Novel oscillation multi-modal information dynamic fusion identification method for wind power generation system

Abstract

The invention relates to the technical field of power systems, and discloses a novel oscillation multi-modal information dynamic fusion identification method for a wind power generation system, which comprises the following steps: firstly, performing frequency analysis on collected data from a wind power system to obtain frequency domain characteristics corresponding to oscillation components of each sampling frequency point, and then, according to attenuation factors in the frequency domain characteristics, calculating the oscillation multi-modal information of each sampling frequency point; according to the method, non-power-frequency oscillation components are recognized, modal energy and damping ratios corresponding to all the non-power-frequency oscillation components are calculated, finally, the modal energy and the damping ratios are comprehensively sorted, the non-power-frequency oscillation component ranked first is selected as the most dominant oscillation mode, and oscillation mode recognition is conducted through sampling frequency points corresponding to the most dominant oscillation mode. Compared with an existing method, the identification method is high in operation speed, good in real-time performance and wide in applicability.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a novel oscillation multi-modal information dynamic fusion identification method for a wind power generation system. Background Art

[0002] In recent years, with the application of high-proportion power electronic equipment in high-proportion wind power generation, the power system has shown a new feature of "double high". In view of this, the "mechanical-electrical-magnetic" interaction between power electronic equipment and between power grids will cause the voltage, current and power information to fluctuate periodically over time, which is easy to cause new oscillations with a large range of frequency changes.

[0003] The occurrence of this new type of oscillation, its oscillation frequency is closely related to information such as wind speed, number of wind turbines and system impedance. In addition, whether in different wind farms or the same wind farm, oscillations of different frequencies will occur due to different system operating parameters. In order to effectively suppress such oscillations, it is necessary to use effective methods to identify the oscillation frequency in advance.

[0004] In the invention patent with announcement number CN106300345, a method for identifying low-frequency oscillation parameters of an improved Prony algorithm is proposed, which mainly uses neural networks and gradient descent methods to adjust weights to achieve the solution of the Prony algorithm, but only detects low-frequency oscillations in the power system; in the invention patent with announcement number CN104852392A, a method for calculating the subsynchronous oscillation modal attenuation coefficient based on the Prony algorithm is proposed, but focuses on the torsional vibration mode and subsynchronous oscillation frequency of the turbine generator shaft system. However, the "wideband" characteristics of the new oscillation of the wind power system are not only manifested in the subsynchronous oscillation frequency, but also in the supersynchronous oscillation frequency. Therefore, how to effectively identify the oscillation frequency in a sufficiently "wideband" area is a problem that technical personnel in this field urgently need to solve. Summary of the invention

[0005] The present invention provides a novel oscillation multi-modal information dynamic fusion identification method for a wind power generation system, which solves the technical problem that the existing method cannot identify subsynchronous and supersynchronous oscillations of a wind power system.

[0006] The present invention can be achieved through the following technical solutions:

[0007] A new type of dynamic fusion identification method for oscillation multi-modal information of a wind power generation system, firstly performs frequency analysis on the collected data from the wind power system to obtain the frequency domain characteristics corresponding to the oscillation components of each sampling frequency point, then identifies the non-power frequency oscillation components according to the attenuation factors in the frequency domain characteristics, and then calculates the modal energy and damping ratio corresponding to each non-power frequency oscillation component, and finally selects the non-power frequency oscillation component ranked first as the most dominant oscillation mode by comprehensively ranking the modal energy and damping ratio, and performs oscillation mode identification at the sampling frequency points corresponding to the most dominant oscillation mode.

[0008] Further, the oscillation modes include a low frequency oscillation mode LFO, a sub / super synchronous oscillation mode SSO and a medium and high frequency oscillation mode MHFO;

[0009] If the sampling frequency point corresponding to the most dominant oscillation mode is between 0 and 10 Hz, it is identified as the low-frequency oscillation mode LFO; if the sampling frequency point corresponding to the most dominant oscillation mode is between 10 and 100 Hz, it is identified as the sub / super synchronous oscillation mode SSO; if the sampling frequency point corresponding to the most dominant oscillation mode is between 100 and several kHz, it is identified as the medium-high frequency oscillation mode MHFO.

