitude extracted based on the trend prediction is output. U m ( i ) and phase angle i ( i ), as the fitting data of the commutation voltage amplitude and phase angle input into the trend prediction model changes continuously, U m ( i ) and phase angle i ( i) are also updated in real time as the commutation voltage amplitude and phase angle extracted by the method proposed in the present invention until there is no need to predict commutation failure.

[0019] For any details not provided in the present invention, please refer to the prior art.

[0020] Different from the existing fitting extraction method, the present invention does not directly output the commutation voltage amplitude and phase angle obtained by fitting, but transmits them to the trend prediction model for processing, and finally calculates the amplitude and phase angle extracted by the method of the present invention.

[0021] In the trend prediction model, three types of prediction components are defined: memory component, trend component and oscillation component. The final amplitude and phase angle extraction results U m ( i )and i ( i ) is obtained by adding the three components respectively. The trend prediction model updates these components in real time according to the input fitting data, and can control the sensitivity of this link to the historical fitting data by adjusting the size of the weighting coefficient, so as to make a more accurate prediction of future changes and ensure the rapidity and accuracy of the extraction of the HVDC commutation voltage amplitude and phase angle.

[0022] The beneficial effects of the present invention are: The method for rapidly extracting the amplitude and phase angle of HVDC commutation voltage based on trend prediction of the present invention first performs trigonometric least square fitting on the real-time collected data of the commutation voltage to preliminarily obtain fitting data of the amplitude and phase angle of the commutation voltage; then, without adding other collected data, the trend of the amplitude and phase angle is predicted based on the current fitting data, and in the prediction process, the amplitude and phase angle after the commutation voltage change is obtained in a shorter time by defining a memory component, a trend component and an oscillation component, thereby realizing rapid extraction of the amplitude and phase angle of the commutation voltage after an HVDC fault, and providing support for rapid prediction and suppression of commutation failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The drawings in the specification, which constitute a part of the present application, are used to provide further understanding of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute improper limitations on the present application.

[0024] Figure 1 It is a flow chart of a method for quickly extracting HVDC commutation voltage amplitude and phase angle based on trend prediction of the present invention; Figure 2 This is the wiring diagram for the 6-pulse inverter; Figure 3 is the commutation voltage waveform near the fault moment; Figure 4The commutation voltage amplitude waveform near the fault moment extracted by the method proposed by the present invention and the existing fitting method; Figure 5 The commutation voltage phase angle waveform near the fault moment is extracted by the method proposed in the present invention and the existing fitting method. DETAILED DESCRIPTION

[0025] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of the present invention are clearly and completely described below in conjunction with the drawings in the implementation of this specification, but are not limited to this. Anything not fully described in the present invention shall be based on the conventional technology in the art.

[0026] Example 1 A fast extraction method of HVDC commutation voltage amplitude and phase angle based on trend prediction, such as Figure 1 As shown, the following steps are included: S1: Set the sampling interval T and the number of fitted data m The instantaneous value of the commutation voltage collected is u The change over time is input into the voltage fitting module in real time, and the counter i The initial value of i ≥1); Sampling interval T and the number of fitted data m Affects the fitting results. In this embodiment, the sampling interval T =0.2ms, number of fitting data m= 5.

[0027] S2: Input the collected instantaneous value of the commutation voltage into the voltage fitting module in real time. Each time the instantaneous value of the commutation voltage is input, the counter i +1 for real-time input m The instantaneous value of the commutation voltage is fitted by trigonometric least squares function to calculate the amplitude and phase angle of the fitting curve; According to formula (5) and (6), the input m The instantaneous value of the commutation voltage is fitted by the least squares trigonometric function to obtain the i The commutation voltage amplitude and phase angle fitting results U m0 ( i )and i 0 ( i ); (5); (6); in, , oh0 is the system angular frequency; u m ( i ) indicates the i Input m The instantaneous value of the commutation voltage; t m ( i ) indicates the i Input m The time signal corresponding to the instantaneous value of the commutation voltage; U m1 ( i )and U m2 ( i ) is an intermediate variable in the calculation process and has no actual physical meaning.

[0028] The fitting process of this step is carried out in real time. m Amplitude and phase fitting extraction is performed on each piece of data.

[0029] S3: input the amplitude and phase angle obtained in step S2 into the trend prediction model, and set the parameters of the trend prediction model; The parameters of the trend prediction model include the amplitude memory factor α u , Phase Angle Memory Factor α θ , Amplitude trend factor β u , phase angle trend factor β θ , Amplitude Oscillation Factor c u , phase angle oscillation factor c θ and the oscillation period length p .

[0030] The choice of parameters affects the prediction results. In this embodiment, the amplitude memory factor α u =0.5, phase angle memory factor α θ =0.2, amplitude trend factor β u =0.5, phase angle trend factor β θ =0.7, amplitude oscillation factor c u =0.3, phase oscillation factor c θ =0.5, oscillation period length p =6.

[0031] S4: When the counter i ≤ p +1, p is the oscillation period length. p + The amplitude and phase angle of the first fitting curve are used to initialize the memory component, trend component and oscillation component of the trend prediction model, and the amplitude and phase angle data of the original fitting curve are used as the prediction results as the amplitude and phase angle after trend prediction respectively; Memory components include amplitude memory components L u and phase angle memory component L θ , the trend component includes the amplitude trend component B u and phase angle trend component B θ , the oscillation component includes the amplitude oscillation component S u and phase angle oscillation component S θ ,in L u , B u , S u , α u , β u , c u For amplitude trend prediction, L θ , B θ , S θ , α θ , β θ , c θ Used for trend prediction of phase angle.

[0032] First, according to the previous p The amplitude and phase angle of the +1 fitting curve are used to initialize the memory component, trend component and oscillation component of the trend prediction model, as shown in equations (7) and (8): (7); (8); in, U m0 (1) represents the first commutation voltage amplitude fitting result; U m0 ( p +1) indicates thep +1 commutation voltage amplitude fitting result; U m0 (1: p ) indicates the first to the p An array consisting of the commutation voltage amplitude fitting results; i 0 (1) represents the first commutation voltage phase angle fitting result; i 0 ( p +1) indicates the p +1 commutation voltage phase angle fitting result; i 0 (1: p ) indicates the first to the p An array consisting of the commutation voltage phase angle fitting results; Then, the amplitude and phase angle are trended and the counter i ≤ p +1, the amplitude and phase angle data of the original fitting curve are stored as prediction results in U m ( i )and i ( i ), as shown in formula (9) and formula (10): (9); (10).

[0033] S5: When the counter i > p +1, the memory component, trend component and oscillation component of the trend prediction model are updated, and the amplitude and phase angle are predicted based on the updated memory component, trend component and oscillation component; With the counter i Increase to greater than p +1, the program will U m0 ( i ), phase angle data i 0 ( i ), L u and L θ , B u and B θ , S u and S θInput to the trend prediction module for update, specifically: L u and L θ Assign values ​​to L u _prev and L θ _ previous ,Right now L u _prev = L u , L θ _prev = L θ , and according to formula (11) and formula (12) respectively L u , B u , S u , L θ , B θ , S θ To update: (11); (12); Among them, mod() is the remainder function; S u Update according to the third line of equation (11), because S u is a length of p Array, each update is to update the array p A value in a quantity; S θ Update according to the third line of equation (12), S θ is a length of p Array, each update is to update the array p A value in a quantity.

[0034] Then, based on the updated L u and L θ , B u and B θ , S u and Sθ , the amplitude and phase angle are predicted by equations (13) and (14) respectively: (13); (14).

[0035] α u , β u , c u Separately control the amplitude memory component L u , amplitude trend component B u and amplitude oscillation component S u The sensitivity of the update process to recent input data; α θ , β θ , c θ Separately control the phase angle memory component L θ , phase angle trend component B θ and phase angle oscillation component S θ The sensitivity of the update process to recent input data.

[0036] S6: Real-time output of amplitude and phase angle after trend prediction.