[0010] Furthermore, when the multimodal information dynamic fusion identification method is used to perform frequency domain analysis on the sampled data, the time domain expression of the sampled data is recorded as

[0011]

[0012] In the formula, p is the model order, fi is the frequency, Ai is the amplitude, θ i is the phase, σ i is the attenuation factor;

[0013] Further, if the attenuation factor corresponding to the oscillation component is less than zero, the corresponding oscillation component is a non-power frequency oscillation component.

[0014] Furthermore, the modal energy and damping ratio of each non-power frequency oscillation component are calculated using the following formula, with modal energy as the main priority and low damping ratio as the secondary priority. The bubble sorting method is used to sort them, and the non-power frequency oscillation component ranked first is selected as the most dominant oscillation mode;

[0015] The modal energy E i and the damping ratio ξ i The mathematical expression of can be expressed as follows.

[0016]

[0017]

[0018] In the formula, A i is the amplitude, z iis the matrix coefficient in the discrete mathematical model after discretization, M is the number of sampling frequency points, and i is the number of the sampling frequency point.

[0019] The beneficial technical effects of the present invention are:

[0020] The present invention can identify the oscillation frequency, amplitude, phase, and attenuation factor in the voltage / current signal for the specific working condition of subsynchronous / supersynchronous oscillation of a high-proportion wind power generation system, and screen out non-power frequency oscillation components with an attenuation factor less than zero and their information, and then autonomously calculate the damping ratio and modal energy of the screened non-power frequency oscillation components, and then sort the non-power frequency oscillation components according to the principle of high modal energy as the main priority and low damping ratio as the secondary priority, and store the corresponding modal information. It is best to distinguish the most dominant oscillation mode based on the sorting result, and determine the most dominant oscillation mode. It can realize wide-area information identification of wind power generation oscillation, and the identification frequency band range reaches several kHz. It can realize subsynchronous oscillation identification and supersynchronous oscillation identification, and can further provide support for oscillation suppression. Compared with the existing methods, the identification method of the present invention has fast computing speed, good real-time performance, and wide applicability. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0022] Figure 2 It is a schematic diagram of the application of the novel oscillation multi-modal information dynamic fusion identification method of the wind power generation system of the present invention in the wind power generation system;

[0023] Figure 3 A schematic diagram showing the comparison between the simulation results and the actual measurement results after discretization processing of the collected current signal by using the novel oscillation multi-modal information dynamic fusion identification method of the wind power generation system of the present invention, wherein red represents the fitting simulation results and blue represents the actual measurement results;

[0024] Figure 4 The result diagram of a specific embodiment of oscillation identification using the novel oscillation multi-modal information dynamic fusion identification method of the wind power generation system of the present invention is shown. DETAILED DESCRIPTION

[0025] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] like Figure 1As shown, the present invention provides a new type of oscillation multi-modal information dynamic fusion identification method for a wind power generation system. First, the collected data from the wind power system is subjected to frequency analysis to obtain the frequency domain characteristics corresponding to the oscillation components of each sampling frequency point. Then, according to the attenuation factor in the frequency domain characteristics, the non-power frequency oscillation components are identified. Then, the modal energy and damping ratio corresponding to each non-power frequency oscillation component are calculated. Finally, by comprehensively ranking the modal energy and damping ratio, the non-power frequency oscillation component ranked first is selected as the most dominant oscillation mode, and the oscillation mode is identified at the sampling frequency points corresponding to the most dominant oscillation mode.

[0027] The specific implementation is as follows:

[0028] Step 1: Oscillation frequency signal identification

[0029] The time domain expression of the oscillation signal x'(t) can be expressed as:

[0030]

[0031] In the formula, p is the model order, f i is the frequency, i is the fitting exponential function number, which can represent 1, 2, 3, ..., A i is the amplitude, t is the time, θ i is the phase, σ i is the attenuation factor.

[0032] The oscillation signal x(t) is discretized, and the discrete mathematical model of the sampling time can be expressed as:

[0033]

[0034] Where, zi and bi are matrix coefficients; Δt is the sampling time interval; N is the number of data sampling points, N≥2p.