[0037] when i > p +1, for one input ( m Each time the prediction component is updated, the prediction component (memory component, trend component and oscillation component) is updated and the trend prediction is calculated. After each prediction component update is completed, their current values ​​are returned to participate in the next component update. As long as the system keeps inputting the collected instantaneous value of the commutation voltage into the extraction system of the present invention, the prediction component is always updated and returned. In other words, the extraction method proposed by the present invention extracts the commutation voltage amplitude and phase angle in real time. Finally, the amplitude extracted based on the trend prediction is output. U m ( i ) and phase angle i ( i ), as the fitting data of the commutation voltage amplitude and phase angle input into the trend prediction model changes continuously, U m ( i ) and phase angle i ( i) are also updated in real time as the commutation voltage amplitude and phase angle extracted by the method proposed in the present invention until there is no need to predict commutation failure.

[0038] In order to verify the feasibility of the method for quickly extracting the commutation voltage amplitude and phase angle proposed in this patent, the following example is used for verification. The parameters of the LCC-HVDC in the embodiment are shown in Table 1.

[0039] Table 1 LCC-HVDC parameters

[0040] The verification was conducted in the LCC-HVDC standard model using the parameters in Table 1. The converter stations on both the rectifier and inverter sides of the LCC-HVDC use 12-pulse converters, which are composed of two 6-pulse converters connected in series. The wiring diagram of the 6-pulse converter is shown in Figure 2 As shown in the figure, VT1-VT6 represent 6 converter valves respectively, each of which is composed of thyristors connected in series. During the operation of the system, the converter valves are controlled by triggering pulses to cyclically conduct in the order of 1 to 6 to achieve AC / DC conversion.

[0041] In the model, the rectifier side (the process from the AC side to the DC side) adopts a constant current control method, and the inverter side (the process from the DC side to the AC side) adopts a control method of constant voltage and fixed arc extinction angle; the converter transformer adopts a three-phase three-winding transformer, the high-voltage side adopts a Yg type connection, and the two low-voltage side windings adopt a Y type connection and a D1 type connection respectively, and provide commutation voltage for two 6-pulse converters respectively. In this embodiment, when the LCC-HVDC is operating normally, an A-phase grounding fault is set at the inverter side AC bus at 0.7s, and the grounding reactance is 0.1H. Taking the commutation process from the D1-type connection converter valve VT1 to the converter valve VT3 as an example, the amplitude and phase angle of the commutation voltage after the fault are extracted to verify the method proposed in this embodiment 1.

[0042] Figure 3 The figure shows the commutation voltage of the D1-connected converter valve VT1 before and after the fault occurs. u com_D1 Instantaneous value waveform, where Δ U m Indicates the change in the commutation voltage amplitude before and after the system fault; Δ i It represents the change in the phase angle of the commutation voltage before and after the system fault. It can be seen that the waveform of the commutation voltage is a stable sine wave under normal system operation before 0.7s. When a ground fault occurs on the AC side of the system 0.7s later, the commutation voltage is distorted, and its amplitude and phase angle also change accordingly, which increases the risk of commutation failure.

[0043] The following is a comparison of the extraction results of the commutation voltage amplitude phase angle extraction method proposed in the present invention and the existing trigonometric function fitting extraction method to verify the rapidity of the proposed method. The relevant parameters are set as follows: Sampling interval T =0.2ms, number of fittings m =5, Amplitude memory factor α u =0.5, phase angle memory factor α θ =0.2, amplitude trend factor β u =0.5, phase angle trend factor β θ =0.7, amplitude oscillation factor c u =0.3, phase oscillation factor c θ =0.5, oscillation period length p =6. From equations (11) and (12), we can see that the amplitude memory component L u and phase angle memory component L θ Describes the basic level and amplitude trend components of the input data B u and phase angle trend component B θ Describes the direction and speed of change of input data, amplitude oscillation component S u and phase angle oscillation component S θ Reflects the periodic fluctuations of the input data. In the calculation process of these components, the amplitude memory factor α u and phase memory factor α θ Controls the model's sensitivity to the overall level of the data, α u and α θ The larger the value, the more the model tends to respond faster to changes in recent data; the amplitude trend factor β u and phase angle trend factor β θ Controls the degree of smoothing of the trend component, β u and β θ The larger the value, the faster the model responds to the trend, which means that the model updates and adjusts the trend information faster. c u and phase oscillation factor c θControls the smoothness of the oscillating component, c u and c θ The larger it is, the faster the model responds to cyclical changes, which means that the model pays more attention to recent cyclical data.

[0044] Figure 4 The commutation voltages for both methods are shown in the fault condition. u com_D1 Amplitude extraction results, the horizontal axis in the figure is the time near the fault, the vertical axis is the per-unit value of the extracted voltage amplitude, and the two curves in the figure represent the amplitude of the commutation voltage near the fault moment obtained by two different extraction methods, where the blue curve represents the amplitude extracted by the existing trigonometric function fitting method, and the red curve represents the amplitude extracted by the method proposed in the present invention.

[0045] from Figure 4 It can be seen that when the LCC-HVDC is operating normally before 0.7s, the commutation voltage amplitudes extracted by the two methods are both stable at around 1.02pu; when the fault occurs at 0.7s, the commutation voltage amplitude drops and eventually drops to 0.68pu. Comparing the two curves, it can be seen that both methods can extract the amplitude after the fault at different times. Among them, the method using trigonometric function fitting completes the amplitude extraction 4.28ms after the fault, and the method proposed in the present invention completes the amplitude extraction 2.16ms after the fault, 2.12ms in advance. Therefore, the proposed method is faster than the traditional method in extracting the commutation voltage amplitude after the fault, and is more conducive to the prediction of commutation failure.

[0046] Figure 5 The results of commutation voltage phase angle extraction by two methods are shown. The horizontal axis in the figure is the time near the fault, and the vertical axis is the extracted commutation voltage phase angle, in rad. The two curves represent the commutation voltage phase angle near the fault moment obtained by the two extraction methods, where the blue curve represents the phase angle extracted by the trigonometric function fitting method, and the red curve represents the phase angle extracted by the proposed method.

[0047] from Figure 5 It can be seen that when the system is operating normally before 0.7s, the commutation voltage phase angles extracted by the two methods are i are all stable around 0.02rad; when the fault occurs at 0.7s, the commutation voltage phase angle iIt increases and finally increases to 0.22rad. By comparing the two curves, it can be seen that the method of trigonometric function fitting completes the phase angle extraction 4.74ms after the fault, and the method proposed in the present invention completes the phase angle extraction 3.35ms after the fault, which is 1.39ms earlier. Therefore, the proposed method is faster than the traditional method in extracting the phase angle of the commutation voltage after a fault, and reserves more time for the defense of commutation failure.

[0048] The present invention proposes a method for extracting the amplitude and phase angle of commutation voltage based on trend prediction. The method can obtain the amplitude and phase angle of commutation voltage after a fault in a shorter time by performing trend prediction on the amplitude and phase angle obtained by trigonometric function fitting without adding other collected data, thereby improving the extraction speed of the amplitude and phase angle. In the implementation process, the method defines three prediction components: memory, trend and oscillation, and updates them in real time according to the input amplitude or phase angle data during the data processing process. Finally, the three components are combined to obtain the final extraction result, which ensures the rapidity of the extraction of the amplitude and phase angle of the commutation voltage, and the stability and accuracy of the final extraction result.

[0049] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for fast extraction of HVDC commutation voltage amplitude and phase angle based on trend prediction, characterized in that: The steps include: S1: Set the sampling interval T and the number of fitted data m The collected instantaneous value of the commutation voltage changes with time and is input into the voltage fitting module in real time. i The initial value of is 1; S2: Input the collected instantaneous value of the commutation voltage into the voltage fitting module in real time. Each time the instantaneous value of the commutation voltage is input, the counter i +1 for real-time input m The instantaneous value of the commutation voltage is fitted by trigonometric least squares function to calculate the amplitude and phase angle of the fitting curve; S3: input the amplitude and phase angle obtained in step S2 into the trend prediction model, and set the parameters of the trend prediction model; S4: When the counter i ≤ p +1, p is the oscillation period length. p +1 The amplitude and phase angle of the fitting curve are used to initialize the memory component, trend component and oscillation component of the trend prediction model, and the amplitude and phase angle data of the original fitting curve are used as the prediction results as the predicted amplitude and phase angle respectively; S5: When the counter i > p +1, the memory component, trend component and oscillation component of the trend prediction model are updated, and the amplitude and phase angle are predicted based on the updated memory component, trend component and oscillation component; S6: Real-time output of amplitude and phase angle after trend prediction.