[0035] The coefficients of the discrete mathematical model are further calculated, and on this basis, the frequency domain information of the obtained sampling frequency points is solved, including the frequency, amplitude, phase, attenuation factor and other information of the sampling frequency points. The specific expression is as follows:

[0036] s i =In|Z i | / △t (5)

[0037] f i =arg[Im(Z i ) / Re(Z i )] / 2p△t (6)

[0038] A i =b i | (7)

[0039] θ i =arg[Im(b i ) / Re(b i )] (8)

[0040] Step 2: Identification of non-power frequency oscillation modes

[0041] The attenuation factor of the corresponding sampling frequency point obtained by the multimodal information fusion identification method is extracted. The attenuation factor can reflect the attenuation trend and degree of the system at the frequency point. The stability of the system at each sampling frequency point is analyzed based on the attenuation factor. When the attenuation factor σ of the sampling frequency point is i When it is greater than zero, the system is very likely to oscillate at this frequency. Extract and store the attenuation factor σ i Dynamic frequency domain information of non-power frequency sampling frequency points greater than zero.

[0042] Step 3: Identify the most dominant oscillation mode

[0043] In determining the attenuation factor σ i Based on the non-power frequency oscillation component less than zero, the modal energy and damping ratio are autonomously calculated in combination with its dynamic frequency domain information to further determine the most dominant oscillation mode in a given oscillation signal.

[0044] The oscillation mode is all the information that characterizes the oscillation frequency, including oscillation frequency, amplitude, phase, energy, damping ratio, attenuation factor, etc. The most dominant oscillation mode is determined by two indicators: 1) modal energy; 2) damping ratio. The following formula can be used to calculate the modal energy and damping ratio corresponding to each non-industrial frequency oscillation component.

[0045] The mathematical expressions of modal energy and damping ratio can be expressed as follows:

[0046]

[0047] In the formula, A i is the amplitude, M is the number of sampling frequency points, and i is the sampling frequency point number.

[0048] The modal energy represents the stability of the system at the non-power frequency sampling frequency point. i and z i When the value of is large, the modal energy E i The larger the damping ratio is, the more serious the oscillation will be, and vice versa. i Characterizes how fast the oscillation decays. The larger the damping ratio, the faster the oscillation amplitude decays, and the smaller the damping ratio, the slower the decay rate. Therefore, the two can better characterize the dynamic change process of the system at this frequency point.

[0049] According to the self-calculated results of the internal modal energy and damping ratio described above, the bubble sort method is used to sort the non-power frequency oscillation components with modal energy as the main priority and low damping ratio as the secondary priority. The two criteria are used at the same time to avoid the consistency of the modal energy of the non-power frequency oscillation components corresponding to the sampling frequency point, and the damping ratio can be further used to identify the most dominant oscillation mode.

[0050] The autonomous calculation results of the modal energy and damping ratio of the sampling frequency points are sorted. The sorting principle gives priority to the modal energy. The larger the modal energy, the earlier the position, and vice versa. When the modal energy of different sampling frequency points is detected to be consistent, the damping ratio is further considered as a sorting rule. The smaller the damping ratio, the earlier the position, and vice versa. The bubble sort method is used to sort the non-power frequency oscillation components in descending order to obtain the descending oscillation frequency, and the frequency domain information of the corresponding oscillation frequency is stored at the same time.

[0051] Step 4: Oscillation pattern identification

[0052] From the descending oscillation frequency sequence obtained in step 3, the non-power frequency oscillation component at the front is selected as the most dominant oscillation mode.

[0053] Oscillation mode positioning

[0054] The descending oscillation pattern obtained after bubble sorting is positioned, and the positioning object is the oscillation frequency. The positioning principle is the specified frequency band range, and the frequency band ranges are 0-10Hz, 10-100Hz, and 100-several kHz.

[0055] Oscillation mode classification

[0056] The oscillation mode positioning results obtained are used to classify the oscillation modes, and the classification categories are: 0~10Hz is the low-frequency oscillation mode LFO (Low Frequency Oscillation), 10~100Hz is the sub / super synchronous oscillation mode SSO (Sub / SupSynchronous Oscillation), 100~several kHz is the medium to high frequency oscillation mode MHFO (Medium to High Frequency Oscillation).

[0057] In order to verify the feasibility of the novel oscillation multi-modal information dynamic fusion identification method of the wind power generation system of the present invention, we conducted the following experiments:

[0058] exist Figure 2In the wind power grid-connected system shown in the figure, the grid frequency is 50Hz. When there is no oscillation, there is only the industrial frequency signal, i.e., 50Hz, in the system. As the number of wind turbines increases, non-industrial frequency signals, i.e., 22Hz and 78Hz, appear in the system. The current signal of the wind power grid-connected system is collected by using the novel oscillation multi-modal information dynamic fusion identification method of the wind power system of the present invention. The sampling frequency interval is 2Hz, and the current signal is discretized. The fitting result is shown in FIG. Figure 3 As shown in the figure, it can be seen that the fitting results are basically consistent with the measurement results.

[0059] In addition, by collecting frequency information and calculating the oscillation energy and damping ratio, the dominant oscillation mode is determined, and finally the oscillation mode is determined. From the data analysis results in Table 1, it can be seen that the attenuation factor of 22Hz is 0.11, and the attenuation factor of 78Hz is 0.12, so it is judged that there is a subsynchronous oscillation frequency of 22Hz and a supersynchronous oscillation frequency of 78Hz. In addition, the energy of 22Hz is 7.9, and the energy of 78Hz is 2. Since the energy is the dominant factor at this time, it can be judged that 22Hz is the dominant oscillation frequency. The specific results are as follows: Figure 4 Therefore, the feasibility of the proposed oscillation multi-modal information dynamic fusion identification method is further verified.

[0060] Table 1 Calculation results of modal energy and damping ratio

[0061]

[0062] Although specific embodiments of the present invention are described above, those skilled in the art should understand that these are merely examples and that various changes or modifications may be made to these embodiments without departing from the principles and essence of the present invention. Therefore, the scope of protection of the present invention is limited by the appended claims.

Claims

1. A new type of oscillation multi-modal information dynamic fusion identification method for wind power generation system, characterized by: First, the collected data from the wind power system is subjected to frequency analysis to obtain the frequency domain characteristics corresponding to the oscillation components at each sampling frequency point. Then, the non-power frequency oscillation components are identified based on the attenuation factors in the frequency domain characteristics. The modal energy and damping ratio corresponding to each non-power frequency oscillation component are calculated. Finally, the modal energy and damping ratio are comprehensively ranked, and the non-power frequency oscillation component ranked first is selected as the most dominant oscillation mode. The oscillation mode is identified at the sampling frequency points corresponding to the most dominant oscillation mode.

2. The novel oscillation multi-modal information dynamic fusion identification method for wind power generation system according to claim 1 is characterized by: The oscillation modes include a low frequency oscillation mode LFO, a sub / super synchronous oscillation mode SSO and a medium and high frequency oscillation mode MHFO; If the sampling frequency point corresponding to the most dominant oscillation mode is between 0 and 10 Hz, it is identified as the low-frequency oscillation mode LFO; if the sampling frequency point corresponding to the most dominant oscillation mode is between 10 and 100 Hz, it is identified as the sub / super synchronous oscillation mode SSO; if the sampling frequency point corresponding to the most dominant oscillation mode is between 100 and several kHz, it is identified as the medium-high frequency oscillation mode MHFO.

3. The novel oscillation multi-modal information dynamic fusion identification method for wind power generation system according to claim 1 is characterized by: When the multi-modal information dynamic fusion identification method is used to perform frequency domain analysis on the sampled data, The time domain expression of the sampled data is In the formula, p is the model order, fi is the frequency, Ai is the amplitude, θ i is the phase, σ i is the attenuation factor;.

4. The novel oscillation multi-modal information dynamic fusion identification method for wind power generation system according to claim 3 is characterized by: If the attenuation factor corresponding to the oscillation component is less than zero, the corresponding oscillation component is a non-power frequency oscillation component.

5. The novel oscillation multi-modal information dynamic fusion identification method for wind power generation system according to claim 4 is characterized by: The following formula is used to calculate the modal energy and damping ratio of each non-power frequency oscillation component. The modal energy is the main priority and the low damping ratio is the secondary priority. The bubble sorting method is used to sort them and the non-power frequency oscillation component ranked first is selected as the most dominant oscillation mode. The modal energy E i and the damping ratio ξ i The mathematical expression of can be expressed as follows. In the formula, A i is the amplitude, z i is the matrix coefficient in the discrete mathematical model after discretization, M is the number of sampling frequency points, and i is the number of the sampling frequency point.

Images