2. The method for rapidly extracting HVDC commutation voltage amplitude and phase angle based on trend prediction according to claim 1 is characterized in that: In step S1, the sampling interval T =0.2ms, number of fitting data m= 5.

3. The method for rapidly extracting HVDC commutation voltage amplitude and phase angle based on trend prediction according to claim 1, characterized in that: In step S2, the input m The instantaneous value of the commutation voltage is fitted by the least squares trigonometric function to obtain the i The commutation voltage amplitude and phase angle fitting results U m0 ( i )and θ 0( i ); （5）； （6）； in, , ω 0 is the system angular frequency; u m ( i ) indicates the i Input m The instantaneous value of the commutation voltage; t m ( i ) indicates the i Input m The time signal corresponding to the instantaneous value of the commutation voltage; U m1 ( i )and U m2 ( i ) is an intermediate variable in the calculation process and has no actual physical meaning.

4. The method for rapidly extracting HVDC commutation voltage amplitude and phase angle based on trend prediction according to claim 1, characterized in that: In step S3, the parameters of the trend prediction model include the amplitude memory factor α u , Phase Angle Memory Factor α θ , Amplitude trend factor β u , phase angle trend factor β θ , Amplitude Oscillation Factor γ u , phase angle oscillation factor γ θ and the oscillation period length p .

5. The method for rapidly extracting HVDC commutation voltage amplitude and phase angle based on trend prediction according to claim 4, characterized in that: Amplitude memory factor α u =0.5, phase angle memory factor α θ =0.2, amplitude trend factor β u =0.5, phase angle trend factor β θ =0.7, amplitude oscillation factor γ u =0.3, phase oscillation factor γ θ =0.5, oscillation period length p =6.

6. The method for rapidly extracting HVDC commutation voltage amplitude and phase angle based on trend prediction according to claim 4, characterized in that: In step S4, the memory component includes an amplitude memory component L u and phase angle memory component L θ , the trend component includes the amplitude trend component B u and phase angle trend component B θ , the oscillation component includes the amplitude oscillation component S u and phase angle oscillation component S θ ; First, according to the previous p The amplitude and phase angle of the +1 fitting curve are used to initialize the memory component, trend component and oscillation component of the trend prediction model, as shown in equations (7) and (8): （7）； （8）； in, U m0 (1) represents the first commutation voltage amplitude fitting result; U m0 ( p +1) indicates the p +1 commutation voltage amplitude fitting result; U m0 (1: p ) indicates the first to the p An array consisting of the commutation voltage amplitude fitting results; θ 0(1) represents the first commutation voltage phase angle fitting result; θ 0( p +1) indicates the p +1 commutation voltage phase angle fitting result; θ 0(1: p ) indicates the first to the p An array consisting of the commutation voltage phase angle fitting results; Then, the amplitude and phase angle are trended and the counter i ≤ p +1, the amplitude and phase angle data of the original fitting curve are stored as prediction results in U m ( i )and θ ( i ), as shown in formula (9) and formula (10): （9）； （10）。 7. The method for rapidly extracting HVDC commutation voltage amplitude and phase angle based on trend prediction according to claim 6, characterized in that: In step S5, L u and L θ Assign values ​​to L u _prev and L θ _prev ,Right now L u _prev = L u , L θ _prev = L θ , and according to formula (11) and formula (12) respectively L u , B u , S u , L θ , B θ , S θ To update: （11）； （12）； Among them, mod() is the remainder function; The trend of amplitude and phase angle is predicted by equations (13) and (14) respectively: （13）； （14）。

Images